diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31189 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998315647633484,
+  "global_step": 5194,
+  "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": 1.7587,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 1.5485,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.7983,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 1.7471,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.41025641025641e-06,
+      "loss": 1.6292,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.6144,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 1.4115,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 1.5227,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.153846153846154e-05,
+      "loss": 1.3025,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-05,
+      "loss": 1.2472,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4102564102564104e-05,
+      "loss": 1.2325,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.1809,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.1912,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 1.0707,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 1.0724,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0512820512820512e-05,
+      "loss": 1.1102,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1794871794871795e-05,
+      "loss": 1.005,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 1.0608,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.435897435897436e-05,
+      "loss": 1.0048,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-05,
+      "loss": 1.047,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6923076923076923e-05,
+      "loss": 0.9852,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8205128205128207e-05,
+      "loss": 1.207,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.948717948717949e-05,
+      "loss": 1.1148,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 0.8741,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.205128205128206e-05,
+      "loss": 0.9966,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.9243,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.461538461538462e-05,
+      "loss": 0.9156,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.58974358974359e-05,
+      "loss": 1.1348,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.717948717948718e-05,
+      "loss": 1.0149,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 0.8106,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.974358974358974e-05,
+      "loss": 0.8989,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1025641025641023e-05,
+      "loss": 0.9304,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.230769230769231e-05,
+      "loss": 0.9586,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.358974358974359e-05,
+      "loss": 1.0163,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4871794871794874e-05,
+      "loss": 0.9543,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 1.0556,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7435897435897435e-05,
+      "loss": 0.9953,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.871794871794872e-05,
+      "loss": 1.0263,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 0.9513,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.128205128205128e-05,
+      "loss": 0.9563,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.256410256410257e-05,
+      "loss": 0.8786,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.384615384615385e-05,
+      "loss": 0.8987,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.512820512820514e-05,
+      "loss": 0.8554,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6410256410256414e-05,
+      "loss": 0.9484,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.769230769230769e-05,
+      "loss": 0.8588,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.897435897435898e-05,
+      "loss": 0.9012,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.025641025641026e-05,
+      "loss": 0.8631,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.153846153846155e-05,
+      "loss": 0.9003,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.282051282051282e-05,
+      "loss": 0.8069,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.410256410256412e-05,
+      "loss": 1.0157,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.538461538461539e-05,
+      "loss": 0.9078,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.8666,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.794871794871795e-05,
+      "loss": 0.8766,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.923076923076924e-05,
+      "loss": 0.8609,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.051282051282052e-05,
+      "loss": 1.0072,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.17948717948718e-05,
+      "loss": 0.9886,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.307692307692307e-05,
+      "loss": 0.8919,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.435897435897436e-05,
+      "loss": 0.9122,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.564102564102564e-05,
+      "loss": 0.8662,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 0.8699,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.820512820512821e-05,
+      "loss": 1.0497,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.948717948717948e-05,
+      "loss": 0.806,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.076923076923078e-05,
+      "loss": 0.8401,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.205128205128205e-05,
+      "loss": 1.0192,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 0.9942,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.461538461538461e-05,
+      "loss": 0.9074,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.58974358974359e-05,
+      "loss": 0.9523,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.717948717948718e-05,
+      "loss": 0.8344,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.846153846153847e-05,
+      "loss": 0.8519,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.974358974358975e-05,
+      "loss": 0.8352,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.102564102564103e-05,
+      "loss": 0.8857,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230769230769232e-05,
+      "loss": 0.782,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.35897435897436e-05,
+      "loss": 0.8555,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.487179487179487e-05,
+      "loss": 0.8677,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615384615384617e-05,
+      "loss": 0.9915,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.743589743589744e-05,
+      "loss": 0.9997,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.871794871794872e-05,
+      "loss": 0.8272,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001,
+      "loss": 0.8185,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010128205128205129,
+      "loss": 0.881,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010256410256410256,
+      "loss": 0.7425,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010384615384615386,
+      "loss": 0.9007,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010512820512820514,
+      "loss": 0.9135,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010641025641025641,
+      "loss": 0.8364,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001076923076923077,
+      "loss": 0.9526,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010897435897435896,
+      "loss": 0.8493,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011025641025641027,
+      "loss": 1.0166,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011153846153846154,
+      "loss": 0.8232,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011282051282051283,
+      "loss": 0.8375,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001141025641025641,
+      "loss": 0.8517,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011538461538461538,
+      "loss": 0.9094,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011666666666666668,
+      "loss": 0.9061,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011794871794871796,
+      "loss": 0.8713,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011923076923076923,
+      "loss": 0.8977,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012051282051282052,
+      "loss": 0.9069,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012179487179487179,
+      "loss": 0.8647,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001230769230769231,
+      "loss": 0.8399,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012435897435897437,
+      "loss": 0.9707,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012564102564102564,
+      "loss": 0.7926,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012692307692307693,
+      "loss": 0.7879,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012820512820512823,
+      "loss": 0.7907,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001294871794871795,
+      "loss": 0.9232,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013076923076923077,
+      "loss": 0.8194,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013205128205128204,
+      "loss": 0.8408,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 1.0193,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013461538461538464,
+      "loss": 0.9132,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001358974358974359,
+      "loss": 0.8642,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013717948717948718,
+      "loss": 0.911,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013846153846153847,
+      "loss": 1.0741,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013974358974358974,
+      "loss": 0.8914,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014102564102564104,
+      "loss": 0.9308,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001423076923076923,
+      "loss": 0.9038,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001435897435897436,
+      "loss": 0.7511,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014487179487179488,
+      "loss": 0.9869,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014615384615384615,
+      "loss": 0.8719,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014743589743589745,
+      "loss": 0.8579,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014871794871794872,
+      "loss": 0.9075,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.7615,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015128205128205128,
+      "loss": 0.8346,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015256410256410255,
+      "loss": 0.7508,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015384615384615385,
+      "loss": 0.8893,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015512820512820515,
+      "loss": 0.901,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015641025641025642,
+      "loss": 0.8284,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001576923076923077,
+      "loss": 0.938,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015897435897435896,
+      "loss": 0.843,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016025641025641028,
+      "loss": 0.8129,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016153846153846155,
+      "loss": 0.9413,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016282051282051282,
+      "loss": 0.7747,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001641025641025641,
+      "loss": 0.7802,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001653846153846154,
+      "loss": 0.8776,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 0.8659,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016794871794871796,
+      "loss": 0.9487,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016923076923076923,
+      "loss": 0.8569,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017051282051282053,
+      "loss": 0.8904,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001717948717948718,
+      "loss": 0.915,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001730769230769231,
+      "loss": 0.8266,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017435897435897436,
+      "loss": 0.8782,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017564102564102566,
+      "loss": 0.7474,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017692307692307693,
+      "loss": 0.8141,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017820512820512823,
+      "loss": 0.8347,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001794871794871795,
+      "loss": 0.7772,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018076923076923077,
+      "loss": 0.9469,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018205128205128207,
+      "loss": 0.7668,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018333333333333334,
+      "loss": 0.8395,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018461538461538463,
+      "loss": 0.8101,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001858974358974359,
+      "loss": 0.8875,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001871794871794872,
+      "loss": 0.8251,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018846153846153847,
+      "loss": 0.8099,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018974358974358974,
+      "loss": 0.8551,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019102564102564104,
+      "loss": 0.8636,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019230769230769233,
+      "loss": 0.8142,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001935897435897436,
+      "loss": 0.8383,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019487179487179487,
+      "loss": 0.9934,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019615384615384615,
+      "loss": 0.7792,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019743589743589744,
+      "loss": 0.8618,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019871794871794874,
+      "loss": 0.9392,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7929,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999998055744174,
+      "loss": 0.8433,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999992222977447,
+      "loss": 0.8666,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999998250170209,
+      "loss": 0.9579,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999968891921884,
+      "loss": 0.7686,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999951393642116,
+      "loss": 1.0083,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999930006869593,
+      "loss": 0.8831,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999904731612635,
+      "loss": 0.7757,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999875567881068,
+      "loss": 0.9098,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999984251568623,
+      "loss": 0.8384,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999805575040975,
+      "loss": 1.0128,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999976474595967,
+      "loss": 0.8163,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999720028458185,
+      "loss": 0.8298,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999671422553914,
+      "loss": 0.9663,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999618928265758,
+      "loss": 0.8845,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999562545614124,
+      "loss": 0.8927,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999502274620941,
+      "loss": 0.8265,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999438115309644,
+      "loss": 0.9483,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999370067705181,
+      "loss": 0.8594,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999298131834014,
+      "loss": 0.9421,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999222307724113,
+      "loss": 1.0282,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999142595404965,
+      "loss": 0.7792,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999058994907564,
+      "loss": 0.8125,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999897150626442,
+      "loss": 0.7371,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999888012950955,
+      "loss": 0.8286,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999878486467849,
+      "loss": 0.8649,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998685711808281,
+      "loss": 0.8065,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999858267093748,
+      "loss": 0.9503,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998475742106155,
+      "loss": 0.9323,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999836492535588,
+      "loss": 0.7784,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998250220729758,
+      "loss": 0.8762,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998131628272382,
+      "loss": 0.7666,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998009148029867,
+      "loss": 0.8752,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997882780049847,
+      "loss": 0.9704,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999775252438145,
+      "loss": 0.8814,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997618381075333,
+      "loss": 0.9225,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999748035018366,
+      "loss": 0.8901,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997338431760095,
+      "loss": 0.8639,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997192625859835,
+      "loss": 0.9736,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997042932539568,
+      "loss": 0.7797,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996889351857504,
+      "loss": 0.9272,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996731883873366,
+      "loss": 0.8389,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996570528648384,
+      "loss": 0.9903,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996405286245298,
+      "loss": 0.7995,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999623615672837,
+      "loss": 0.703,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999606314016336,
+      "loss": 0.8421,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995886236617546,
+      "loss": 0.8648,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995705446159722,
+      "loss": 0.966,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995520768860184,
+      "loss": 0.8135,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995332204790744,
+      "loss": 0.7829,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995139754024727,
+      "loss": 0.8987,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994943416636968,
+      "loss": 0.9417,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994743192703812,
+      "loss": 0.7744,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994539082303118,
+      "loss": 0.8953,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994331085514252,
+      "loss": 0.8649,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994119202418098,
+      "loss": 0.8756,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993903433097038,
+      "loss": 0.842,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993683777634987,
+      "loss": 0.8587,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993460236117347,
+      "loss": 0.8689,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993232808631048,
+      "loss": 0.9195,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993001495264526,
+      "loss": 0.768,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992766296107726,
+      "loss": 0.8462,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992527211252101,
+      "loss": 0.8263,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992284240790626,
+      "loss": 0.809,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999203738481778,
+      "loss": 0.9111,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991786643429552,
+      "loss": 0.8752,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991532016723439,
+      "loss": 0.9716,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991273504798456,
+      "loss": 0.7736,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991011107755127,
+      "loss": 0.9797,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990744825695484,
+      "loss": 0.8006,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990474658723072,
+      "loss": 0.8718,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990200606942945,
+      "loss": 0.8377,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989922670461669,
+      "loss": 0.8429,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998964084938732,
+      "loss": 0.8466,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989355143829483,
+      "loss": 0.8131,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989065553899257,
+      "loss": 0.7982,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988772079709244,
+      "loss": 0.7865,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988474721373568,
+      "loss": 0.861,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988173479007856,
+      "loss": 0.8551,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987868352729248,
+      "loss": 0.8261,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987559342656385,
+      "loss": 0.7111,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987246448909435,
+      "loss": 0.8661,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986929671610062,
+      "loss": 0.8711,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986609010881447,
+      "loss": 0.9537,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998628446684828,
+      "loss": 0.7686,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985956039636756,
+      "loss": 0.9092,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998562372937459,
+      "loss": 0.8042,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985287536190995,
+      "loss": 0.7676,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984947460216707,
+      "loss": 0.8045,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984603501583962,
+      "loss": 0.8355,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984255660426508,
+      "loss": 0.9295,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983903936879603,
+      "loss": 0.8698,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983548331080017,
+      "loss": 0.8131,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983188843166025,
+      "loss": 0.8695,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998282547327742,
+      "loss": 0.933,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998245822155549,
+      "loss": 0.8455,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998208708814305,
+      "loss": 0.9243,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998171207318441,
+      "loss": 0.8395,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981333176825398,
+      "loss": 0.762,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980950399213344,
+      "loss": 0.802,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980563740497096,
+      "loss": 0.8629,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980173200827005,
+      "loss": 0.743,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979778780354935,
+      "loss": 0.7819,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997938047923425,
+      "loss": 0.8245,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978978297619838,
+      "loss": 0.9308,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978572235668086,
+      "loss": 0.8862,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978162293536887,
+      "loss": 0.93,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977748471385653,
+      "loss": 0.8252,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977330769375298,
+      "loss": 0.841,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976909187668245,
+      "loss": 0.8945,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976483726428422,
+      "loss": 0.764,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976054385821278,
+      "loss": 0.9599,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997562116601376,
+      "loss": 0.7883,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975184067174323,
+      "loss": 0.7896,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974743089472936,
+      "loss": 0.8387,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997429823308107,
+      "loss": 0.8063,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973849498171716,
+      "loss": 0.8648,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997339688491936,
+      "loss": 0.818,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972940393499998,
+      "loss": 1.005,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972480024091143,
+      "loss": 0.891,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972015776871803,
+      "loss": 0.8679,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997154765202251,
+      "loss": 0.8565,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019971075649725294,
+      "loss": 0.982,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970599770163686,
+      "loss": 0.7612,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970120013522736,
+      "loss": 0.8562,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969636379989,
+      "loss": 0.8757,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996914886975054,
+      "loss": 0.8556,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996865748299692,
+      "loss": 0.8201,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996816221991922,
+      "loss": 0.8874,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019967663080710027,
+      "loss": 0.8736,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019967160065563423,
+      "loss": 0.8442,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966653174675013,
+      "loss": 0.7872,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966142408241901,
+      "loss": 0.7151,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019965627766462696,
+      "loss": 0.765,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019965109249537523,
+      "loss": 0.7609,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964586857667998,
+      "loss": 0.7904,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964060591057262,
+      "loss": 0.8337,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963530449909953,
+      "loss": 0.8325,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019962996434432214,
+      "loss": 0.7783,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019962458544831705,
+      "loss": 0.9296,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961916781317576,
+      "loss": 0.7292,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961371144100495,
+      "loss": 0.9165,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960821633392635,
+      "loss": 0.8514,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960268249407675,
+      "loss": 0.9129,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959710992360798,
+      "loss": 0.9198,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959149862468692,
+      "loss": 0.9026,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958584859949556,
+      "loss": 0.8582,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958015985023088,
+      "loss": 0.9179,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019957443237910504,
+      "loss": 0.8584,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956866618834506,
+      "loss": 0.7285,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956286128019319,
+      "loss": 0.7032,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995570176569067,
+      "loss": 0.9153,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955113532075784,
+      "loss": 0.94,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954521427403398,
+      "loss": 0.8248,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019953925451903756,
+      "loss": 0.9939,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199533256058086,
+      "loss": 0.7194,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995272188935118,
+      "loss": 0.7792,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019952114302766255,
+      "loss": 0.808,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995150284629008,
+      "loss": 0.927,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019950887520160432,
+      "loss": 0.8047,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995026832461657,
+      "loss": 0.8116,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949645259899272,
+      "loss": 0.8436,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949018326250823,
+      "loss": 0.8016,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019948387523914998,
+      "loss": 0.7963,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994775285313709,
+      "loss": 0.7838,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994711431416389,
+      "loss": 0.8209,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019946471907243698,
+      "loss": 0.9113,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019945825632626313,
+      "loss": 0.7819,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019945175490563037,
+      "loss": 0.8728,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019944521481306683,
+      "loss": 0.873,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019943863605111556,
+      "loss": 0.923,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994320186223348,
+      "loss": 0.7797,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019942536252929771,
+      "loss": 0.7485,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941866777459252,
+      "loss": 0.7899,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994119343608225,
+      "loss": 0.9315,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019940516229060594,
+      "loss": 0.691,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939835156657616,
+      "loss": 0.8453,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001993915021913815,
+      "loss": 0.8723,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019938461416768545,
+      "loss": 0.8917,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019937768749816626,
+      "loss": 0.9366,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001993707221855175,
+      "loss": 0.8699,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019936371823244762,
+      "loss": 0.8863,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993566756416801,
+      "loss": 0.8947,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993495944159534,
+      "loss": 0.7547,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019934247455802117,
+      "loss": 0.7952,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993353160706519,
+      "loss": 0.8812,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019932811895662921,
+      "loss": 0.9074,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019932088321875172,
+      "loss": 0.7438,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199313608859833,
+      "loss": 0.9063,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019930629588270175,
+      "loss": 0.7372,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992989442902016,
+      "loss": 0.7409,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929155408519124,
+      "loss": 0.8199,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019928412527054433,
+      "loss": 0.837,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019927665784914958,
+      "loss": 0.8845,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019926915182391075,
+      "loss": 0.837,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019926160719774654,
+      "loss": 1.0273,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019925402397359066,
+      "loss": 0.7631,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019924640215439188,
+      "loss": 0.7932,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923874174311394,
+      "loss": 0.9311,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923104274273566,
+      "loss": 0.7258,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992233051562507,
+      "loss": 0.8335,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992155289866679,
+      "loss": 0.748,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019920771423701105,
+      "loss": 0.9806,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019919986091031887,
+      "loss": 0.9572,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019919196900964516,
+      "loss": 0.9127,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019918403853805869,
+      "loss": 0.7973,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019917606949864324,
+      "loss": 0.8708,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019916806189449759,
+      "loss": 0.8037,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991600157287355,
+      "loss": 0.7987,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991519310044857,
+      "loss": 0.7476,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019914380772489195,
+      "loss": 0.8573,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019913564589311302,
+      "loss": 0.7849,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019912744551232268,
+      "loss": 0.8173,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991192065857096,
+      "loss": 0.867,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991109291164775,
+      "loss": 0.7672,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019910261310784514,
+      "loss": 0.7539,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019909425856304614,
+      "loss": 0.81,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990858654853292,
+      "loss": 0.7502,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019907743387795798,
+      "loss": 0.7776,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019906896374421114,
+      "loss": 0.8568,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019906045508738228,
+      "loss": 0.9128,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019905190791077998,
+      "loss": 0.6812,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019904332221772788,
+      "loss": 0.8406,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990346980115645,
+      "loss": 0.7789,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019902603529564338,
+      "loss": 0.8958,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199017334073333,
+      "loss": 0.8661,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019900859434801685,
+      "loss": 0.7882,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019899981612309342,
+      "loss": 0.8868,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989909994019761,
+      "loss": 0.8238,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989821441880933,
+      "loss": 0.853,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019897325048488833,
+      "loss": 0.8408,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989643182958196,
+      "loss": 0.867,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019895534762436032,
+      "loss": 0.7756,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989463384739988,
+      "loss": 0.9209,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019893729084823826,
+      "loss": 0.9125,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019892820475059687,
+      "loss": 0.8314,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019891908018460774,
+      "loss": 0.7679,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198909917153819,
+      "loss": 0.8164,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001989007156617937,
+      "loss": 0.8609,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019889147571210982,
+      "loss": 0.7568,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019888219730836038,
+      "loss": 0.7799,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019887288045415327,
+      "loss": 0.9154,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019886352515311134,
+      "loss": 0.7848,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019885413140887243,
+      "loss": 0.8916,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019884469922508934,
+      "loss": 0.7659,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019883522860542971,
+      "loss": 0.765,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019882571955357628,
+      "loss": 0.7442,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019881617207322661,
+      "loss": 0.7247,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019880658616809325,
+      "loss": 0.8138,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987969618419037,
+      "loss": 0.8807,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019878729909840042,
+      "loss": 0.801,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019877759794134073,
+      "loss": 0.9438,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019876785837449695,
+      "loss": 0.8732,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987580804016563,
+      "loss": 0.7199,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019874826402662104,
+      "loss": 0.937,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987384092532082,
+      "loss": 0.9229,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019872851608524985,
+      "loss": 0.8843,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019871858452659294,
+      "loss": 0.811,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019870861458109936,
+      "loss": 0.7292,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198698606252646,
+      "loss": 0.8418,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019868855954512454,
+      "loss": 0.7562,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986784744624417,
+      "loss": 0.7796,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019866835100851903,
+      "loss": 0.9022,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986581891872931,
+      "loss": 0.9029,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019864798900271532,
+      "loss": 0.7534,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019863775045875204,
+      "loss": 0.9456,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019862747355938452,
+      "loss": 0.8277,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019861715830860895,
+      "loss": 0.8423,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019860680471043643,
+      "loss": 0.8546,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198596412768893,
+      "loss": 0.8326,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019858598248801953,
+      "loss": 0.7939,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985755138718719,
+      "loss": 0.7727,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985650069245208,
+      "loss": 0.7735,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019855446165005186,
+      "loss": 0.8447,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019854387805256563,
+      "loss": 0.7694,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985332561361776,
+      "loss": 0.9685,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019852259590501806,
+      "loss": 0.8606,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985118973632323,
+      "loss": 0.7599,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019850116051498042,
+      "loss": 0.8513,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019849038536443747,
+      "loss": 0.8061,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984795719157934,
+      "loss": 0.9101,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198468720173253,
+      "loss": 0.7812,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198457830141036,
+      "loss": 0.8807,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019844690182337698,
+      "loss": 0.8003,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019843593522452543,
+      "loss": 0.7834,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984249303487458,
+      "loss": 0.7968,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019841388720031727,
+      "loss": 0.7313,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019840280578353397,
+      "loss": 0.7413,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019839168610270497,
+      "loss": 0.9299,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019838052816215416,
+      "loss": 0.7926,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019836933196622033,
+      "loss": 0.85,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983580975192571,
+      "loss": 0.9117,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000198346824825633,
+      "loss": 0.8205,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019833551388973147,
+      "loss": 0.7638,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019832416471595075,
+      "loss": 0.8679,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019831277730870394,
+      "loss": 0.8863,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019830135167241914,
+      "loss": 0.7653,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019828988781153917,
+      "loss": 0.8478,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019827838573052174,
+      "loss": 0.8279,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019826684543383947,
+      "loss": 0.8182,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982552669259799,
+      "loss": 0.7777,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982436502114452,
+      "loss": 0.8075,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019823199529475261,
+      "loss": 0.7378,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019822030218043422,
+      "loss": 0.8399,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982085708730368,
+      "loss": 0.8065,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019819680137712214,
+      "loss": 0.8663,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019818499369726684,
+      "loss": 0.894,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019817314783806227,
+      "loss": 0.8022,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019816126380411476,
+      "loss": 0.9105,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019814934160004543,
+      "loss": 0.8313,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019813738123049021,
+      "loss": 0.8506,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019812538270009994,
+      "loss": 0.7622,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981133460135402,
+      "loss": 0.8555,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019810127117549157,
+      "loss": 0.9503,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019808915819064925,
+      "loss": 0.7068,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980770070637235,
+      "loss": 0.7978,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019806481779943922,
+      "loss": 0.7967,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019805259040253623,
+      "loss": 0.9074,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980403248777692,
+      "loss": 0.8076,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019802802122990758,
+      "loss": 0.7614,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019801567946373564,
+      "loss": 0.6846,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019800329958405253,
+      "loss": 0.7508,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979908815956721,
+      "loss": 0.7946,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979784255034232,
+      "loss": 0.7659,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019796593131214937,
+      "loss": 0.8474,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019795339902670893,
+      "loss": 0.8295,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979408286519751,
+      "loss": 0.805,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019792822019283594,
+      "loss": 0.7411,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979155736541942,
+      "loss": 0.8505,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019790288904096754,
+      "loss": 0.7258,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019789016635808837,
+      "loss": 0.7553,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001978774056105039,
+      "loss": 0.8094,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001978646068031762,
+      "loss": 0.8805,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019785176994108208,
+      "loss": 0.8135,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019783889502921316,
+      "loss": 0.8239,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001978259820725759,
+      "loss": 0.8666,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019781303107619148,
+      "loss": 0.778,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019780004204509596,
+      "loss": 0.7656,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019778701498434007,
+      "loss": 0.9149,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019777394989898945,
+      "loss": 0.7674,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019776084679412443,
+      "loss": 0.8535,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019774770567484022,
+      "loss": 0.8744,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019773452654624674,
+      "loss": 0.7913,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019772130941346868,
+      "loss": 0.8593,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019770805428164558,
+      "loss": 0.8761,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019769476115593168,
+      "loss": 0.8261,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019768143004149607,
+      "loss": 0.8816,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976680609435225,
+      "loss": 0.9343,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019765465386720963,
+      "loss": 0.9002,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019764120881777077,
+      "loss": 0.7591,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019762772580043407,
+      "loss": 0.8986,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019761420482044238,
+      "loss": 0.8562,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019760064588305345,
+      "loss": 0.8902,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019758704899353955,
+      "loss": 0.8514,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019757341415718794,
+      "loss": 0.9313,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019755974137930052,
+      "loss": 0.7524,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019754603066519396,
+      "loss": 0.8278,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019753228202019964,
+      "loss": 0.7587,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019751849544966384,
+      "loss": 0.777,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019750467095894738,
+      "loss": 0.791,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019749080855342602,
+      "loss": 0.9554,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001974769082384901,
+      "loss": 0.8691,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001974629700195448,
+      "loss": 0.7651,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019744899390201006,
+      "loss": 0.8743,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019743497989132042,
+      "loss": 0.87,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019742092799292528,
+      "loss": 0.8734,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001974068382122888,
+      "loss": 0.8726,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973927105548897,
+      "loss": 0.8263,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019737854502622164,
+      "loss": 0.8439,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019736434163179282,
+      "loss": 0.9115,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973501003771263,
+      "loss": 0.7814,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019733582126775975,
+      "loss": 0.8396,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973215043092457,
+      "loss": 0.8948,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019730714950715123,
+      "loss": 0.9026,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972927568670583,
+      "loss": 0.912,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019727832639456345,
+      "loss": 0.7538,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000197263858095278,
+      "loss": 0.8032,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019724935197482798,
+      "loss": 0.7164,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972348080388541,
+      "loss": 0.8365,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972202262930118,
+      "loss": 0.8633,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019720560674297118,
+      "loss": 0.8429,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019719094939441706,
+      "loss": 0.7253,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019717625425304903,
+      "loss": 0.7515,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019716152132458127,
+      "loss": 0.7771,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019714675061474274,
+      "loss": 0.8063,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019713194212927696,
+      "loss": 0.9514,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019711709587394224,
+      "loss": 0.939,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019710221185451166,
+      "loss": 0.8652,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019708729007677283,
+      "loss": 1.0151,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019707233054652807,
+      "loss": 0.7764,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019705733326959445,
+      "loss": 0.8174,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019704229825180368,
+      "loss": 0.746,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019702722549900213,
+      "loss": 0.8539,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019701211501705083,
+      "loss": 0.7976,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019699696681182557,
+      "loss": 0.834,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969817808892167,
+      "loss": 0.7577,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019696655725512933,
+      "loss": 0.7145,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019695129591548315,
+      "loss": 0.8048,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019693599687621252,
+      "loss": 0.7091,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019692066014326655,
+      "loss": 0.7325,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019690528572260894,
+      "loss": 0.7954,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000196889873620218,
+      "loss": 0.7596,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019687442384208682,
+      "loss": 0.7823,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000196858936394223,
+      "loss": 0.7082,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019684341128264885,
+      "loss": 0.8912,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019682784851340136,
+      "loss": 0.8913,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019681224809253215,
+      "loss": 0.858,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019679661002610743,
+      "loss": 0.811,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967809343202081,
+      "loss": 0.8679,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019676522098092963,
+      "loss": 0.7506,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967494700143822,
+      "loss": 0.9627,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019673368142669064,
+      "loss": 0.7389,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019671785522399425,
+      "loss": 0.8909,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019670199141244717,
+      "loss": 0.7478,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019668608999821806,
+      "loss": 0.7812,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019667015098749012,
+      "loss": 0.7126,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966541743864613,
+      "loss": 0.7651,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019663816020134412,
+      "loss": 0.8624,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019662210843836574,
+      "loss": 0.7964,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019660601910376782,
+      "loss": 0.8941,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019658989220380684,
+      "loss": 0.9246,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965737277447537,
+      "loss": 0.852,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019655752573289394,
+      "loss": 0.7558,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965412861745278,
+      "loss": 0.8287,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019652500907596995,
+      "loss": 0.8632,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019650869444354986,
+      "loss": 0.8108,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964923422836115,
+      "loss": 0.871,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019647595260251333,
+      "loss": 0.7347,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019645952540662856,
+      "loss": 0.9728,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964430607023449,
+      "loss": 0.8235,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964265584960647,
+      "loss": 0.9119,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019641001879420482,
+      "loss": 0.8325,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019639344160319677,
+      "loss": 0.8082,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019637682692948664,
+      "loss": 0.8815,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000196360174779535,
+      "loss": 0.7525,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963434851598171,
+      "loss": 0.8904,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963267580768227,
+      "loss": 0.8057,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019630999353705618,
+      "loss": 0.7946,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001962931915470364,
+      "loss": 0.748,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019627635211329686,
+      "loss": 0.7896,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019625947524238563,
+      "loss": 0.8082,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019624256094086522,
+      "loss": 0.6715,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019622560921531284,
+      "loss": 0.8853,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019620862007232014,
+      "loss": 0.8397,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019619159351849338,
+      "loss": 0.7822,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019617452956045343,
+      "loss": 0.8117,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019615742820483555,
+      "loss": 0.7478,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019614028945828962,
+      "loss": 0.8115,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019612311332748008,
+      "loss": 0.8734,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961058998190859,
+      "loss": 0.8472,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019608864893980057,
+      "loss": 0.8557,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019607136069633212,
+      "loss": 0.9451,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019605403509540307,
+      "loss": 0.8364,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019603667214375053,
+      "loss": 0.7687,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019601927184812612,
+      "loss": 0.9463,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019600183421529592,
+      "loss": 0.8574,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959843592520406,
+      "loss": 0.8525,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019596684696515532,
+      "loss": 0.7865,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019594929736144976,
+      "loss": 0.8382,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019593171044774804,
+      "loss": 1.0194,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000195914086230889,
+      "loss": 0.755,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019589642471772565,
+      "loss": 0.8587,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958787259151258,
+      "loss": 0.915,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958609898299717,
+      "loss": 0.8164,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958432164691599,
+      "loss": 0.9353,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958254058396017,
+      "loss": 0.8034,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019580755794822277,
+      "loss": 0.8161,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019578967280196326,
+      "loss": 0.7718,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001957717504077778,
+      "loss": 0.792,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001957537907726356,
+      "loss": 0.9444,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019573579390352028,
+      "loss": 0.7905,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001957177598074299,
+      "loss": 0.7561,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019569968849137705,
+      "loss": 0.8497,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019568157996238884,
+      "loss": 0.8526,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001956634342275067,
+      "loss": 0.8054,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019564525129378668,
+      "loss": 0.7596,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019562703116829923,
+      "loss": 0.8958,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019560877385812924,
+      "loss": 0.8913,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019559047937037612,
+      "loss": 0.8185,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019557214771215366,
+      "loss": 0.817,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955537788905902,
+      "loss": 0.8395,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019553537291282843,
+      "loss": 0.8261,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955169297860256,
+      "loss": 0.7834,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019549844951735325,
+      "loss": 0.8471,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954799321139975,
+      "loss": 0.8872,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954613775831589,
+      "loss": 0.7833,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019544278593205237,
+      "loss": 0.801,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019542415716790727,
+      "loss": 0.8582,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019540549129796745,
+      "loss": 0.8319,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019538678832949113,
+      "loss": 0.7644,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019536804826975103,
+      "loss": 0.7332,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019534927112603416,
+      "loss": 0.7622,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019533045690564213,
+      "loss": 0.841,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019531160561589078,
+      "loss": 0.7598,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019529271726411053,
+      "loss": 0.8936,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019527379185764612,
+      "loss": 0.8244,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019525482940385672,
+      "loss": 0.824,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019523582991011585,
+      "loss": 0.9465,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001952167933838115,
+      "loss": 0.8087,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001951977198323461,
+      "loss": 0.6899,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019517860926313642,
+      "loss": 0.8386,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019515946168361357,
+      "loss": 0.7289,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019514027710122313,
+      "loss": 0.8429,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019512105552342505,
+      "loss": 0.8417,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019510179695769367,
+      "loss": 0.857,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950825014115177,
+      "loss": 0.9167,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019506316889240027,
+      "loss": 0.9209,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950437994078588,
+      "loss": 0.7913,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019502439296542514,
+      "loss": 0.7809,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019500494957264554,
+      "loss": 0.823,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001949854692370806,
+      "loss": 0.9008,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019496595196630517,
+      "loss": 0.831,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001949463977679087,
+      "loss": 0.885,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019492680664949478,
+      "loss": 0.8001,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019490717861868147,
+      "loss": 0.8809,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948875136831011,
+      "loss": 0.892,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948678118504005,
+      "loss": 0.7714,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019484807312824067,
+      "loss": 0.8833,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019482829752429703,
+      "loss": 0.696,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948084850462594,
+      "loss": 0.7979,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019478863570183188,
+      "loss": 0.7868,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019476874949873288,
+      "loss": 0.966,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019474882644469516,
+      "loss": 0.8578,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019472886654746586,
+      "loss": 0.8,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019470886981480642,
+      "loss": 0.9097,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019468883625449256,
+      "loss": 0.8457,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019466876587431435,
+      "loss": 0.7471,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001946486586820762,
+      "loss": 0.9344,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001946285146855968,
+      "loss": 1.0653,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001946083338927092,
+      "loss": 0.8458,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019458811631126069,
+      "loss": 0.7703,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019456786194911292,
+      "loss": 0.7772,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001945475708141418,
+      "loss": 0.7361,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019452724291423756,
+      "loss": 0.9043,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019450687825730477,
+      "loss": 0.6742,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944864768512622,
+      "loss": 0.8571,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019446603870404297,
+      "loss": 0.8461,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019444556382359454,
+      "loss": 0.8085,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944250522178785,
+      "loss": 0.77,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944045038948709,
+      "loss": 0.8131,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001943839188625619,
+      "loss": 0.7971,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019436329712895604,
+      "loss": 0.6855,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001943426387020721,
+      "loss": 0.9092,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019432194358994317,
+      "loss": 0.8274,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019430121180061654,
+      "loss": 0.9359,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001942804433421538,
+      "loss": 0.752,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019425963822263076,
+      "loss": 0.7997,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019423879645013756,
+      "loss": 0.9048,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001942179180327785,
+      "loss": 0.7083,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019419700297867225,
+      "loss": 0.8732,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019417605129595157,
+      "loss": 0.7897,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001941550629927636,
+      "loss": 0.8542,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019413403807726963,
+      "loss": 0.8669,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019411297655764527,
+      "loss": 0.7723,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019409187844208025,
+      "loss": 0.8392,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019407074373877867,
+      "loss": 0.7819,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940495724559587,
+      "loss": 0.791,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019402836460185292,
+      "loss": 0.821,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019400712018470794,
+      "loss": 0.8605,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019398583921278472,
+      "loss": 0.8187,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001939645216943584,
+      "loss": 0.706,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001939431676377183,
+      "loss": 0.8988,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019392177705116796,
+      "loss": 0.7863,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019390034994302516,
+      "loss": 0.7509,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019387888632162184,
+      "loss": 0.7905,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019385738619530414,
+      "loss": 0.8405,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019383584957243243,
+      "loss": 0.8825,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001938142764613813,
+      "loss": 0.7175,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937926668705394,
+      "loss": 0.7741,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937710208083096,
+      "loss": 0.9028,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019374933828310918,
+      "loss": 0.7914,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019372761930336926,
+      "loss": 0.8043,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937058638775353,
+      "loss": 0.8395,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000193684072014067,
+      "loss": 0.7716,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001936622437214381,
+      "loss": 0.8745,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019364037900813656,
+      "loss": 0.816,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019361847788266454,
+      "loss": 0.8226,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019359654035353824,
+      "loss": 0.7887,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019357456642928814,
+      "loss": 0.6934,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019355255611845884,
+      "loss": 0.7995,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019353050942960906,
+      "loss": 0.8599,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001935084263713117,
+      "loss": 0.7992,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001934863069521537,
+      "loss": 0.9085,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019346415118073632,
+      "loss": 0.8074,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019344195906567487,
+      "loss": 0.8951,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019341973061559866,
+      "loss": 0.813,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019339746583915134,
+      "loss": 0.7939,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019337516474499057,
+      "loss": 0.8125,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019335282734178817,
+      "loss": 0.882,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019333045363823006,
+      "loss": 0.6576,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019330804364301626,
+      "loss": 0.8276,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000193285597364861,
+      "loss": 0.7483,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001932631148124924,
+      "loss": 0.8933,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019324059599465295,
+      "loss": 0.8101,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019321804092009906,
+      "loss": 0.8496,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019319544959760135,
+      "loss": 0.8244,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001931728220359444,
+      "loss": 0.7134,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019315015824392703,
+      "loss": 0.8016,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019312745823036204,
+      "loss": 0.7398,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019310472200407638,
+      "loss": 0.6835,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019308194957391107,
+      "loss": 0.9115,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019305914094872118,
+      "loss": 0.8504,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001930362961373759,
+      "loss": 0.8769,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001930134151487584,
+      "loss": 0.7788,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019299049799176602,
+      "loss": 0.734,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019296754467531014,
+      "loss": 0.9509,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001929445552083161,
+      "loss": 0.8496,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019292152959972348,
+      "loss": 0.706,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928984678584858,
+      "loss": 0.9267,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928753699935706,
+      "loss": 0.8838,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019285223601395952,
+      "loss": 0.8494,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019282906592864825,
+      "loss": 0.7971,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928058597466465,
+      "loss": 0.82,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019278261747697804,
+      "loss": 0.8499,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019275933912868062,
+      "loss": 0.8375,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001927360247108061,
+      "loss": 0.7974,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019271267423242024,
+      "loss": 0.8226,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019268928770260296,
+      "loss": 0.8184,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019266586513044817,
+      "loss": 0.8554,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001926424065250637,
+      "loss": 0.7673,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019261891189557146,
+      "loss": 0.9081,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001925953812511074,
+      "loss": 0.7936,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019257181460082142,
+      "loss": 0.7491,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019254821195387742,
+      "loss": 0.838,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019252457331945334,
+      "loss": 0.8203,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019250089870674112,
+      "loss": 0.7344,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019247718812494662,
+      "loss": 0.7223,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019245344158328972,
+      "loss": 0.8257,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019242965909100433,
+      "loss": 0.8785,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019240584065733827,
+      "loss": 0.8532,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019238198629155335,
+      "loss": 0.8753,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923580960029254,
+      "loss": 0.7818,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923341698007442,
+      "loss": 0.8665,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019231020769431341,
+      "loss": 0.7909,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001922862096929508,
+      "loss": 0.8952,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019226217580598802,
+      "loss": 0.8959,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001922381060427706,
+      "loss": 0.8523,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019221400041265815,
+      "loss": 0.8291,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921898589250242,
+      "loss": 0.7738,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019216568158925612,
+      "loss": 0.8784,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019214146841475536,
+      "loss": 0.8463,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921172194109372,
+      "loss": 0.7691,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019209293458723092,
+      "loss": 0.6997,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019206861395307972,
+      "loss": 0.8271,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019204425751794064,
+      "loss": 0.7712,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019201986529128476,
+      "loss": 0.8477,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000191995437282597,
+      "loss": 0.8656,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019197097350137624,
+      "loss": 0.9166,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019194647395713523,
+      "loss": 0.7936,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001919219386594007,
+      "loss": 0.784,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019189736761771316,
+      "loss": 0.7842,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019187276084162714,
+      "loss": 0.8204,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019184811834071095,
+      "loss": 0.7393,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001918234401245469,
+      "loss": 0.892,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019179872620273116,
+      "loss": 0.8105,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019177397658487374,
+      "loss": 0.8483,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019174919128059855,
+      "loss": 0.7424,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001917243702995434,
+      "loss": 0.7617,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019169951365135998,
+      "loss": 0.9177,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019167462134571378,
+      "loss": 0.7401,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019164969339228422,
+      "loss": 0.7358,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001916247298007646,
+      "loss": 0.749,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019159973058086198,
+      "loss": 0.7927,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019157469574229737,
+      "loss": 0.9215,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001915496252948056,
+      "loss": 0.8526,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001915245192481353,
+      "loss": 0.911,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019149937761204906,
+      "loss": 0.6995,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019147420039632317,
+      "loss": 0.7766,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019144898761074786,
+      "loss": 0.7257,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019142373926512711,
+      "loss": 0.7624,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019139845536927884,
+      "loss": 0.8614,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019137313593303463,
+      "loss": 0.8455,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019134778096624002,
+      "loss": 0.8139,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019132239047875432,
+      "loss": 0.86,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019129696448045067,
+      "loss": 0.7719,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019127150298121594,
+      "loss": 0.9012,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019124600599095093,
+      "loss": 0.7499,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019122047351957013,
+      "loss": 0.8162,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001911949055770019,
+      "loss": 0.8629,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019116930217318834,
+      "loss": 0.7428,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019114366331808536,
+      "loss": 0.921,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001911179890216627,
+      "loss": 0.8398,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019109227929390378,
+      "loss": 0.7731,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001910665341448059,
+      "loss": 0.7101,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001910407535843801,
+      "loss": 0.842,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019101493762265114,
+      "loss": 0.8032,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019098908626965762,
+      "loss": 0.8351,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019096319953545185,
+      "loss": 0.9239,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019093727743009993,
+      "loss": 0.9068,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019091131996368168,
+      "loss": 0.8995,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001908853271462907,
+      "loss": 0.7885,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019085929898803437,
+      "loss": 0.8289,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001908332354990337,
+      "loss": 0.8943,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019080713668942356,
+      "loss": 0.7237,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019078100256935247,
+      "loss": 0.8345,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019075483314898276,
+      "loss": 0.9355,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019072862843849035,
+      "loss": 0.7665,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019070238844806506,
+      "loss": 0.728,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001906761131879103,
+      "loss": 0.8455,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019064980266824324,
+      "loss": 0.8009,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001906234568992948,
+      "loss": 0.7312,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019059707589130946,
+      "loss": 0.8418,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001905706596545456,
+      "loss": 0.8271,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019054420819927521,
+      "loss": 0.6729,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019051772153578389,
+      "loss": 0.8403,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019049119967437105,
+      "loss": 0.8364,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019046464262534977,
+      "loss": 0.7688,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019043805039904677,
+      "loss": 0.8539,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019041142300580245,
+      "loss": 0.7942,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019038476045597094,
+      "loss": 0.8635,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019035806275991996,
+      "loss": 0.8418,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019033132992803099,
+      "loss": 0.841,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001903045619706991,
+      "loss": 0.8571,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019027775889833302,
+      "loss": 0.8108,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019025092072135516,
+      "loss": 0.8073,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019022404745020163,
+      "loss": 0.828,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019019713909532206,
+      "loss": 0.8367,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019017019566717984,
+      "loss": 0.7423,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019014321717625196,
+      "loss": 0.7879,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019011620363302897,
+      "loss": 0.8257,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019008915504801523,
+      "loss": 0.7937,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019006207143172852,
+      "loss": 0.8426,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019003495279470037,
+      "loss": 0.9203,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019000779914747588,
+      "loss": 0.8296,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001899806105006138,
+      "loss": 0.7896,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018995338686468647,
+      "loss": 0.8052,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018992612825027976,
+      "loss": 0.7874,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018989883466799333,
+      "loss": 0.8992,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018987150612844028,
+      "loss": 0.8217,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018984414264224726,
+      "loss": 0.8072,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001898167442200547,
+      "loss": 0.8651,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018978931087251647,
+      "loss": 0.7628,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018976184261030004,
+      "loss": 0.8224,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018973433944408652,
+      "loss": 0.6592,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001897068013845705,
+      "loss": 0.7806,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018967922844246023,
+      "loss": 0.8055,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018965162062847742,
+      "loss": 0.7887,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001896239779533575,
+      "loss": 0.857,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018959630042784926,
+      "loss": 0.8404,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018956858806271518,
+      "loss": 0.8122,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018954084086873123,
+      "loss": 0.8886,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018951305885668695,
+      "loss": 0.8408,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018948524203738542,
+      "loss": 0.8045,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018945739042164323,
+      "loss": 0.8539,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018942950402029053,
+      "loss": 0.8227,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018940158284417094,
+      "loss": 0.7538,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018937362690414164,
+      "loss": 0.8897,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001893456362110734,
+      "loss": 0.815,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018931761077585035,
+      "loss": 0.7294,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018928955060937028,
+      "loss": 0.8375,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018926145572254437,
+      "loss": 0.8792,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018923332612629738,
+      "loss": 0.7892,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001892051618315675,
+      "loss": 0.839,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018917696284930652,
+      "loss": 0.7791,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018914872919047958,
+      "loss": 0.759,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018912046086606537,
+      "loss": 0.806,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001890921578870561,
+      "loss": 0.8368,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001890638202644574,
+      "loss": 0.7989,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001890354480092884,
+      "loss": 0.8493,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018900704113258165,
+      "loss": 0.7569,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018897859964538323,
+      "loss": 0.8958,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018895012355875262,
+      "loss": 0.7348,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018892161288376283,
+      "loss": 0.8985,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888930676315002,
+      "loss": 0.7298,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018886448781306464,
+      "loss": 0.7011,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018883587343956939,
+      "loss": 0.8753,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018880722452214123,
+      "loss": 0.9008,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018877854107192033,
+      "loss": 0.8927,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018874982310006022,
+      "loss": 0.8048,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018872107061772799,
+      "loss": 0.8566,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018869228363610404,
+      "loss": 0.7238,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018866346216638225,
+      "loss": 0.7806,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018863460621976983,
+      "loss": 0.8111,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001886057158074875,
+      "loss": 0.7991,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885767909407693,
+      "loss": 0.8756,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018854783163086268,
+      "loss": 0.8176,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018851883788902857,
+      "loss": 0.9303,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018848980972654115,
+      "loss": 0.7602,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001884607471546881,
+      "loss": 0.7656,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018843165018477044,
+      "loss": 0.7716,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001884025188281025,
+      "loss": 0.8259,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018837335309601213,
+      "loss": 0.7951,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018834415299984035,
+      "loss": 0.7721,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018831491855094174,
+      "loss": 0.8297,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018828564976068412,
+      "loss": 0.7751,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001882563466404487,
+      "loss": 0.8625,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018822700920163,
+      "loss": 0.7755,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018819763745563597,
+      "loss": 0.8322,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001881682314138878,
+      "loss": 0.8363,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018813879108782014,
+      "loss": 0.8643,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018810931648888078,
+      "loss": 0.873,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018807980762853107,
+      "loss": 0.793,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018805026451824546,
+      "loss": 0.75,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018802068716951186,
+      "loss": 0.8548,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001879910755938315,
+      "loss": 0.8551,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018796142980271886,
+      "loss": 0.7696,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018793174980770167,
+      "loss": 0.8553,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001879020356203211,
+      "loss": 0.856,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018787228725213154,
+      "loss": 0.9647,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001878425047147007,
+      "loss": 0.7828,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018781268801960945,
+      "loss": 0.866,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018778283717845216,
+      "loss": 0.8528,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018775295220283634,
+      "loss": 0.7924,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018772303310438275,
+      "loss": 0.8491,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001876930798947255,
+      "loss": 0.8457,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018766309258551193,
+      "loss": 0.7063,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018763307118840262,
+      "loss": 0.9853,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001876030157150715,
+      "loss": 0.6824,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018757292617720557,
+      "loss": 0.838,
+      "step": 964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018754280258650522,
+      "loss": 0.8543,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001875126449546841,
+      "loss": 0.7855,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018748245329346897,
+      "loss": 0.8967,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018745222761459992,
+      "loss": 0.7815,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018742196792983023,
+      "loss": 0.8521,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018739167425092644,
+      "loss": 0.9105,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018736134658966824,
+      "loss": 0.7819,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001873309849578486,
+      "loss": 0.9291,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018730058936727368,
+      "loss": 0.7584,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018727015982976286,
+      "loss": 0.8476,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018723969635714863,
+      "loss": 0.738,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018720919896127685,
+      "loss": 0.6796,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018717866765400638,
+      "loss": 0.8156,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871481024472094,
+      "loss": 0.8035,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018711750335277118,
+      "loss": 0.7973,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018708687038259026,
+      "loss": 0.7089,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018705620354857833,
+      "loss": 0.8793,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018702550286266013,
+      "loss": 0.8946,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001869947683367738,
+      "loss": 0.8421,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018696399998287033,
+      "loss": 0.8516,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018693319781291413,
+      "loss": 0.8441,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018690236183888263,
+      "loss": 0.7955,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018687149207276648,
+      "loss": 0.7998,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018684058852656938,
+      "loss": 0.8136,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001868096512123082,
+      "loss": 0.7698,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000186778680142013,
+      "loss": 0.7792,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018674767532772687,
+      "loss": 0.8661,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018671663678150607,
+      "loss": 0.8855,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018668556451542,
+      "loss": 0.8925,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018665445854155116,
+      "loss": 0.8144,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018662331887199507,
+      "loss": 0.8333,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018659214551886052,
+      "loss": 0.7224,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018656093849426925,
+      "loss": 0.8437,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018652969781035616,
+      "loss": 0.7843,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018649842347926923,
+      "loss": 0.7747,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001864671155131695,
+      "loss": 0.8336,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018643577392423114,
+      "loss": 0.7588,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018640439872464134,
+      "loss": 0.7193,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001863729899266004,
+      "loss": 0.8756,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018634154754232167,
+      "loss": 0.7053,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018631007158403156,
+      "loss": 0.7586,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018627856206396949,
+      "loss": 0.8873,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018624701899438802,
+      "loss": 0.8711,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001862154423875527,
+      "loss": 0.6552,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001861838322557421,
+      "loss": 0.7215,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018615218861124788,
+      "loss": 0.8258,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018612051146637475,
+      "loss": 0.7871,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001860888008334403,
+      "loss": 0.8587,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018605705672477537,
+      "loss": 0.7448,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001860252791527236,
+      "loss": 0.8262,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018599346812964181,
+      "loss": 0.8097,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018596162366789966,
+      "loss": 0.7844,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018592974577987998,
+      "loss": 0.8862,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001858978344779785,
+      "loss": 0.8626,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000185865889774604,
+      "loss": 0.8478,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018583391168217815,
+      "loss": 0.7642,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001858019002131357,
+      "loss": 0.7514,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018576985537992438,
+      "loss": 0.7234,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001857377771950048,
+      "loss": 0.7618,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018570566567085065,
+      "loss": 0.7084,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018567352081994852,
+      "loss": 0.7808,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018564134265479796,
+      "loss": 0.8448,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001856091311879115,
+      "loss": 0.8159,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001855768864318146,
+      "loss": 0.8882,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018554460839904566,
+      "loss": 0.8075,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018551229710215604,
+      "loss": 0.876,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018547995255371004,
+      "loss": 0.8231,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018544757476628485,
+      "loss": 0.7901,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018541516375247062,
+      "loss": 0.7975,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018538271952487043,
+      "loss": 0.8114,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018535024209610025,
+      "loss": 0.9073,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018531773147878895,
+      "loss": 0.8159,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018528518768557833,
+      "loss": 0.7765,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018525261072912304,
+      "loss": 0.7783,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018522000062209078,
+      "loss": 0.9497,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001851873573771619,
+      "loss": 0.727,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018515468100702983,
+      "loss": 0.8592,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001851219715244008,
+      "loss": 0.8661,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018508922894199393,
+      "loss": 0.9118,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018505645327254123,
+      "loss": 0.8398,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018502364452878753,
+      "loss": 0.8801,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018499080272349057,
+      "loss": 0.8928,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001849579278694209,
+      "loss": 0.8362,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018492501997936195,
+      "loss": 0.8465,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018489207906611003,
+      "loss": 0.8175,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001848591051424742,
+      "loss": 0.782,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018482609822127649,
+      "loss": 0.8174,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018479305831535156,
+      "loss": 0.7552,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001847599854375471,
+      "loss": 0.7865,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018472687960072352,
+      "loss": 0.9156,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018469374081775405,
+      "loss": 0.8454,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018466056910152476,
+      "loss": 0.7574,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001846273644649345,
+      "loss": 0.8492,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018459412692089494,
+      "loss": 0.8088,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018456085648233054,
+      "loss": 0.8074,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018452755316217853,
+      "loss": 0.8184,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018449421697338893,
+      "loss": 0.9,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018446084792892462,
+      "loss": 0.8921,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018442744604176117,
+      "loss": 0.8503,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001843940113248869,
+      "loss": 0.8218,
+      "step": 1064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018436054379130295,
+      "loss": 0.7818,
+      "step": 1065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018432704345402323,
+      "loss": 0.8568,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018429351032607443,
+      "loss": 0.746,
+      "step": 1067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018425994442049587,
+      "loss": 0.8706,
+      "step": 1068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001842263457503397,
+      "loss": 0.8648,
+      "step": 1069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001841927143286708,
+      "loss": 0.8261,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018415905016856682,
+      "loss": 0.7925,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018412535328311814,
+      "loss": 0.7776,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018409162368542776,
+      "loss": 0.7346,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018405786138861148,
+      "loss": 0.7489,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018402406640579785,
+      "loss": 0.8415,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001839902387501281,
+      "loss": 0.8244,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018395637843475606,
+      "loss": 0.8811,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018392248547284844,
+      "loss": 0.8115,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018388855987758456,
+      "loss": 0.7971,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018385460166215638,
+      "loss": 0.8089,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001838206108397686,
+      "loss": 0.7371,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001837865874236386,
+      "loss": 0.785,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018375253142699646,
+      "loss": 0.8319,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018371844286308483,
+      "loss": 0.7772,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018368432174515915,
+      "loss": 0.7576,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001836501680864874,
+      "loss": 0.8699,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001836159819003503,
+      "loss": 0.7792,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018358176320004119,
+      "loss": 0.8582,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018354751199886607,
+      "loss": 0.8022,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018351322831014347,
+      "loss": 0.7819,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018347891214720477,
+      "loss": 0.8399,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018344456352339377,
+      "loss": 0.7692,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018341018245206697,
+      "loss": 0.7916,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018337576894659354,
+      "loss": 0.7319,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018334132302035515,
+      "loss": 0.7288,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001833068446867462,
+      "loss": 0.8207,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018327233395917358,
+      "loss": 0.9212,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018323779085105684,
+      "loss": 0.7109,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018320321537582815,
+      "loss": 0.8265,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018316860754693213,
+      "loss": 0.727,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001831339673778262,
+      "loss": 0.8073,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018309929488198012,
+      "loss": 0.7583,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018306459007287637,
+      "loss": 0.7087,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018302985296401,
+      "loss": 0.7789,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001829950835688885,
+      "loss": 0.763,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018296028190103203,
+      "loss": 0.8218,
+      "step": 1106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018292544797397325,
+      "loss": 0.8547,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001828905818012574,
+      "loss": 0.8877,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001828556833964422,
+      "loss": 0.8403,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001828207527730979,
+      "loss": 0.8151,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001827857899448074,
+      "loss": 0.8383,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018275079492516597,
+      "loss": 0.7537,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018271576772778154,
+      "loss": 0.8007,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001826807083662744,
+      "loss": 0.7478,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001826456168542774,
+      "loss": 0.7622,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000182610493205436,
+      "loss": 0.8015,
+      "step": 1116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018257533743340804,
+      "loss": 0.7607,
+      "step": 1117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018254014955186386,
+      "loss": 0.7407,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018250492957448634,
+      "loss": 0.7795,
+      "step": 1119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018246967751497082,
+      "loss": 0.8873,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018243439338702503,
+      "loss": 0.8149,
+      "step": 1121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018239907720436933,
+      "loss": 0.7614,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018236372898073643,
+      "loss": 0.8493,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018232834872987147,
+      "loss": 0.771,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018229293646553223,
+      "loss": 0.8542,
+      "step": 1125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018225749220148867,
+      "loss": 0.9387,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018222201595152342,
+      "loss": 0.8924,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018218650772943145,
+      "loss": 0.9183,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018215096754902013,
+      "loss": 0.7453,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018211539542410936,
+      "loss": 0.7526,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018207979136853136,
+      "loss": 0.8613,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018204415539613085,
+      "loss": 0.8854,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018200848752076486,
+      "loss": 0.7886,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018197278775630292,
+      "loss": 0.7654,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018193705611662696,
+      "loss": 0.7849,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001819012926156312,
+      "loss": 0.8787,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018186549726722235,
+      "loss": 0.6751,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018182967008531948,
+      "loss": 0.7596,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000181793811083854,
+      "loss": 0.7294,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018175792027676978,
+      "loss": 0.879,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000181721997678023,
+      "loss": 0.729,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018168604330158212,
+      "loss": 0.7521,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018165005716142812,
+      "loss": 0.7071,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018161403927155422,
+      "loss": 0.8946,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018157798964596605,
+      "loss": 0.7791,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001815419082986815,
+      "loss": 0.7721,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001815057952437309,
+      "loss": 0.7845,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018146965049515684,
+      "loss": 0.8252,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018143347406701418,
+      "loss": 0.8085,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018139726597337025,
+      "loss": 0.8626,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018136102622830458,
+      "loss": 0.8099,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018132475484590902,
+      "loss": 0.8185,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018128845184028778,
+      "loss": 0.7123,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018125211722555727,
+      "loss": 0.81,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001812157510158463,
+      "loss": 0.8261,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001811793532252959,
+      "loss": 0.8665,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018114292386805936,
+      "loss": 0.8522,
+      "step": 1157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001811064629583023,
+      "loss": 0.8598,
+      "step": 1158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001810699705102026,
+      "loss": 0.9054,
+      "step": 1159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001810334465379504,
+      "loss": 0.8068,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018099689105574804,
+      "loss": 0.7353,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001809603040778102,
+      "loss": 0.8945,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018092368561836378,
+      "loss": 0.7143,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018088703569164787,
+      "loss": 0.8239,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018085035431191385,
+      "loss": 0.9075,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018081364149342535,
+      "loss": 0.7542,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018077689725045815,
+      "loss": 0.8501,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018074012159730032,
+      "loss": 0.8493,
+      "step": 1168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018070331454825208,
+      "loss": 0.8095,
+      "step": 1169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018066647611762594,
+      "loss": 0.6746,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018062960631974657,
+      "loss": 0.7455,
+      "step": 1171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001805927051689508,
+      "loss": 0.9047,
+      "step": 1172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018055577267958768,
+      "loss": 0.8218,
+      "step": 1173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018051880886601844,
+      "loss": 0.7378,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018048181374261658,
+      "loss": 0.8714,
+      "step": 1175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001804447873237676,
+      "loss": 0.7605,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018040772962386935,
+      "loss": 0.7679,
+      "step": 1177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001803706406573317,
+      "loss": 0.7341,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018033352043857675,
+      "loss": 0.8195,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018029636898203875,
+      "loss": 0.9468,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001802591863021641,
+      "loss": 0.6776,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018022197241341132,
+      "loss": 0.7709,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018018472733025108,
+      "loss": 0.8587,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018014745106716612,
+      "loss": 0.712,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018011014363865144,
+      "loss": 0.7528,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018007280505921402,
+      "loss": 0.9146,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018003543534337302,
+      "loss": 0.7305,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017999803450565973,
+      "loss": 0.8298,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017996060256061748,
+      "loss": 0.8553,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017992313952280172,
+      "loss": 0.719,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017988564540678,
+      "loss": 0.7807,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017984812022713196,
+      "loss": 0.8119,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017981056399844934,
+      "loss": 0.7889,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017977297673533587,
+      "loss": 0.8205,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017973535845240744,
+      "loss": 0.8289,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001796977091642919,
+      "loss": 0.7225,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001796600288856293,
+      "loss": 0.6791,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001796223176310716,
+      "loss": 0.7316,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001795845754152829,
+      "loss": 0.8097,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017954680225293928,
+      "loss": 0.7284,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017950899815872892,
+      "loss": 0.7407,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017947116314735194,
+      "loss": 0.7384,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017943329723352057,
+      "loss": 0.7562,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017939540043195897,
+      "loss": 0.8944,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001793574727574034,
+      "loss": 0.8006,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017931951422460202,
+      "loss": 0.7747,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001792815248483151,
+      "loss": 0.7966,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017924350464331487,
+      "loss": 0.7063,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001792054536243855,
+      "loss": 0.8706,
+      "step": 1209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017916737180632316,
+      "loss": 0.7421,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017912925920393602,
+      "loss": 0.833,
+      "step": 1211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017909111583204422,
+      "loss": 0.8415,
+      "step": 1212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001790529417054798,
+      "loss": 0.7839,
+      "step": 1213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017901473683908694,
+      "loss": 0.8765,
+      "step": 1214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017897650124772154,
+      "loss": 0.9031,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017893823494625157,
+      "loss": 0.7628,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017889993794955694,
+      "loss": 0.7525,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001788616102725295,
+      "loss": 0.7494,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017882325193007299,
+      "loss": 0.798,
+      "step": 1219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017878486293710307,
+      "loss": 0.7565,
+      "step": 1220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017874644330854742,
+      "loss": 0.8138,
+      "step": 1221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017870799305934546,
+      "loss": 0.8712,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001786695122044487,
+      "loss": 0.8801,
+      "step": 1223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017863100075882042,
+      "loss": 0.722,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017859245873743586,
+      "loss": 0.7262,
+      "step": 1225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017855388615528213,
+      "loss": 0.8731,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017851528302735822,
+      "loss": 0.8207,
+      "step": 1227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017847664936867494,
+      "loss": 0.7944,
+      "step": 1228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017843798519425513,
+      "loss": 0.9224,
+      "step": 1229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017839929051913337,
+      "loss": 0.7562,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017836056535835607,
+      "loss": 0.8163,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017832180972698165,
+      "loss": 0.7147,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017828302364008024,
+      "loss": 0.8035,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001782442071127338,
+      "loss": 0.8095,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017820536016003625,
+      "loss": 0.7363,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017816648279709324,
+      "loss": 0.8285,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017812757503902232,
+      "loss": 0.9139,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017808863690095281,
+      "loss": 0.7789,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017804966839802579,
+      "loss": 0.7477,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017801066954539424,
+      "loss": 0.8452,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017797164035822296,
+      "loss": 0.7522,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017793258085168842,
+      "loss": 0.9189,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000177893491040979,
+      "loss": 0.7588,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001778543709412948,
+      "loss": 0.692,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001778152205678477,
+      "loss": 0.8209,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001777760399358614,
+      "loss": 0.7603,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017773682906057134,
+      "loss": 0.7835,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001776975879572247,
+      "loss": 0.8421,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001776583166410804,
+      "loss": 0.7767,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017761901512740918,
+      "loss": 0.7606,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017757968343149347,
+      "loss": 0.8051,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017754032156862742,
+      "loss": 0.8026,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017750092955411698,
+      "loss": 0.7543,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017746150740327975,
+      "loss": 0.7961,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001774220551314451,
+      "loss": 0.7527,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017738257275395404,
+      "loss": 0.7592,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001773430602861594,
+      "loss": 0.7362,
+      "step": 1257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001773035177434256,
+      "loss": 0.7995,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017726394514112884,
+      "loss": 0.7195,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017722434249465695,
+      "loss": 0.8471,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017718470981940954,
+      "loss": 0.7477,
+      "step": 1261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001771450471307977,
+      "loss": 0.7861,
+      "step": 1262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017710535444424437,
+      "loss": 0.8538,
+      "step": 1263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017706563177518413,
+      "loss": 0.8421,
+      "step": 1264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017702587913906312,
+      "loss": 0.7705,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017698609655133923,
+      "loss": 0.6895,
+      "step": 1266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017694628402748202,
+      "loss": 0.7508,
+      "step": 1267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017690644158297256,
+      "loss": 0.8051,
+      "step": 1268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017686656923330364,
+      "loss": 0.7761,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001768266669939797,
+      "loss": 0.8029,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017678673488051675,
+      "loss": 0.8137,
+      "step": 1271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017674677290844247,
+      "loss": 0.7011,
+      "step": 1272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017670678109329608,
+      "loss": 0.6732,
+      "step": 1273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017666675945062846,
+      "loss": 0.8318,
+      "step": 1274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017662670799600206,
+      "loss": 0.8238,
+      "step": 1275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017658662674499098,
+      "loss": 0.696,
+      "step": 1276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001765465157131808,
+      "loss": 0.7233,
+      "step": 1277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001765063749161688,
+      "loss": 0.9003,
+      "step": 1278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001764662043695637,
+      "loss": 0.7928,
+      "step": 1279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017642600408898597,
+      "loss": 0.8167,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017638577409006745,
+      "loss": 0.8075,
+      "step": 1281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017634551438845163,
+      "loss": 0.7884,
+      "step": 1282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001763052249997936,
+      "loss": 0.7829,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017626490593975987,
+      "loss": 0.8468,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017622455722402856,
+      "loss": 0.868,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017618417886828934,
+      "loss": 0.7524,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017614377088824337,
+      "loss": 0.8656,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017610333329960338,
+      "loss": 0.7903,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017606286611809353,
+      "loss": 0.703,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017602236935944954,
+      "loss": 0.7988,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001759818430394186,
+      "loss": 0.7208,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017594128717375945,
+      "loss": 0.8613,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017590070177824228,
+      "loss": 0.7738,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017586008686864874,
+      "loss": 0.8677,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017581944246077203,
+      "loss": 0.7768,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017577876857041672,
+      "loss": 0.8037,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017573806521339896,
+      "loss": 0.8322,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017569733240554628,
+      "loss": 0.7677,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017565657016269765,
+      "loss": 0.7791,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017561577850070355,
+      "loss": 0.8761,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017557495743542585,
+      "loss": 0.7577,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017553410698273786,
+      "loss": 0.7071,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001754932271585243,
+      "loss": 0.6975,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017545231797868143,
+      "loss": 0.768,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017541137945911673,
+      "loss": 0.8203,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001753704116157493,
+      "loss": 0.7424,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001753294144645094,
+      "loss": 0.9403,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017528838802133892,
+      "loss": 0.8725,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000175247332302191,
+      "loss": 0.7452,
+      "step": 1309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017520624732303024,
+      "loss": 0.6295,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017516513309983253,
+      "loss": 0.8463,
+      "step": 1311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017512398964858523,
+      "loss": 0.7559,
+      "step": 1312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000175082816985287,
+      "loss": 0.8343,
+      "step": 1313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017504161512594784,
+      "loss": 0.9025,
+      "step": 1314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017500038408658922,
+      "loss": 0.7504,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001749591238832438,
+      "loss": 0.7329,
+      "step": 1316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001749178345319557,
+      "loss": 0.8222,
+      "step": 1317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017487651604878034,
+      "loss": 0.8026,
+      "step": 1318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017483516844978445,
+      "loss": 0.7455,
+      "step": 1319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001747937917510461,
+      "loss": 0.7917,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017475238596865463,
+      "loss": 0.6788,
+      "step": 1321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017471095111871074,
+      "loss": 0.8046,
+      "step": 1322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017466948721732645,
+      "loss": 0.7711,
+      "step": 1323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017462799428062504,
+      "loss": 0.8014,
+      "step": 1324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017458647232474104,
+      "loss": 0.7622,
+      "step": 1325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017454492136582042,
+      "loss": 0.8394,
+      "step": 1326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017450334142002022,
+      "loss": 0.7813,
+      "step": 1327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001744617325035088,
+      "loss": 0.7306,
+      "step": 1328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017442009463246598,
+      "loss": 0.8193,
+      "step": 1329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017437842782308261,
+      "loss": 0.8207,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017433673209156088,
+      "loss": 0.7601,
+      "step": 1331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017429500745411427,
+      "loss": 0.8976,
+      "step": 1332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001742532539269674,
+      "loss": 0.8324,
+      "step": 1333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017421147152635623,
+      "loss": 0.7652,
+      "step": 1334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017416966026852785,
+      "loss": 0.7124,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017412782016974058,
+      "loss": 0.8706,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001740859512462641,
+      "loss": 0.884,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017404405351437912,
+      "loss": 0.7405,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001740021269903776,
+      "loss": 0.8677,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017396017169056278,
+      "loss": 0.7568,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000173918187631249,
+      "loss": 0.7751,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001738761748287618,
+      "loss": 0.6534,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001738341332994379,
+      "loss": 0.8207,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017379206305962526,
+      "loss": 0.8107,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017374996412568286,
+      "loss": 0.7613,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017370783651398094,
+      "loss": 0.7745,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017366568024090092,
+      "loss": 0.7419,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001736234953228353,
+      "loss": 0.8273,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017358128177618766,
+      "loss": 0.8054,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017353903961737287,
+      "loss": 0.8244,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017349676886281682,
+      "loss": 0.8958,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001734544695289565,
+      "loss": 0.9362,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017341214163224016,
+      "loss": 0.8592,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017336978518912695,
+      "loss": 0.6941,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017332740021608722,
+      "loss": 0.7323,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017328498672960252,
+      "loss": 0.7703,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017324254474616527,
+      "loss": 0.8522,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017320007428227915,
+      "loss": 0.7741,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017315757535445886,
+      "loss": 0.7129,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017311504797923007,
+      "loss": 0.7576,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001730724921731297,
+      "loss": 0.7583,
+      "step": 1361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017302990795270556,
+      "loss": 0.7417,
+      "step": 1362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001729872953345166,
+      "loss": 0.7837,
+      "step": 1363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017294465433513278,
+      "loss": 0.7002,
+      "step": 1364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017290198497113513,
+      "loss": 0.7556,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017285928725911562,
+      "loss": 0.8405,
+      "step": 1366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001728165612156774,
+      "loss": 0.7134,
+      "step": 1367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017277380685743442,
+      "loss": 0.8852,
+      "step": 1368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017273102420101188,
+      "loss": 0.7661,
+      "step": 1369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017268821326304578,
+      "loss": 0.7786,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017264537406018324,
+      "loss": 0.7812,
+      "step": 1371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017260250660908232,
+      "loss": 0.8596,
+      "step": 1372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001725596109264121,
+      "loss": 0.84,
+      "step": 1373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001725166870288526,
+      "loss": 0.7837,
+      "step": 1374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017247373493309485,
+      "loss": 0.7511,
+      "step": 1375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017243075465584078,
+      "loss": 0.8836,
+      "step": 1376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017238774621380337,
+      "loss": 0.7084,
+      "step": 1377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017234470962370642,
+      "loss": 0.8299,
+      "step": 1378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017230164490228487,
+      "loss": 0.9272,
+      "step": 1379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001722585520662844,
+      "loss": 0.7426,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001722154311324618,
+      "loss": 0.7661,
+      "step": 1381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001721722821175846,
+      "loss": 0.9009,
+      "step": 1382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017212910503843135,
+      "loss": 0.7698,
+      "step": 1383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001720858999117916,
+      "loss": 0.8855,
+      "step": 1384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017204266675446562,
+      "loss": 0.7403,
+      "step": 1385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001719994055832647,
+      "loss": 0.7438,
+      "step": 1386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000171956116415011,
+      "loss": 0.7675,
+      "step": 1387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017191279926653761,
+      "loss": 0.822,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017186945415468836,
+      "loss": 0.7774,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017182608109631814,
+      "loss": 0.7417,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017178268010829254,
+      "loss": 0.7322,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017173925120748815,
+      "loss": 0.7213,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017169579441079228,
+      "loss": 0.656,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001716523097351032,
+      "loss": 0.825,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017160879719733,
+      "loss": 0.7174,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017156525681439252,
+      "loss": 0.7609,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017152168860322156,
+      "loss": 0.838,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001714780925807586,
+      "loss": 0.7952,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017143446876395602,
+      "loss": 0.7692,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017139081716977702,
+      "loss": 0.7867,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017134713781519556,
+      "loss": 0.6875,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001713034307171964,
+      "loss": 0.8517,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001712596958927751,
+      "loss": 0.7362,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000171215933358938,
+      "loss": 0.762,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017117214313270223,
+      "loss": 0.7033,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001711283252310956,
+      "loss": 0.8872,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017108447967115686,
+      "loss": 0.7255,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017104060646993535,
+      "loss": 0.7802,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017099670564449116,
+      "loss": 0.8436,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017095277721189528,
+      "loss": 0.7454,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001709088211892293,
+      "loss": 0.854,
+      "step": 1411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017086483759358554,
+      "loss": 0.8968,
+      "step": 1412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017082082644206706,
+      "loss": 0.7611,
+      "step": 1413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017077678775178772,
+      "loss": 1.0092,
+      "step": 1414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017073272153987198,
+      "loss": 0.8094,
+      "step": 1415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017068862782345498,
+      "loss": 0.8008,
+      "step": 1416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001706445066196827,
+      "loss": 0.7405,
+      "step": 1417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001706003579457117,
+      "loss": 0.7075,
+      "step": 1418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001705561818187092,
+      "loss": 0.8171,
+      "step": 1419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017051197825585318,
+      "loss": 0.8052,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017046774727433222,
+      "loss": 0.7004,
+      "step": 1421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001704234888913456,
+      "loss": 0.8062,
+      "step": 1422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017037920312410325,
+      "loss": 0.791,
+      "step": 1423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017033488998982575,
+      "loss": 0.7895,
+      "step": 1424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001702905495057443,
+      "loss": 0.6929,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017024618168910073,
+      "loss": 0.8037,
+      "step": 1426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017020178655714754,
+      "loss": 0.6972,
+      "step": 1427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017015736412714786,
+      "loss": 0.8015,
+      "step": 1428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017011291441637537,
+      "loss": 0.7635,
+      "step": 1429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017006843744211438,
+      "loss": 0.7793,
+      "step": 1430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001700239332216598,
+      "loss": 0.7021,
+      "step": 1431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016997940177231722,
+      "loss": 0.83,
+      "step": 1432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001699348431114027,
+      "loss": 0.7822,
+      "step": 1433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001698902572562429,
+      "loss": 0.9027,
+      "step": 1434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001698456442241751,
+      "loss": 0.7243,
+      "step": 1435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016980100403254716,
+      "loss": 0.8783,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016975633669871746,
+      "loss": 0.732,
+      "step": 1437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001697116422400549,
+      "loss": 0.8226,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016966692067393904,
+      "loss": 0.8361,
+      "step": 1439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016962217201775984,
+      "loss": 0.8383,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016957739628891796,
+      "loss": 0.7816,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016953259350482443,
+      "loss": 0.8198,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016948776368290084,
+      "loss": 0.786,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016944290684057936,
+      "loss": 0.72,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016939802299530265,
+      "loss": 0.7843,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001693531121645238,
+      "loss": 0.753,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016930817436570646,
+      "loss": 0.812,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016926320961632474,
+      "loss": 0.7635,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016921821793386324,
+      "loss": 0.8189,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016917319933581698,
+      "loss": 0.7505,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016912815383969155,
+      "loss": 0.8439,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016908308146300298,
+      "loss": 0.7747,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016903798222327765,
+      "loss": 0.775,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016899285613805246,
+      "loss": 0.7762,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016894770322487476,
+      "loss": 0.7384,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016890252350130227,
+      "loss": 0.8135,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001688573169849032,
+      "loss": 0.7325,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016881208369325622,
+      "loss": 0.7743,
+      "step": 1458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001687668236439503,
+      "loss": 0.9351,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016872153685458483,
+      "loss": 0.6859,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016867622334276966,
+      "loss": 0.8179,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000168630883126125,
+      "loss": 0.864,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016858551622228146,
+      "loss": 0.8495,
+      "step": 1463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016854012264888,
+      "loss": 0.7495,
+      "step": 1464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016849470242357196,
+      "loss": 0.8333,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016844925556401904,
+      "loss": 0.7035,
+      "step": 1466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016840378208789333,
+      "loss": 0.7832,
+      "step": 1467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016835828201287723,
+      "loss": 0.8319,
+      "step": 1468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001683127553566635,
+      "loss": 0.6857,
+      "step": 1469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016826720213695523,
+      "loss": 0.7642,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016822162237146583,
+      "loss": 0.8161,
+      "step": 1471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016817601607791908,
+      "loss": 0.8222,
+      "step": 1472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016813038327404903,
+      "loss": 0.8666,
+      "step": 1473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001680847239776,
+      "loss": 0.9637,
+      "step": 1474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016803903820632673,
+      "loss": 0.7775,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016799332597799413,
+      "loss": 0.9069,
+      "step": 1476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001679475873103775,
+      "loss": 0.8753,
+      "step": 1477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016790182222126235,
+      "loss": 0.7322,
+      "step": 1478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016785603072844447,
+      "loss": 0.9481,
+      "step": 1479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016781021284972998,
+      "loss": 0.7468,
+      "step": 1480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001677643686029352,
+      "loss": 0.8297,
+      "step": 1481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016771849800588666,
+      "loss": 0.8022,
+      "step": 1482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016767260107642128,
+      "loss": 0.83,
+      "step": 1483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001676266778323861,
+      "loss": 0.7954,
+      "step": 1484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016758072829163841,
+      "loss": 0.7111,
+      "step": 1485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016753475247204577,
+      "loss": 0.768,
+      "step": 1486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016748875039148593,
+      "loss": 0.7642,
+      "step": 1487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016744272206784685,
+      "loss": 0.827,
+      "step": 1488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016739666751902667,
+      "loss": 0.7702,
+      "step": 1489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001673505867629338,
+      "loss": 0.7804,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016730447981748672,
+      "loss": 0.8452,
+      "step": 1491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016725834670061425,
+      "loss": 0.8194,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001672121874302553,
+      "loss": 0.6982,
+      "step": 1493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016716600202435888,
+      "loss": 0.7673,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016711979050088434,
+      "loss": 0.865,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000167073552877801,
+      "loss": 0.7446,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016702728917308845,
+      "loss": 0.7966,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001669809994047364,
+      "loss": 0.9252,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016693468359074468,
+      "loss": 0.9316,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016688834174912324,
+      "loss": 0.8575,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016684197389789214,
+      "loss": 0.7494,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016679558005508157,
+      "loss": 0.8017,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016674916023873184,
+      "loss": 0.832,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016670271446689338,
+      "loss": 0.7881,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016665624275762663,
+      "loss": 0.7308,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016660974512900225,
+      "loss": 0.8473,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001665632215991008,
+      "loss": 0.8649,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016651667218601306,
+      "loss": 1.0703,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001664700969078398,
+      "loss": 0.7893,
+      "step": 1509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001664234957826919,
+      "loss": 0.7582,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001663768688286903,
+      "loss": 0.7786,
+      "step": 1511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001663302160639658,
+      "loss": 0.7049,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016628353750665955,
+      "loss": 0.8418,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016623683317492245,
+      "loss": 0.8273,
+      "step": 1514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016619010308691556,
+      "loss": 0.8978,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016614334726080998,
+      "loss": 0.7214,
+      "step": 1516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016609656571478675,
+      "loss": 0.7932,
+      "step": 1517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016604975846703685,
+      "loss": 0.7632,
+      "step": 1518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001660029255357614,
+      "loss": 0.8747,
+      "step": 1519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016595606693917142,
+      "loss": 0.8135,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016590918269548797,
+      "loss": 0.7405,
+      "step": 1521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016586227282294196,
+      "loss": 0.8312,
+      "step": 1522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016581533733977448,
+      "loss": 0.7732,
+      "step": 1523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016576837626423632,
+      "loss": 0.8589,
+      "step": 1524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016572138961458842,
+      "loss": 0.8681,
+      "step": 1525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016567437740910154,
+      "loss": 0.8466,
+      "step": 1526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016562733966605646,
+      "loss": 0.825,
+      "step": 1527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016558027640374388,
+      "loss": 0.8174,
+      "step": 1528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016553318764046438,
+      "loss": 0.8154,
+      "step": 1529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016548607339452853,
+      "loss": 0.9366,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016543893368425666,
+      "loss": 0.793,
+      "step": 1531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001653917685279792,
+      "loss": 0.8431,
+      "step": 1532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016534457794403632,
+      "loss": 0.7261,
+      "step": 1533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016529736195077815,
+      "loss": 0.7826,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016525012056656463,
+      "loss": 0.8081,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016520285380976574,
+      "loss": 0.8541,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016515556169876115,
+      "loss": 0.8353,
+      "step": 1537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016510824425194042,
+      "loss": 0.8595,
+      "step": 1538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016506090148770306,
+      "loss": 0.7801,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016501353342445832,
+      "loss": 0.7972,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016496614008062533,
+      "loss": 0.7897,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016491872147463306,
+      "loss": 0.7481,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016487127762492026,
+      "loss": 0.7804,
+      "step": 1543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016482380854993559,
+      "loss": 0.7532,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001647763142681374,
+      "loss": 0.7477,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016472879479799388,
+      "loss": 0.8663,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016468125015798308,
+      "loss": 0.7974,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016463368036659278,
+      "loss": 0.7627,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016458608544232052,
+      "loss": 0.8501,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016453846540367368,
+      "loss": 0.7854,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016449082026916934,
+      "loss": 0.6854,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016444315005733438,
+      "loss": 0.7811,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016439545478670543,
+      "loss": 0.7608,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001643477344758288,
+      "loss": 0.7499,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016429998914326065,
+      "loss": 0.7155,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016425221880756675,
+      "loss": 0.8807,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001642044234873227,
+      "loss": 0.7619,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016415660320111375,
+      "loss": 0.8105,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001641087579675349,
+      "loss": 0.8929,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016406088780519072,
+      "loss": 0.7662,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001640129927326957,
+      "loss": 0.7007,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016396507276867386,
+      "loss": 0.8112,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016391712793175888,
+      "loss": 0.7946,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016386915824059427,
+      "loss": 0.8583,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000163821163713833,
+      "loss": 0.8415,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016377314437013784,
+      "loss": 0.9031,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016372510022818118,
+      "loss": 0.8324,
+      "step": 1567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000163677031306645,
+      "loss": 0.7805,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000163628937624221,
+      "loss": 0.752,
+      "step": 1569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016358081919961044,
+      "loss": 0.8047,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016353267605152423,
+      "loss": 0.8009,
+      "step": 1571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016348450819868288,
+      "loss": 0.7904,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016343631565981655,
+      "loss": 0.751,
+      "step": 1573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016338809845366491,
+      "loss": 0.7955,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016333985659897735,
+      "loss": 0.7716,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001632915901145127,
+      "loss": 0.802,
+      "step": 1576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016324329901903946,
+      "loss": 0.8436,
+      "step": 1577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001631949833313357,
+      "loss": 0.7659,
+      "step": 1578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016314664307018903,
+      "loss": 0.7661,
+      "step": 1579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001630982782543966,
+      "loss": 0.7191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016304988890276513,
+      "loss": 0.8997,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001630014750341109,
+      "loss": 0.7865,
+      "step": 1582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016295303666725964,
+      "loss": 0.9552,
+      "step": 1583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016290457382104668,
+      "loss": 0.7629,
+      "step": 1584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001628560865143169,
+      "loss": 0.7415,
+      "step": 1585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016280757476592466,
+      "loss": 0.7052,
+      "step": 1586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016275903859473372,
+      "loss": 0.8707,
+      "step": 1587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001627104780196175,
+      "loss": 0.8223,
+      "step": 1588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016266189305945877,
+      "loss": 0.7982,
+      "step": 1589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016261328373314994,
+      "loss": 0.755,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016256465005959277,
+      "loss": 0.8464,
+      "step": 1591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016251599205769845,
+      "loss": 0.8291,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001624673097463878,
+      "loss": 0.7361,
+      "step": 1593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016241860314459094,
+      "loss": 0.8333,
+      "step": 1594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016236987227124746,
+      "loss": 0.8494,
+      "step": 1595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016232111714530652,
+      "loss": 0.7968,
+      "step": 1596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001622723377857265,
+      "loss": 0.827,
+      "step": 1597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016222353421147537,
+      "loss": 0.6331,
+      "step": 1598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016217470644153045,
+      "loss": 0.8054,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016212585449487842,
+      "loss": 0.8148,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016207697839051553,
+      "loss": 0.6907,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016202807814744721,
+      "loss": 0.6693,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001619791537846884,
+      "loss": 0.7286,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016193020532126343,
+      "loss": 0.7692,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001618812327762059,
+      "loss": 0.8445,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001618322361685589,
+      "loss": 0.7248,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001617832155173748,
+      "loss": 0.7988,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016173417084171536,
+      "loss": 0.8193,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016168510216065164,
+      "loss": 0.8054,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016163600949326403,
+      "loss": 0.7714,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001615868928586423,
+      "loss": 0.8673,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016153775227588545,
+      "loss": 0.7226,
+      "step": 1612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016148858776410198,
+      "loss": 0.8931,
+      "step": 1613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001614393993424095,
+      "loss": 0.7608,
+      "step": 1614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016139018702993494,
+      "loss": 0.7259,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001613409508458146,
+      "loss": 0.8101,
+      "step": 1616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016129169080919404,
+      "loss": 0.8509,
+      "step": 1617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016124240693922808,
+      "loss": 0.7304,
+      "step": 1618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016119309925508078,
+      "loss": 0.8283,
+      "step": 1619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016114376777592556,
+      "loss": 0.7322,
+      "step": 1620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016109441252094495,
+      "loss": 0.7716,
+      "step": 1621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016104503350933082,
+      "loss": 0.7934,
+      "step": 1622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016099563076028427,
+      "loss": 0.8103,
+      "step": 1623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001609462042930156,
+      "loss": 0.7529,
+      "step": 1624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016089675412674437,
+      "loss": 0.6899,
+      "step": 1625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016084728028069933,
+      "loss": 0.7617,
+      "step": 1626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016079778277411843,
+      "loss": 0.8961,
+      "step": 1627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016074826162624882,
+      "loss": 0.8053,
+      "step": 1628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001606987168563469,
+      "loss": 0.8064,
+      "step": 1629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001606491484836782,
+      "loss": 0.7642,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001605995565275174,
+      "loss": 0.7671,
+      "step": 1631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001605499410071484,
+      "loss": 0.7216,
+      "step": 1632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001605003019418643,
+      "loss": 0.7782,
+      "step": 1633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016045063935096727,
+      "loss": 0.6951,
+      "step": 1634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016040095325376864,
+      "loss": 0.8703,
+      "step": 1635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000160351243669589,
+      "loss": 0.8869,
+      "step": 1636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016030151061775788,
+      "loss": 0.8229,
+      "step": 1637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001602517541176141,
+      "loss": 0.8359,
+      "step": 1638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016020197418850548,
+      "loss": 0.769,
+      "step": 1639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016015217084978906,
+      "loss": 0.7596,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016010234412083086,
+      "loss": 0.8224,
+      "step": 1641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016005249402100612,
+      "loss": 0.7877,
+      "step": 1642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001600026205696991,
+      "loss": 0.8037,
+      "step": 1643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015995272378630312,
+      "loss": 0.7254,
+      "step": 1644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015990280369022064,
+      "loss": 0.8104,
+      "step": 1645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015985286030086314,
+      "loss": 0.7582,
+      "step": 1646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015980289363765114,
+      "loss": 0.8757,
+      "step": 1647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015975290372001423,
+      "loss": 0.8047,
+      "step": 1648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015970289056739108,
+      "loss": 0.829,
+      "step": 1649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015965285419922934,
+      "loss": 0.7705,
+      "step": 1650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015960279463498575,
+      "loss": 0.7337,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015955271189412598,
+      "loss": 0.7855,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001595026059961248,
+      "loss": 0.7888,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001594524769604659,
+      "loss": 0.742,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015940232480664206,
+      "loss": 0.7443,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015935214955415495,
+      "loss": 0.8018,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015930195122251537,
+      "loss": 0.8186,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001592517298312429,
+      "loss": 0.8322,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015920148539986623,
+      "loss": 0.7503,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000159151217947923,
+      "loss": 0.7144,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001591009274949597,
+      "loss": 0.7693,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015905061406053185,
+      "loss": 0.9401,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015900027766420393,
+      "loss": 0.7768,
+      "step": 1663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015894991832554926,
+      "loss": 0.906,
+      "step": 1664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015889953606415013,
+      "loss": 0.8206,
+      "step": 1665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001588491308995978,
+      "loss": 0.8895,
+      "step": 1666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001587987028514923,
+      "loss": 0.789,
+      "step": 1667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015874825193944265,
+      "loss": 0.8742,
+      "step": 1668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001586977781830668,
+      "loss": 0.822,
+      "step": 1669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015864728160199144,
+      "loss": 0.8154,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001585967622158523,
+      "loss": 0.8905,
+      "step": 1671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015854622004429393,
+      "loss": 0.7778,
+      "step": 1672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001584956551069696,
+      "loss": 0.7542,
+      "step": 1673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015844506742354164,
+      "loss": 0.7453,
+      "step": 1674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015839445701368106,
+      "loss": 0.7806,
+      "step": 1675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001583438238970678,
+      "loss": 0.7168,
+      "step": 1676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015829316809339068,
+      "loss": 0.8565,
+      "step": 1677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015824248962234715,
+      "loss": 0.7753,
+      "step": 1678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001581917885036437,
+      "loss": 0.6988,
+      "step": 1679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015814106475699546,
+      "loss": 0.7074,
+      "step": 1680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001580903184021264,
+      "loss": 0.8297,
+      "step": 1681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015803954945876932,
+      "loss": 0.911,
+      "step": 1682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015798875794666582,
+      "loss": 0.7455,
+      "step": 1683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015793794388556617,
+      "loss": 0.7552,
+      "step": 1684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015788710729522953,
+      "loss": 0.7559,
+      "step": 1685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015783624819542378,
+      "loss": 0.7522,
+      "step": 1686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015778536660592546,
+      "loss": 0.8207,
+      "step": 1687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015773446254652002,
+      "loss": 0.6915,
+      "step": 1688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015768353603700153,
+      "loss": 0.8903,
+      "step": 1689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015763258709717282,
+      "loss": 0.752,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015758161574684545,
+      "loss": 0.8325,
+      "step": 1691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015753062200583967,
+      "loss": 0.7703,
+      "step": 1692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015747960589398448,
+      "loss": 0.7657,
+      "step": 1693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001574285674311175,
+      "loss": 0.6389,
+      "step": 1694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015737750663708522,
+      "loss": 0.8562,
+      "step": 1695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015732642353174259,
+      "loss": 0.8242,
+      "step": 1696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015727531813495332,
+      "loss": 0.7243,
+      "step": 1697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015722419046658987,
+      "loss": 0.7563,
+      "step": 1698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015717304054653326,
+      "loss": 0.8072,
+      "step": 1699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001571218683946732,
+      "loss": 0.8445,
+      "step": 1700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015707067403090803,
+      "loss": 0.8034,
+      "step": 1701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015701945747514478,
+      "loss": 0.7936,
+      "step": 1702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015696821874729898,
+      "loss": 0.7173,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015691695786729498,
+      "loss": 0.6299,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001568656748550655,
+      "loss": 0.6927,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015681436973055216,
+      "loss": 0.643,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001567630425137049,
+      "loss": 0.8087,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001567116932244824,
+      "loss": 0.6949,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001566603218828519,
+      "loss": 0.8297,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001566089285087892,
+      "loss": 0.7597,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015655751312227865,
+      "loss": 0.8023,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015650607574331323,
+      "loss": 0.8446,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015645461639189438,
+      "loss": 0.764,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015640313508803215,
+      "loss": 0.7772,
+      "step": 1714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001563516318517451,
+      "loss": 0.7637,
+      "step": 1715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015630010670306035,
+      "loss": 0.7926,
+      "step": 1716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015624855966201348,
+      "loss": 0.6572,
+      "step": 1717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015619699074864864,
+      "loss": 0.8187,
+      "step": 1718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015614539998301842,
+      "loss": 0.7874,
+      "step": 1719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015609378738518402,
+      "loss": 0.8057,
+      "step": 1720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000156042152975215,
+      "loss": 0.6892,
+      "step": 1721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015599049677318945,
+      "loss": 0.8208,
+      "step": 1722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000155938818799194,
+      "loss": 0.7676,
+      "step": 1723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015588711907332366,
+      "loss": 0.8202,
+      "step": 1724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015583539761568194,
+      "loss": 0.7652,
+      "step": 1725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001557836544463808,
+      "loss": 0.9497,
+      "step": 1726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015573188958554056,
+      "loss": 0.7822,
+      "step": 1727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015568010305329013,
+      "loss": 0.8824,
+      "step": 1728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015562829486976673,
+      "loss": 0.8273,
+      "step": 1729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015557646505511604,
+      "loss": 0.8153,
+      "step": 1730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015552461362949215,
+      "loss": 0.8412,
+      "step": 1731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001554727406130575,
+      "loss": 0.7713,
+      "step": 1732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015542084602598305,
+      "loss": 0.8472,
+      "step": 1733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000155368929888448,
+      "loss": 0.7503,
+      "step": 1734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015531699222064006,
+      "loss": 0.7199,
+      "step": 1735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001552650330427552,
+      "loss": 0.7979,
+      "step": 1736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015521305237499783,
+      "loss": 0.7501,
+      "step": 1737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015516105023758068,
+      "loss": 0.8203,
+      "step": 1738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015510902665072488,
+      "loss": 0.8042,
+      "step": 1739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015505698163465986,
+      "loss": 0.7061,
+      "step": 1740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015500491520962332,
+      "loss": 0.7036,
+      "step": 1741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001549528273958614,
+      "loss": 0.7913,
+      "step": 1742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015490071821362851,
+      "loss": 0.7677,
+      "step": 1743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015484858768318737,
+      "loss": 0.6761,
+      "step": 1744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015479643582480898,
+      "loss": 0.8591,
+      "step": 1745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015474426265877264,
+      "loss": 0.7697,
+      "step": 1746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015469206820536595,
+      "loss": 0.802,
+      "step": 1747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015463985248488483,
+      "loss": 0.6684,
+      "step": 1748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015458761551763337,
+      "loss": 0.9191,
+      "step": 1749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015453535732392397,
+      "loss": 0.8374,
+      "step": 1750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015448307792407734,
+      "loss": 0.7752,
+      "step": 1751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015443077733842236,
+      "loss": 0.7326,
+      "step": 1752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015437845558729618,
+      "loss": 0.7918,
+      "step": 1753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015432611269104414,
+      "loss": 0.6906,
+      "step": 1754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001542737486700199,
+      "loss": 0.8214,
+      "step": 1755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015422136354458524,
+      "loss": 0.8076,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015416895733511013,
+      "loss": 0.8036,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015411653006197284,
+      "loss": 0.9482,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015406408174555976,
+      "loss": 0.7035,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001540116124062655,
+      "loss": 0.8692,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001539591220644928,
+      "loss": 0.8277,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015390661074065256,
+      "loss": 0.86,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015385407845516395,
+      "loss": 0.7483,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015380152522845414,
+      "loss": 0.8051,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015374895108095858,
+      "loss": 0.834,
+      "step": 1765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015369635603312072,
+      "loss": 0.7709,
+      "step": 1766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015364374010539225,
+      "loss": 0.694,
+      "step": 1767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001535911033182329,
+      "loss": 0.8403,
+      "step": 1768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015353844569211054,
+      "loss": 0.6996,
+      "step": 1769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015348576724750123,
+      "loss": 0.7242,
+      "step": 1770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015343306800488896,
+      "loss": 0.7833,
+      "step": 1771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001533803479847659,
+      "loss": 0.8353,
+      "step": 1772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015332760720763232,
+      "loss": 0.7576,
+      "step": 1773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015327484569399656,
+      "loss": 0.745,
+      "step": 1774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015322206346437493,
+      "loss": 0.6685,
+      "step": 1775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001531692605392919,
+      "loss": 0.7208,
+      "step": 1776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015311643693927994,
+      "loss": 0.8363,
+      "step": 1777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015306359268487958,
+      "loss": 0.7315,
+      "step": 1778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015301072779663933,
+      "loss": 0.8308,
+      "step": 1779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015295784229511582,
+      "loss": 0.736,
+      "step": 1780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015290493620087362,
+      "loss": 0.7775,
+      "step": 1781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001528520095344853,
+      "loss": 0.7208,
+      "step": 1782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015279906231653145,
+      "loss": 0.7031,
+      "step": 1783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015274609456760073,
+      "loss": 0.7168,
+      "step": 1784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015269310630828961,
+      "loss": 0.7639,
+      "step": 1785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015264009755920274,
+      "loss": 0.8396,
+      "step": 1786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015258706834095256,
+      "loss": 0.7666,
+      "step": 1787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015253401867415956,
+      "loss": 0.8177,
+      "step": 1788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001524809485794522,
+      "loss": 0.8211,
+      "step": 1789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015242785807746678,
+      "loss": 0.6826,
+      "step": 1790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015237474718884765,
+      "loss": 0.7861,
+      "step": 1791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015232161593424704,
+      "loss": 0.7792,
+      "step": 1792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015226846433432509,
+      "loss": 0.9449,
+      "step": 1793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015221529240974987,
+      "loss": 0.7967,
+      "step": 1794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015216210018119733,
+      "loss": 0.7977,
+      "step": 1795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015210888766935136,
+      "loss": 0.7726,
+      "step": 1796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015205565489490367,
+      "loss": 0.7017,
+      "step": 1797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001520024018785539,
+      "loss": 0.8737,
+      "step": 1798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001519491286410096,
+      "loss": 0.7501,
+      "step": 1799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015189583520298604,
+      "loss": 0.6991,
+      "step": 1800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015184252158520646,
+      "loss": 0.7112,
+      "step": 1801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015178918780840196,
+      "loss": 0.8479,
+      "step": 1802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015173583389331143,
+      "loss": 0.7041,
+      "step": 1803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001516824598606816,
+      "loss": 0.8502,
+      "step": 1804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000151629065731267,
+      "loss": 0.8034,
+      "step": 1805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015157565152583002,
+      "loss": 0.7479,
+      "step": 1806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015152221726514082,
+      "loss": 0.8499,
+      "step": 1807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015146876296997738,
+      "loss": 0.7369,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001514152886611255,
+      "loss": 0.7933,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001513617943593787,
+      "loss": 0.9856,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001513082800855383,
+      "loss": 0.9473,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015125474586041336,
+      "loss": 0.8165,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015120119170482078,
+      "loss": 0.7696,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015114761763958514,
+      "loss": 0.7965,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015109402368553875,
+      "loss": 0.7556,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015104040986352172,
+      "loss": 0.8313,
+      "step": 1816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001509867761943818,
+      "loss": 0.8594,
+      "step": 1817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015093312269897458,
+      "loss": 0.8382,
+      "step": 1818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015087944939816322,
+      "loss": 0.7694,
+      "step": 1819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001508257563128186,
+      "loss": 0.7796,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015077204346381946,
+      "loss": 0.8602,
+      "step": 1821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015071831087205205,
+      "loss": 0.821,
+      "step": 1822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015066455855841028,
+      "loss": 0.8187,
+      "step": 1823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001506107865437959,
+      "loss": 0.7526,
+      "step": 1824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001505569948491182,
+      "loss": 0.9345,
+      "step": 1825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015050318349529415,
+      "loss": 0.733,
+      "step": 1826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001504493525032483,
+      "loss": 0.8931,
+      "step": 1827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015039550189391298,
+      "loss": 0.8775,
+      "step": 1828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015034163168822797,
+      "loss": 0.9367,
+      "step": 1829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015028774190714082,
+      "loss": 0.7193,
+      "step": 1830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015023383257160662,
+      "loss": 0.7636,
+      "step": 1831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015017990370258806,
+      "loss": 0.7569,
+      "step": 1832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015012595532105547,
+      "loss": 0.7964,
+      "step": 1833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001500719874479867,
+      "loss": 0.8335,
+      "step": 1834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001500180001043673,
+      "loss": 0.765,
+      "step": 1835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014996399331119027,
+      "loss": 0.6917,
+      "step": 1836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014990996708945617,
+      "loss": 0.778,
+      "step": 1837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014985592146017323,
+      "loss": 0.7645,
+      "step": 1838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001498018564443571,
+      "loss": 0.6768,
+      "step": 1839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014974777206303106,
+      "loss": 0.7395,
+      "step": 1840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014969366833722587,
+      "loss": 0.6207,
+      "step": 1841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014963954528797983,
+      "loss": 0.861,
+      "step": 1842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014958540293633874,
+      "loss": 0.7374,
+      "step": 1843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014953124130335593,
+      "loss": 0.7174,
+      "step": 1844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014947706041009223,
+      "loss": 0.7593,
+      "step": 1845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001494228602776159,
+      "loss": 0.8511,
+      "step": 1846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014936864092700278,
+      "loss": 0.6801,
+      "step": 1847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014931440237933604,
+      "loss": 0.817,
+      "step": 1848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001492601446557065,
+      "loss": 0.7219,
+      "step": 1849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001492058677772123,
+      "loss": 0.7671,
+      "step": 1850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001491515717649591,
+      "loss": 0.7018,
+      "step": 1851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014909725664005987,
+      "loss": 0.7247,
+      "step": 1852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001490429224236352,
+      "loss": 0.7512,
+      "step": 1853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014898856913681297,
+      "loss": 0.6663,
+      "step": 1854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014893419680072855,
+      "loss": 0.7416,
+      "step": 1855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014887980543652467,
+      "loss": 0.7357,
+      "step": 1856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001488253950653515,
+      "loss": 0.7512,
+      "step": 1857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014877096570836652,
+      "loss": 0.8244,
+      "step": 1858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014871651738673468,
+      "loss": 0.8328,
+      "step": 1859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001486620501216283,
+      "loss": 0.7,
+      "step": 1860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000148607563934227,
+      "loss": 0.7326,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014855305884571782,
+      "loss": 0.8953,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014849853487729512,
+      "loss": 0.7898,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001484439920501606,
+      "loss": 0.8134,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001483894303855233,
+      "loss": 0.7614,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014833484990459963,
+      "loss": 0.7682,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001482802506286132,
+      "loss": 0.7945,
+      "step": 1867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014822563257879508,
+      "loss": 0.6957,
+      "step": 1868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014817099577638348,
+      "loss": 0.83,
+      "step": 1869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001481163402426241,
+      "loss": 0.9322,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014806166599876967,
+      "loss": 0.7678,
+      "step": 1871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014800697306608044,
+      "loss": 0.8238,
+      "step": 1872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014795226146582376,
+      "loss": 0.8781,
+      "step": 1873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014789753121927435,
+      "loss": 0.8083,
+      "step": 1874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014784278234771406,
+      "loss": 0.7466,
+      "step": 1875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014778801487243213,
+      "loss": 0.8326,
+      "step": 1876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001477332288147249,
+      "loss": 0.8289,
+      "step": 1877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014767842419589598,
+      "loss": 0.7645,
+      "step": 1878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001476236010372563,
+      "loss": 0.7555,
+      "step": 1879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014756875936012385,
+      "loss": 0.8015,
+      "step": 1880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014751389918582386,
+      "loss": 0.7283,
+      "step": 1881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001474590205356888,
+      "loss": 0.7467,
+      "step": 1882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014740412343105828,
+      "loss": 0.7846,
+      "step": 1883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014734920789327913,
+      "loss": 0.7673,
+      "step": 1884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014729427394370533,
+      "loss": 0.733,
+      "step": 1885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014723932160369793,
+      "loss": 0.6799,
+      "step": 1886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014718435089462528,
+      "loss": 0.8114,
+      "step": 1887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001471293618378628,
+      "loss": 0.7931,
+      "step": 1888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014707435445479304,
+      "loss": 0.6887,
+      "step": 1889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014701932876680571,
+      "loss": 0.8028,
+      "step": 1890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014696428479529759,
+      "loss": 0.7636,
+      "step": 1891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014690922256167257,
+      "loss": 0.8156,
+      "step": 1892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014685414208734168,
+      "loss": 0.7713,
+      "step": 1893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014679904339372302,
+      "loss": 0.8633,
+      "step": 1894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001467439265022418,
+      "loss": 0.8068,
+      "step": 1895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001466887914343303,
+      "loss": 0.8644,
+      "step": 1896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014663363821142784,
+      "loss": 0.776,
+      "step": 1897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014657846685498077,
+      "loss": 0.8673,
+      "step": 1898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001465232773864426,
+      "loss": 0.831,
+      "step": 1899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014646806982727379,
+      "loss": 0.7378,
+      "step": 1900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014641284419894188,
+      "loss": 0.7291,
+      "step": 1901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001463576005229214,
+      "loss": 0.7944,
+      "step": 1902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014630233882069395,
+      "loss": 0.8708,
+      "step": 1903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001462470591137481,
+      "loss": 0.8314,
+      "step": 1904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014619176142357935,
+      "loss": 0.7931,
+      "step": 1905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014613644577169034,
+      "loss": 0.7771,
+      "step": 1906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014608111217959068,
+      "loss": 0.7361,
+      "step": 1907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001460257606687968,
+      "loss": 0.7312,
+      "step": 1908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014597039126083223,
+      "loss": 0.7018,
+      "step": 1909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014591500397722747,
+      "loss": 0.7957,
+      "step": 1910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014585959883951988,
+      "loss": 0.7672,
+      "step": 1911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014580417586925384,
+      "loss": 0.6183,
+      "step": 1912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014574873508798063,
+      "loss": 0.8566,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014569327651725846,
+      "loss": 0.8011,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014563780017865248,
+      "loss": 0.7937,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001455823060937347,
+      "loss": 0.8051,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014552679428408402,
+      "loss": 0.871,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014547126477128637,
+      "loss": 0.7499,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001454157175769344,
+      "loss": 0.8153,
+      "step": 1919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014536015272262773,
+      "loss": 0.9623,
+      "step": 1920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014530457022997282,
+      "loss": 0.859,
+      "step": 1921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014524897012058296,
+      "loss": 0.7543,
+      "step": 1922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001451933524160783,
+      "loss": 0.7046,
+      "step": 1923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014513771713808594,
+      "loss": 0.7044,
+      "step": 1924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014508206430823963,
+      "loss": 0.6751,
+      "step": 1925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014502639394818007,
+      "loss": 0.7666,
+      "step": 1926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014497070607955476,
+      "loss": 0.7341,
+      "step": 1927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014491500072401795,
+      "loss": 0.7187,
+      "step": 1928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014485927790323079,
+      "loss": 0.8445,
+      "step": 1929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001448035376388611,
+      "loss": 0.7855,
+      "step": 1930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001447477799525836,
+      "loss": 0.8281,
+      "step": 1931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001446920048660797,
+      "loss": 0.8691,
+      "step": 1932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001446362124010377,
+      "loss": 0.877,
+      "step": 1933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001445804025791524,
+      "loss": 0.7023,
+      "step": 1934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001445245754221256,
+      "loss": 0.7227,
+      "step": 1935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014446873095166579,
+      "loss": 0.7708,
+      "step": 1936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001444128691894881,
+      "loss": 0.7077,
+      "step": 1937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014435699015731448,
+      "loss": 0.8191,
+      "step": 1938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001443010938768735,
+      "loss": 0.8233,
+      "step": 1939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001442451803699006,
+      "loss": 0.8319,
+      "step": 1940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014418924965813767,
+      "loss": 0.7025,
+      "step": 1941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014413330176333354,
+      "loss": 0.7095,
+      "step": 1942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001440773367072436,
+      "loss": 0.805,
+      "step": 1943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001440213545116299,
+      "loss": 0.6981,
+      "step": 1944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001439653551982612,
+      "loss": 0.7356,
+      "step": 1945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014390933878891287,
+      "loss": 0.7548,
+      "step": 1946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014385330530536697,
+      "loss": 0.7222,
+      "step": 1947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014379725476941222,
+      "loss": 0.8095,
+      "step": 1948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014374118720284388,
+      "loss": 0.7722,
+      "step": 1949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014368510262746393,
+      "loss": 0.7594,
+      "step": 1950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014362900106508093,
+      "loss": 0.7531,
+      "step": 1951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014357288253750996,
+      "loss": 0.8255,
+      "step": 1952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014351674706657288,
+      "loss": 0.8617,
+      "step": 1953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014346059467409798,
+      "loss": 0.7985,
+      "step": 1954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014340442538192017,
+      "loss": 0.7821,
+      "step": 1955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014334823921188097,
+      "loss": 0.7662,
+      "step": 1956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014329203618582845,
+      "loss": 0.8558,
+      "step": 1957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014323581632561715,
+      "loss": 0.7351,
+      "step": 1958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014317957965310834,
+      "loss": 0.8906,
+      "step": 1959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014312332619016965,
+      "loss": 0.671,
+      "step": 1960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014306705595867528,
+      "loss": 0.7326,
+      "step": 1961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000143010768980506,
+      "loss": 0.6602,
+      "step": 1962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001429544652775491,
+      "loss": 0.7635,
+      "step": 1963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001428981448716983,
+      "loss": 0.7783,
+      "step": 1964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014284180778485387,
+      "loss": 0.8682,
+      "step": 1965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014278545403892257,
+      "loss": 0.7064,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001427290836558176,
+      "loss": 0.7083,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014267269665745863,
+      "loss": 0.8153,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014261629306577182,
+      "loss": 0.797,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001425598729026898,
+      "loss": 0.7561,
+      "step": 1970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001425034361901516,
+      "loss": 0.7499,
+      "step": 1971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014244698295010267,
+      "loss": 0.856,
+      "step": 1972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000142390513204495,
+      "loss": 0.7182,
+      "step": 1973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014233402697528684,
+      "loss": 0.754,
+      "step": 1974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014227752428444293,
+      "loss": 0.7937,
+      "step": 1975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014222100515393445,
+      "loss": 0.7398,
+      "step": 1976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014216446960573891,
+      "loss": 0.8633,
+      "step": 1977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014210791766184025,
+      "loss": 0.835,
+      "step": 1978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001420513493442287,
+      "loss": 0.7271,
+      "step": 1979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014199476467490095,
+      "loss": 0.7476,
+      "step": 1980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014193816367585998,
+      "loss": 0.7655,
+      "step": 1981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014188154636911524,
+      "loss": 0.8306,
+      "step": 1982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014182491277668234,
+      "loss": 0.8322,
+      "step": 1983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014176826292058338,
+      "loss": 0.6456,
+      "step": 1984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001417115968228467,
+      "loss": 0.7711,
+      "step": 1985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014165491450550698,
+      "loss": 0.6696,
+      "step": 1986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001415982159906052,
+      "loss": 0.7793,
+      "step": 1987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014154150130018866,
+      "loss": 0.7408,
+      "step": 1988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014148477045631094,
+      "loss": 0.7394,
+      "step": 1989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014142802348103185,
+      "loss": 0.7593,
+      "step": 1990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014137126039641756,
+      "loss": 0.843,
+      "step": 1991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001413144812245404,
+      "loss": 0.5874,
+      "step": 1992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001412576859874791,
+      "loss": 0.6967,
+      "step": 1993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014120087470731853,
+      "loss": 0.802,
+      "step": 1994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014114404740614978,
+      "loss": 0.6969,
+      "step": 1995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001410872041060703,
+      "loss": 0.7678,
+      "step": 1996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014103034482918357,
+      "loss": 0.7736,
+      "step": 1997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014097346959759943,
+      "loss": 0.7105,
+      "step": 1998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001409165784334339,
+      "loss": 0.7738,
+      "step": 1999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014085967135880916,
+      "loss": 0.6754,
+      "step": 2000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001408027483958536,
+      "loss": 0.8113,
+      "step": 2001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014074580956670177,
+      "loss": 0.7924,
+      "step": 2002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014068885489349438,
+      "loss": 0.7568,
+      "step": 2003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014063188439837832,
+      "loss": 0.7761,
+      "step": 2004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014057489810350668,
+      "loss": 0.679,
+      "step": 2005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014051789603103864,
+      "loss": 0.9505,
+      "step": 2006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001404608782031395,
+      "loss": 0.7172,
+      "step": 2007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001404038446419807,
+      "loss": 0.8545,
+      "step": 2008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014034679536973984,
+      "loss": 0.8009,
+      "step": 2009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014028973040860056,
+      "loss": 0.8263,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014023264978075265,
+      "loss": 0.7641,
+      "step": 2011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000140175553508392,
+      "loss": 0.7529,
+      "step": 2012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014011844161372053,
+      "loss": 0.75,
+      "step": 2013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014006131411894627,
+      "loss": 0.7011,
+      "step": 2014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001400041710462833,
+      "loss": 0.7413,
+      "step": 2015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013994701241795186,
+      "loss": 0.7499,
+      "step": 2016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013988983825617804,
+      "loss": 0.7475,
+      "step": 2017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013983264858319415,
+      "loss": 0.7768,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013977544342123844,
+      "loss": 0.717,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013971822279255518,
+      "loss": 0.7632,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001396609867193947,
+      "loss": 0.812,
+      "step": 2021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001396037352240133,
+      "loss": 0.8141,
+      "step": 2022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001395464683286733,
+      "loss": 0.7067,
+      "step": 2023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000139489186055643,
+      "loss": 0.8249,
+      "step": 2024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013943188842719667,
+      "loss": 0.6545,
+      "step": 2025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001393745754656146,
+      "loss": 0.8079,
+      "step": 2026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013931724719318294,
+      "loss": 0.8125,
+      "step": 2027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001392599036321939,
+      "loss": 0.8068,
+      "step": 2028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013920254480494557,
+      "loss": 0.8346,
+      "step": 2029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000139145170733742,
+      "loss": 0.6912,
+      "step": 2030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013908778144089317,
+      "loss": 0.7674,
+      "step": 2031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000139030376948715,
+      "loss": 0.711,
+      "step": 2032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001389729572795292,
+      "loss": 0.8299,
+      "step": 2033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013891552245566356,
+      "loss": 0.8294,
+      "step": 2034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013885807249945165,
+      "loss": 0.7047,
+      "step": 2035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000138800607433233,
+      "loss": 0.8129,
+      "step": 2036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013874312727935292,
+      "loss": 0.8436,
+      "step": 2037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013868563206016264,
+      "loss": 0.8595,
+      "step": 2038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001386281217980192,
+      "loss": 0.7091,
+      "step": 2039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013857059651528562,
+      "loss": 0.9186,
+      "step": 2040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013851305623433063,
+      "loss": 0.8183,
+      "step": 2041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013845550097752883,
+      "loss": 0.6912,
+      "step": 2042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001383979307672607,
+      "loss": 0.8226,
+      "step": 2043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013834034562591236,
+      "loss": 0.8326,
+      "step": 2044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013828274557587598,
+      "loss": 0.7403,
+      "step": 2045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013822513063954935,
+      "loss": 0.7497,
+      "step": 2046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013816750083933612,
+      "loss": 0.7288,
+      "step": 2047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013810985619764572,
+      "loss": 0.7432,
+      "step": 2048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013805219673689332,
+      "loss": 0.711,
+      "step": 2049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013799452247949983,
+      "loss": 0.8411,
+      "step": 2050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013793683344789204,
+      "loss": 0.7417,
+      "step": 2051
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013787912966450235,
+      "loss": 0.7294,
+      "step": 2052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013782141115176893,
+      "loss": 0.758,
+      "step": 2053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001377636779321357,
+      "loss": 0.8395,
+      "step": 2054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013770593002805228,
+      "loss": 0.7872,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013764816746197404,
+      "loss": 0.8711,
+      "step": 2056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000137590390256362,
+      "loss": 0.6577,
+      "step": 2057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013753259843368287,
+      "loss": 0.7249,
+      "step": 2058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013747479201640914,
+      "loss": 0.7553,
+      "step": 2059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013741697102701882,
+      "loss": 0.8256,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013735913548799575,
+      "loss": 0.8278,
+      "step": 2061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013730128542182926,
+      "loss": 0.6765,
+      "step": 2062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001372434208510145,
+      "loss": 0.7418,
+      "step": 2063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001371855417980521,
+      "loss": 0.8493,
+      "step": 2064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013712764828544845,
+      "loss": 0.7839,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013706974033571547,
+      "loss": 0.9343,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013701181797137077,
+      "loss": 0.7745,
+      "step": 2067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013695388121493752,
+      "loss": 0.8125,
+      "step": 2068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013689593008894447,
+      "loss": 0.7834,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013683796461592604,
+      "loss": 0.9089,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013677998481842214,
+      "loss": 0.8449,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013672199071897823,
+      "loss": 0.7676,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013666398234014546,
+      "loss": 0.7749,
+      "step": 2073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013660595970448042,
+      "loss": 0.7747,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001365479228345453,
+      "loss": 0.6807,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013648987175290776,
+      "loss": 0.7121,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013643180648214103,
+      "loss": 0.7363,
+      "step": 2077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013637372704482397,
+      "loss": 0.8006,
+      "step": 2078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001363156334635407,
+      "loss": 0.8369,
+      "step": 2079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013625752576088105,
+      "loss": 0.7637,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013619940395944027,
+      "loss": 0.7896,
+      "step": 2081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013614126808181903,
+      "loss": 0.7915,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001360831181506236,
+      "loss": 0.8064,
+      "step": 2083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013602495418846562,
+      "loss": 0.7703,
+      "step": 2084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001359667762179622,
+      "loss": 0.809,
+      "step": 2085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013590858426173593,
+      "loss": 0.8466,
+      "step": 2086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013585037834241485,
+      "loss": 0.7589,
+      "step": 2087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013579215848263235,
+      "loss": 0.7524,
+      "step": 2088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013573392470502728,
+      "loss": 0.8053,
+      "step": 2089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013567567703224397,
+      "loss": 0.7954,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013561741548693206,
+      "loss": 0.6935,
+      "step": 2091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013555914009174663,
+      "loss": 0.6916,
+      "step": 2092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013550085086934814,
+      "loss": 0.6925,
+      "step": 2093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013544254784240243,
+      "loss": 0.7624,
+      "step": 2094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013538423103358064,
+      "loss": 0.789,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013532590046555937,
+      "loss": 0.8336,
+      "step": 2096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001352675561610206,
+      "loss": 0.8309,
+      "step": 2097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013520919814265144,
+      "loss": 0.8185,
+      "step": 2098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013515082643314455,
+      "loss": 0.772,
+      "step": 2099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013509244105519788,
+      "loss": 0.8404,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013503404203151458,
+      "loss": 0.7761,
+      "step": 2101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001349756293848032,
+      "loss": 0.7914,
+      "step": 2102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013491720313777756,
+      "loss": 0.8326,
+      "step": 2103
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001348587633131568,
+      "loss": 0.7888,
+      "step": 2104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001348003099336653,
+      "loss": 0.8079,
+      "step": 2105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013474184302203273,
+      "loss": 0.6855,
+      "step": 2106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013468336260099401,
+      "loss": 0.8309,
+      "step": 2107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001346248686932893,
+      "loss": 0.7857,
+      "step": 2108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013456636132166405,
+      "loss": 0.7107,
+      "step": 2109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001345078405088689,
+      "loss": 0.8287,
+      "step": 2110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013444930627765978,
+      "loss": 0.6642,
+      "step": 2111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013439075865079775,
+      "loss": 0.7067,
+      "step": 2112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013433219765104912,
+      "loss": 0.8536,
+      "step": 2113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013427362330118543,
+      "loss": 0.7674,
+      "step": 2114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001342150356239834,
+      "loss": 0.8544,
+      "step": 2115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013415643464222487,
+      "loss": 0.7888,
+      "step": 2116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001340978203786969,
+      "loss": 0.752,
+      "step": 2117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013403919285619179,
+      "loss": 0.7499,
+      "step": 2118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001339805520975068,
+      "loss": 0.7637,
+      "step": 2119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013392189812544455,
+      "loss": 0.7236,
+      "step": 2120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001338632309628127,
+      "loss": 0.8659,
+      "step": 2121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000133804550632424,
+      "loss": 0.6572,
+      "step": 2122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001337458571570964,
+      "loss": 0.832,
+      "step": 2123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001336871505596529,
+      "loss": 0.8393,
+      "step": 2124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001336284308629216,
+      "loss": 0.8237,
+      "step": 2125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013356969808973583,
+      "loss": 0.768,
+      "step": 2126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013351095226293383,
+      "loss": 0.7814,
+      "step": 2127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013345219340535897,
+      "loss": 0.7434,
+      "step": 2128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013339342153985973,
+      "loss": 0.7029,
+      "step": 2129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001333346366892896,
+      "loss": 0.6944,
+      "step": 2130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001332758388765071,
+      "loss": 0.6496,
+      "step": 2131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001332170281243759,
+      "loss": 0.7943,
+      "step": 2132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001331582044557646,
+      "loss": 0.8348,
+      "step": 2133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013309936789354687,
+      "loss": 0.7022,
+      "step": 2134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013304051846060134,
+      "loss": 0.7516,
+      "step": 2135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013298165617981172,
+      "loss": 0.7553,
+      "step": 2136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013292278107406662,
+      "loss": 0.7597,
+      "step": 2137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013286389316625972,
+      "loss": 0.7509,
+      "step": 2138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001328049924792897,
+      "loss": 0.7726,
+      "step": 2139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001327460790360601,
+      "loss": 0.6825,
+      "step": 2140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001326871528594795,
+      "loss": 0.7477,
+      "step": 2141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013262821397246142,
+      "loss": 0.7719,
+      "step": 2142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013256926239792432,
+      "loss": 0.7902,
+      "step": 2143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013251029815879156,
+      "loss": 0.8266,
+      "step": 2144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013245132127799147,
+      "loss": 0.8921,
+      "step": 2145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001323923317784573,
+      "loss": 0.7923,
+      "step": 2146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013233332968312715,
+      "loss": 0.7085,
+      "step": 2147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013227431501494405,
+      "loss": 0.9045,
+      "step": 2148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013221528779685596,
+      "loss": 0.7666,
+      "step": 2149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001321562480518157,
+      "loss": 0.6715,
+      "step": 2150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013209719580278083,
+      "loss": 0.7319,
+      "step": 2151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000132038131072714,
+      "loss": 0.7538,
+      "step": 2152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001319790538845826,
+      "loss": 0.7505,
+      "step": 2153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013191996426135876,
+      "loss": 0.7597,
+      "step": 2154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013186086222601964,
+      "loss": 0.7577,
+      "step": 2155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013180174780154713,
+      "loss": 0.6913,
+      "step": 2156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013174262101092792,
+      "loss": 0.6851,
+      "step": 2157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001316834818771535,
+      "loss": 0.8105,
+      "step": 2158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013162433042322028,
+      "loss": 0.7804,
+      "step": 2159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013156516667212927,
+      "loss": 0.6464,
+      "step": 2160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013150599064688643,
+      "loss": 0.6755,
+      "step": 2161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013144680237050238,
+      "loss": 0.7959,
+      "step": 2162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013138760186599263,
+      "loss": 0.6653,
+      "step": 2163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013132838915637726,
+      "loss": 0.7026,
+      "step": 2164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013126916426468128,
+      "loss": 0.7577,
+      "step": 2165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013120992721393432,
+      "loss": 0.7619,
+      "step": 2166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001311506780271708,
+      "loss": 0.6595,
+      "step": 2167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001310914167274298,
+      "loss": 0.7121,
+      "step": 2168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013103214333775521,
+      "loss": 0.7585,
+      "step": 2169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001309728578811955,
+      "loss": 0.7958,
+      "step": 2170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013091356038080388,
+      "loss": 0.6903,
+      "step": 2171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013085425085963828,
+      "loss": 0.8246,
+      "step": 2172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001307949293407613,
+      "loss": 0.6639,
+      "step": 2173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013073559584724013,
+      "loss": 0.8108,
+      "step": 2174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013067625040214672,
+      "loss": 0.778,
+      "step": 2175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013061689302855758,
+      "loss": 0.7571,
+      "step": 2176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013055752374955388,
+      "loss": 0.722,
+      "step": 2177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013049814258822147,
+      "loss": 0.8272,
+      "step": 2178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013043874956765076,
+      "loss": 0.6888,
+      "step": 2179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013037934471093682,
+      "loss": 0.7643,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001303199280411793,
+      "loss": 0.7277,
+      "step": 2181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013026049958148237,
+      "loss": 0.8046,
+      "step": 2182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013020105935495493,
+      "loss": 0.7721,
+      "step": 2183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013014160738471038,
+      "loss": 0.8077,
+      "step": 2184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013008214369386665,
+      "loss": 0.7466,
+      "step": 2185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001300226683055463,
+      "loss": 0.7518,
+      "step": 2186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001299631812428764,
+      "loss": 0.7007,
+      "step": 2187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001299036825289885,
+      "loss": 0.7859,
+      "step": 2188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012984417218701882,
+      "loss": 0.7952,
+      "step": 2189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012978465024010802,
+      "loss": 0.7349,
+      "step": 2190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012972511671140125,
+      "loss": 0.7656,
+      "step": 2191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012966557162404823,
+      "loss": 0.6597,
+      "step": 2192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012960601500120307,
+      "loss": 0.8059,
+      "step": 2193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012954644686602449,
+      "loss": 0.7727,
+      "step": 2194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012948686724167558,
+      "loss": 0.7661,
+      "step": 2195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000129427276151324,
+      "loss": 0.7899,
+      "step": 2196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012936767361814179,
+      "loss": 0.7811,
+      "step": 2197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012930805966530548,
+      "loss": 0.714,
+      "step": 2198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000129248434315996,
+      "loss": 0.6827,
+      "step": 2199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001291887975933987,
+      "loss": 0.8389,
+      "step": 2200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001291291495207035,
+      "loss": 0.7046,
+      "step": 2201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012906949012110456,
+      "loss": 0.7957,
+      "step": 2202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012900981941780048,
+      "loss": 0.7879,
+      "step": 2203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012895013743399432,
+      "loss": 0.7788,
+      "step": 2204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012889044419289348,
+      "loss": 0.685,
+      "step": 2205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012883073971770972,
+      "loss": 0.8318,
+      "step": 2206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012877102403165922,
+      "loss": 0.7315,
+      "step": 2207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012871129715796254,
+      "loss": 0.7381,
+      "step": 2208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012865155911984444,
+      "loss": 0.8765,
+      "step": 2209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012859180994053424,
+      "loss": 0.7885,
+      "step": 2210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012853204964326535,
+      "loss": 0.7555,
+      "step": 2211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012847227825127578,
+      "loss": 0.817,
+      "step": 2212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012841249578780757,
+      "loss": 0.7362,
+      "step": 2213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012835270227610728,
+      "loss": 0.7156,
+      "step": 2214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012829289773942564,
+      "loss": 0.7265,
+      "step": 2215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012823308220101778,
+      "loss": 0.7851,
+      "step": 2216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012817325568414297,
+      "loss": 0.7819,
+      "step": 2217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012811341821206487,
+      "loss": 0.7705,
+      "step": 2218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012805356980805132,
+      "loss": 0.7937,
+      "step": 2219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012799371049537447,
+      "loss": 0.7179,
+      "step": 2220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012793384029731065,
+      "loss": 0.7544,
+      "step": 2221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012787395923714048,
+      "loss": 0.8841,
+      "step": 2222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012781406733814878,
+      "loss": 0.7742,
+      "step": 2223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012775416462362457,
+      "loss": 0.914,
+      "step": 2224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001276942511168611,
+      "loss": 0.6712,
+      "step": 2225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012763432684115582,
+      "loss": 0.6533,
+      "step": 2226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012757439181981033,
+      "loss": 0.769,
+      "step": 2227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012751444607613045,
+      "loss": 0.7838,
+      "step": 2228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012745448963342613,
+      "loss": 0.7641,
+      "step": 2229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001273945225150115,
+      "loss": 0.8266,
+      "step": 2230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012733454474420493,
+      "loss": 0.7181,
+      "step": 2231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012727455634432873,
+      "loss": 0.6717,
+      "step": 2232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001272145573387095,
+      "loss": 0.846,
+      "step": 2233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012715454775067797,
+      "loss": 0.7924,
+      "step": 2234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012709452760356884,
+      "loss": 0.7467,
+      "step": 2235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012703449692072111,
+      "loss": 0.8324,
+      "step": 2236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012697445572547772,
+      "loss": 0.7598,
+      "step": 2237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012691440404118581,
+      "loss": 0.8315,
+      "step": 2238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001268543418911965,
+      "loss": 0.7186,
+      "step": 2239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012679426929886506,
+      "loss": 0.7892,
+      "step": 2240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012673418628755077,
+      "loss": 0.8212,
+      "step": 2241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000126674092880617,
+      "loss": 0.7814,
+      "step": 2242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001266139891014311,
+      "loss": 0.7006,
+      "step": 2243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012655387497336455,
+      "loss": 0.8434,
+      "step": 2244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012649375051979275,
+      "loss": 0.7501,
+      "step": 2245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012643361576409516,
+      "loss": 0.7842,
+      "step": 2246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012637347072965528,
+      "loss": 0.8896,
+      "step": 2247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012631331543986061,
+      "loss": 0.6256,
+      "step": 2248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012625314991810254,
+      "loss": 0.8742,
+      "step": 2249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012619297418777652,
+      "loss": 0.7076,
+      "step": 2250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012613278827228197,
+      "loss": 0.7436,
+      "step": 2251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001260725921950222,
+      "loss": 0.7831,
+      "step": 2252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001260123859794046,
+      "loss": 0.7242,
+      "step": 2253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012595216964884042,
+      "loss": 0.8717,
+      "step": 2254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001258919432267448,
+      "loss": 0.7074,
+      "step": 2255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012583170673653688,
+      "loss": 0.7462,
+      "step": 2256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012577146020163968,
+      "loss": 0.7901,
+      "step": 2257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012571120364548015,
+      "loss": 0.7574,
+      "step": 2258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012565093709148911,
+      "loss": 0.7553,
+      "step": 2259
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001255906605631013,
+      "loss": 0.6609,
+      "step": 2260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012553037408375532,
+      "loss": 0.8533,
+      "step": 2261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001254700776768936,
+      "loss": 0.817,
+      "step": 2262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012540977136596247,
+      "loss": 0.738,
+      "step": 2263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012534945517441215,
+      "loss": 0.7948,
+      "step": 2264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012528912912569665,
+      "loss": 0.7824,
+      "step": 2265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012522879324327382,
+      "loss": 0.7168,
+      "step": 2266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012516844755060532,
+      "loss": 0.8451,
+      "step": 2267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012510809207115666,
+      "loss": 0.7444,
+      "step": 2268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012504772682839712,
+      "loss": 0.8802,
+      "step": 2269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012498735184579982,
+      "loss": 0.7473,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012492696714684165,
+      "loss": 0.6505,
+      "step": 2271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001248665727550032,
+      "loss": 0.771,
+      "step": 2272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012480616869376903,
+      "loss": 0.8078,
+      "step": 2273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012474575498662722,
+      "loss": 0.7371,
+      "step": 2274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001246853316570697,
+      "loss": 0.7423,
+      "step": 2275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012462489872859224,
+      "loss": 0.776,
+      "step": 2276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012456445622469422,
+      "loss": 0.8441,
+      "step": 2277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001245040041688787,
+      "loss": 0.7379,
+      "step": 2278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012444354258465268,
+      "loss": 0.8247,
+      "step": 2279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012438307149552659,
+      "loss": 0.8359,
+      "step": 2280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012432259092501474,
+      "loss": 0.8098,
+      "step": 2281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012426210089663505,
+      "loss": 0.6843,
+      "step": 2282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012420160143390918,
+      "loss": 0.7478,
+      "step": 2283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012414109256036237,
+      "loss": 0.7645,
+      "step": 2284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001240805742995236,
+      "loss": 0.8145,
+      "step": 2285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001240200466749254,
+      "loss": 0.6706,
+      "step": 2286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012395950971010408,
+      "loss": 0.7518,
+      "step": 2287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001238989634285995,
+      "loss": 0.7895,
+      "step": 2288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012383840785395512,
+      "loss": 0.8826,
+      "step": 2289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012377784300971807,
+      "loss": 0.826,
+      "step": 2290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012371726891943904,
+      "loss": 0.8465,
+      "step": 2291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012365668560667236,
+      "loss": 0.72,
+      "step": 2292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012359609309497592,
+      "loss": 0.6482,
+      "step": 2293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012353549140791116,
+      "loss": 0.8928,
+      "step": 2294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012347488056904316,
+      "loss": 0.7407,
+      "step": 2295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012341426060194047,
+      "loss": 0.7955,
+      "step": 2296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012335363153017528,
+      "loss": 0.6996,
+      "step": 2297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012329299337732321,
+      "loss": 0.7337,
+      "step": 2298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012323234616696353,
+      "loss": 0.7618,
+      "step": 2299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012317168992267895,
+      "loss": 0.9158,
+      "step": 2300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001231110246680558,
+      "loss": 0.7772,
+      "step": 2301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012305035042668372,
+      "loss": 0.7695,
+      "step": 2302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012298966722215602,
+      "loss": 0.8214,
+      "step": 2303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001229289750780694,
+      "loss": 0.7234,
+      "step": 2304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012286827401802416,
+      "loss": 0.7796,
+      "step": 2305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012280756406562389,
+      "loss": 0.7987,
+      "step": 2306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012274684524447575,
+      "loss": 0.7773,
+      "step": 2307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012268611757819032,
+      "loss": 0.7618,
+      "step": 2308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012262538109038162,
+      "loss": 0.783,
+      "step": 2309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012256463580466714,
+      "loss": 0.5895,
+      "step": 2310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001225038817446677,
+      "loss": 0.8624,
+      "step": 2311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012244311893400763,
+      "loss": 0.646,
+      "step": 2312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001223823473963146,
+      "loss": 0.7571,
+      "step": 2313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001223215671552197,
+      "loss": 0.8809,
+      "step": 2314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012226077823435737,
+      "loss": 0.7727,
+      "step": 2315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012219998065736553,
+      "loss": 0.7141,
+      "step": 2316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001221391744478853,
+      "loss": 0.8435,
+      "step": 2317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001220783596295613,
+      "loss": 0.6227,
+      "step": 2318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012201753622604144,
+      "loss": 0.7381,
+      "step": 2319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012195670426097691,
+      "loss": 0.79,
+      "step": 2320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001218958637580224,
+      "loss": 0.8276,
+      "step": 2321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001218350147408357,
+      "loss": 0.7016,
+      "step": 2322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012177415723307808,
+      "loss": 0.7558,
+      "step": 2323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012171329125841404,
+      "loss": 0.799,
+      "step": 2324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001216524168405114,
+      "loss": 0.7556,
+      "step": 2325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012159153400304124,
+      "loss": 0.8097,
+      "step": 2326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001215306427696779,
+      "loss": 0.8487,
+      "step": 2327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012146974316409902,
+      "loss": 0.6005,
+      "step": 2328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012140883520998551,
+      "loss": 0.7865,
+      "step": 2329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012134791893102149,
+      "loss": 0.7856,
+      "step": 2330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012128699435089428,
+      "loss": 0.664,
+      "step": 2331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012122606149329451,
+      "loss": 0.8178,
+      "step": 2332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012116512038191598,
+      "loss": 0.8464,
+      "step": 2333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012110417104045575,
+      "loss": 0.7673,
+      "step": 2334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012104321349261401,
+      "loss": 0.7867,
+      "step": 2335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012098224776209417,
+      "loss": 0.7735,
+      "step": 2336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012092127387260284,
+      "loss": 0.8108,
+      "step": 2337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012086029184784977,
+      "loss": 0.7692,
+      "step": 2338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012079930171154792,
+      "loss": 0.8162,
+      "step": 2339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012073830348741335,
+      "loss": 0.7985,
+      "step": 2340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001206772971991653,
+      "loss": 0.849,
+      "step": 2341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012061628287052611,
+      "loss": 0.8906,
+      "step": 2342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012055526052522132,
+      "loss": 0.6732,
+      "step": 2343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012049423018697949,
+      "loss": 0.704,
+      "step": 2344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012043319187953241,
+      "loss": 0.7146,
+      "step": 2345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012037214562661483,
+      "loss": 0.7253,
+      "step": 2346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012031109145196469,
+      "loss": 0.6857,
+      "step": 2347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012025002937932297,
+      "loss": 0.6851,
+      "step": 2348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012018895943243371,
+      "loss": 0.7283,
+      "step": 2349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012012788163504405,
+      "loss": 0.6784,
+      "step": 2350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012006679601090414,
+      "loss": 0.722,
+      "step": 2351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012000570258376725,
+      "loss": 0.6482,
+      "step": 2352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011994460137738954,
+      "loss": 0.8312,
+      "step": 2353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011988349241553038,
+      "loss": 0.79,
+      "step": 2354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011982237572195202,
+      "loss": 0.7729,
+      "step": 2355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011976125132041974,
+      "loss": 0.7,
+      "step": 2356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011970011923470187,
+      "loss": 0.8522,
+      "step": 2357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011963897948856967,
+      "loss": 0.7527,
+      "step": 2358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001195778321057974,
+      "loss": 0.8099,
+      "step": 2359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011951667711016232,
+      "loss": 0.6717,
+      "step": 2360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011945551452544456,
+      "loss": 0.7504,
+      "step": 2361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011939434437542733,
+      "loss": 0.7521,
+      "step": 2362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001193331666838967,
+      "loss": 0.8936,
+      "step": 2363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011927198147464162,
+      "loss": 0.7858,
+      "step": 2364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011921078877145412,
+      "loss": 0.8593,
+      "step": 2365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011914958859812901,
+      "loss": 0.7666,
+      "step": 2366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011908838097846404,
+      "loss": 0.8257,
+      "step": 2367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011902716593625991,
+      "loss": 0.8508,
+      "step": 2368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011896594349532013,
+      "loss": 0.6803,
+      "step": 2369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011890471367945108,
+      "loss": 0.7585,
+      "step": 2370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011884347651246213,
+      "loss": 0.7779,
+      "step": 2371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011878223201816534,
+      "loss": 0.8022,
+      "step": 2372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011872098022037575,
+      "loss": 0.7244,
+      "step": 2373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011865972114291115,
+      "loss": 0.7895,
+      "step": 2374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011859845480959228,
+      "loss": 0.8556,
+      "step": 2375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011853718124424258,
+      "loss": 0.7326,
+      "step": 2376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011847590047068832,
+      "loss": 0.7197,
+      "step": 2377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011841461251275867,
+      "loss": 0.8502,
+      "step": 2378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011835331739428544,
+      "loss": 0.7995,
+      "step": 2379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011829201513910338,
+      "loss": 0.8234,
+      "step": 2380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011823070577104989,
+      "loss": 0.7905,
+      "step": 2381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011816938931396523,
+      "loss": 0.736,
+      "step": 2382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011810806579169235,
+      "loss": 0.7466,
+      "step": 2383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011804673522807699,
+      "loss": 0.6865,
+      "step": 2384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011798539764696756,
+      "loss": 0.8089,
+      "step": 2385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011792405307221534,
+      "loss": 0.839,
+      "step": 2386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001178627015276742,
+      "loss": 0.6471,
+      "step": 2387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011780134303720074,
+      "loss": 0.7908,
+      "step": 2388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011773997762465429,
+      "loss": 0.7061,
+      "step": 2389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011767860531389685,
+      "loss": 0.852,
+      "step": 2390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011761722612879316,
+      "loss": 0.8515,
+      "step": 2391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011755584009321054,
+      "loss": 0.7815,
+      "step": 2392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011749444723101906,
+      "loss": 0.852,
+      "step": 2393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011743304756609132,
+      "loss": 0.7237,
+      "step": 2394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011737164112230279,
+      "loss": 0.8576,
+      "step": 2395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001173102279235313,
+      "loss": 0.7798,
+      "step": 2396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011724880799365756,
+      "loss": 0.6844,
+      "step": 2397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011718738135656471,
+      "loss": 0.7419,
+      "step": 2398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001171259480361386,
+      "loss": 0.8444,
+      "step": 2399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001170645080562676,
+      "loss": 0.7453,
+      "step": 2400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011700306144084279,
+      "loss": 0.8103,
+      "step": 2401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011694160821375768,
+      "loss": 0.7019,
+      "step": 2402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011688014839890851,
+      "loss": 0.6949,
+      "step": 2403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011681868202019397,
+      "loss": 0.6761,
+      "step": 2404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011675720910151528,
+      "loss": 0.6631,
+      "step": 2405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011669572966677633,
+      "loss": 0.7928,
+      "step": 2406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011663424373988341,
+      "loss": 0.7987,
+      "step": 2407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011657275134474546,
+      "loss": 0.7893,
+      "step": 2408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011651125250527383,
+      "loss": 0.7254,
+      "step": 2409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011644974724538243,
+      "loss": 0.8152,
+      "step": 2410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011638823558898762,
+      "loss": 0.7042,
+      "step": 2411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001163267175600083,
+      "loss": 0.847,
+      "step": 2412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011626519318236581,
+      "loss": 0.8691,
+      "step": 2413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011620366247998405,
+      "loss": 0.7597,
+      "step": 2414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011614212547678922,
+      "loss": 0.7352,
+      "step": 2415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011608058219671006,
+      "loss": 0.7333,
+      "step": 2416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001160190326636778,
+      "loss": 0.7834,
+      "step": 2417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011595747690162602,
+      "loss": 0.7079,
+      "step": 2418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011589591493449072,
+      "loss": 0.8449,
+      "step": 2419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011583434678621039,
+      "loss": 0.647,
+      "step": 2420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011577277248072587,
+      "loss": 0.8157,
+      "step": 2421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011571119204198037,
+      "loss": 0.7838,
+      "step": 2422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011564960549391956,
+      "loss": 0.8028,
+      "step": 2423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011558801286049139,
+      "loss": 0.7751,
+      "step": 2424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011552641416564624,
+      "loss": 0.7029,
+      "step": 2425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011546480943333686,
+      "loss": 0.7287,
+      "step": 2426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001154031986875183,
+      "loss": 0.7908,
+      "step": 2427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011534158195214799,
+      "loss": 0.7387,
+      "step": 2428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011527995925118566,
+      "loss": 0.7629,
+      "step": 2429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011521833060859338,
+      "loss": 0.7543,
+      "step": 2430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011515669604833546,
+      "loss": 0.7449,
+      "step": 2431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011509505559437865,
+      "loss": 0.7812,
+      "step": 2432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011503340927069189,
+      "loss": 0.7132,
+      "step": 2433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001149717571012464,
+      "loss": 0.7218,
+      "step": 2434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011491009911001571,
+      "loss": 0.6824,
+      "step": 2435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011484843532097558,
+      "loss": 0.7437,
+      "step": 2436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001147867657581041,
+      "loss": 0.834,
+      "step": 2437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011472509044538149,
+      "loss": 0.8021,
+      "step": 2438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011466340940679033,
+      "loss": 0.8063,
+      "step": 2439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011460172266631532,
+      "loss": 0.8433,
+      "step": 2440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011454003024794344,
+      "loss": 0.7809,
+      "step": 2441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011447833217566384,
+      "loss": 0.8273,
+      "step": 2442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011441662847346793,
+      "loss": 0.8903,
+      "step": 2443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011435491916534919,
+      "loss": 0.8073,
+      "step": 2444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011429320427530343,
+      "loss": 0.685,
+      "step": 2445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011423148382732853,
+      "loss": 0.769,
+      "step": 2446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011416975784542454,
+      "loss": 0.8548,
+      "step": 2447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011410802635359372,
+      "loss": 0.7562,
+      "step": 2448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011404628937584036,
+      "loss": 0.7512,
+      "step": 2449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011398454693617106,
+      "loss": 0.8532,
+      "step": 2450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011392279905859438,
+      "loss": 0.7571,
+      "step": 2451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011386104576712105,
+      "loss": 0.6876,
+      "step": 2452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011379928708576393,
+      "loss": 0.7552,
+      "step": 2453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011373752303853791,
+      "loss": 0.7054,
+      "step": 2454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011367575364946006,
+      "loss": 0.771,
+      "step": 2455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011361397894254948,
+      "loss": 0.7315,
+      "step": 2456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011355219894182731,
+      "loss": 0.8226,
+      "step": 2457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011349041367131676,
+      "loss": 0.7637,
+      "step": 2458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011342862315504316,
+      "loss": 0.737,
+      "step": 2459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001133668274170338,
+      "loss": 0.8074,
+      "step": 2460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011330502648131799,
+      "loss": 0.777,
+      "step": 2461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011324322037192717,
+      "loss": 0.7097,
+      "step": 2462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011318140911289466,
+      "loss": 0.7682,
+      "step": 2463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011311959272825585,
+      "loss": 0.789,
+      "step": 2464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001130577712420481,
+      "loss": 0.8435,
+      "step": 2465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011299594467831078,
+      "loss": 0.7771,
+      "step": 2466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011293411306108526,
+      "loss": 0.7878,
+      "step": 2467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011287227641441478,
+      "loss": 0.8425,
+      "step": 2468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011281043476234461,
+      "loss": 0.7042,
+      "step": 2469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011274858812892197,
+      "loss": 0.7277,
+      "step": 2470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011268673653819595,
+      "loss": 0.7477,
+      "step": 2471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011262488001421765,
+      "loss": 0.7136,
+      "step": 2472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011256301858104005,
+      "loss": 0.6751,
+      "step": 2473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011250115226271804,
+      "loss": 0.7486,
+      "step": 2474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011243928108330842,
+      "loss": 0.7508,
+      "step": 2475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011237740506686983,
+      "loss": 0.7757,
+      "step": 2476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011231552423746283,
+      "loss": 0.8345,
+      "step": 2477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011225363861914991,
+      "loss": 0.8137,
+      "step": 2478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011219174823599532,
+      "loss": 0.7375,
+      "step": 2479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011212985311206521,
+      "loss": 0.8509,
+      "step": 2480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011206795327142758,
+      "loss": 0.7752,
+      "step": 2481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011200604873815225,
+      "loss": 0.791,
+      "step": 2482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011194413953631089,
+      "loss": 0.7013,
+      "step": 2483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011188222568997697,
+      "loss": 0.7918,
+      "step": 2484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011182030722322571,
+      "loss": 0.8117,
+      "step": 2485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011175838416013424,
+      "loss": 0.8488,
+      "step": 2486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011169645652478138,
+      "loss": 0.695,
+      "step": 2487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011163452434124773,
+      "loss": 0.6863,
+      "step": 2488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011157258763361575,
+      "loss": 0.7885,
+      "step": 2489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011151064642596958,
+      "loss": 0.7411,
+      "step": 2490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011144870074239513,
+      "loss": 0.7694,
+      "step": 2491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011138675060698008,
+      "loss": 0.748,
+      "step": 2492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011132479604381376,
+      "loss": 0.7213,
+      "step": 2493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011126283707698731,
+      "loss": 0.7527,
+      "step": 2494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011120087373059351,
+      "loss": 0.7856,
+      "step": 2495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011113890602872694,
+      "loss": 0.8501,
+      "step": 2496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011107693399548378,
+      "loss": 0.6705,
+      "step": 2497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001110149576549619,
+      "loss": 0.7306,
+      "step": 2498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011095297703126093,
+      "loss": 0.7557,
+      "step": 2499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011089099214848203,
+      "loss": 0.7059,
+      "step": 2500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011082900303072814,
+      "loss": 0.8175,
+      "step": 2501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011076700970210383,
+      "loss": 0.7703,
+      "step": 2502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011070501218671525,
+      "loss": 0.7348,
+      "step": 2503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001106430105086702,
+      "loss": 0.7112,
+      "step": 2504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011058100469207811,
+      "loss": 0.7898,
+      "step": 2505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011051899476105001,
+      "loss": 0.8188,
+      "step": 2506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011045698073969855,
+      "loss": 0.6328,
+      "step": 2507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011039496265213791,
+      "loss": 0.5856,
+      "step": 2508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011033294052248395,
+      "loss": 0.7781,
+      "step": 2509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011027091437485404,
+      "loss": 0.6637,
+      "step": 2510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011020888423336705,
+      "loss": 0.7948,
+      "step": 2511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011014685012214356,
+      "loss": 0.7065,
+      "step": 2512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011008481206530558,
+      "loss": 0.7938,
+      "step": 2513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011002277008697667,
+      "loss": 0.7604,
+      "step": 2514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010996072421128197,
+      "loss": 0.728,
+      "step": 2515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000109898674462348,
+      "loss": 0.8167,
+      "step": 2516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010983662086430293,
+      "loss": 0.7529,
+      "step": 2517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010977456344127639,
+      "loss": 0.7116,
+      "step": 2518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010971250221739946,
+      "loss": 0.6478,
+      "step": 2519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010965043721680472,
+      "loss": 0.6342,
+      "step": 2520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010958836846362621,
+      "loss": 0.6789,
+      "step": 2521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010952629598199946,
+      "loss": 0.8155,
+      "step": 2522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001094642197960614,
+      "loss": 0.8179,
+      "step": 2523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010940213992995047,
+      "loss": 0.7099,
+      "step": 2524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010934005640780647,
+      "loss": 0.7146,
+      "step": 2525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010927796925377063,
+      "loss": 0.7422,
+      "step": 2526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010921587849198565,
+      "loss": 0.7291,
+      "step": 2527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010915378414659556,
+      "loss": 0.7243,
+      "step": 2528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010909168624174583,
+      "loss": 0.7477,
+      "step": 2529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010902958480158334,
+      "loss": 0.7801,
+      "step": 2530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010896747985025627,
+      "loss": 0.7014,
+      "step": 2531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010890537141191417,
+      "loss": 0.7806,
+      "step": 2532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010884325951070804,
+      "loss": 0.8185,
+      "step": 2533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010878114417079013,
+      "loss": 0.7082,
+      "step": 2534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010871902541631409,
+      "loss": 0.704,
+      "step": 2535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010865690327143487,
+      "loss": 0.8,
+      "step": 2536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010859477776030871,
+      "loss": 0.7794,
+      "step": 2537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001085326489070932,
+      "loss": 0.7551,
+      "step": 2538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010847051673594724,
+      "loss": 0.7619,
+      "step": 2539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010840838127103092,
+      "loss": 0.7698,
+      "step": 2540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010834624253650578,
+      "loss": 0.7194,
+      "step": 2541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001082841005565345,
+      "loss": 0.8206,
+      "step": 2542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010822195535528106,
+      "loss": 0.7356,
+      "step": 2543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010815980695691068,
+      "loss": 0.69,
+      "step": 2544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010809765538558989,
+      "loss": 0.82,
+      "step": 2545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010803550066548633,
+      "loss": 0.7748,
+      "step": 2546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010797334282076898,
+      "loss": 0.7815,
+      "step": 2547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000107911181875608,
+      "loss": 0.8178,
+      "step": 2548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010784901785417474,
+      "loss": 0.766,
+      "step": 2549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010778685078064173,
+      "loss": 0.7771,
+      "step": 2550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010772468067918268,
+      "loss": 0.8452,
+      "step": 2551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010766250757397256,
+      "loss": 0.7919,
+      "step": 2552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010760033148918747,
+      "loss": 0.8062,
+      "step": 2553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010753815244900458,
+      "loss": 0.8265,
+      "step": 2554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010747597047760236,
+      "loss": 0.7846,
+      "step": 2555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010741378559916029,
+      "loss": 0.8579,
+      "step": 2556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010735159783785905,
+      "loss": 0.7803,
+      "step": 2557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010728940721788041,
+      "loss": 0.7242,
+      "step": 2558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010722721376340729,
+      "loss": 0.7989,
+      "step": 2559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010716501749862369,
+      "loss": 0.8053,
+      "step": 2560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010710281844771468,
+      "loss": 0.7568,
+      "step": 2561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010704061663486644,
+      "loss": 0.7819,
+      "step": 2562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010697841208426618,
+      "loss": 0.7646,
+      "step": 2563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010691620482010228,
+      "loss": 0.7734,
+      "step": 2564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010685399486656406,
+      "loss": 0.7528,
+      "step": 2565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010679178224784197,
+      "loss": 0.8225,
+      "step": 2566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010672956698812742,
+      "loss": 0.7601,
+      "step": 2567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010666734911161291,
+      "loss": 0.7555,
+      "step": 2568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010660512864249191,
+      "loss": 0.7535,
+      "step": 2569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010654290560495897,
+      "loss": 0.7533,
+      "step": 2570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010648068002320952,
+      "loss": 0.6786,
+      "step": 2571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010641845192144013,
+      "loss": 0.7447,
+      "step": 2572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010635622132384821,
+      "loss": 0.8103,
+      "step": 2573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010629398825463222,
+      "loss": 0.7937,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010623175273799157,
+      "loss": 0.7065,
+      "step": 2575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010616951479812658,
+      "loss": 0.7899,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010610727445923859,
+      "loss": 0.7038,
+      "step": 2577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010604503174552982,
+      "loss": 0.7109,
+      "step": 2578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001059827866812034,
+      "loss": 0.8509,
+      "step": 2579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010592053929046339,
+      "loss": 0.8086,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010585828959751478,
+      "loss": 0.7698,
+      "step": 2581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010579603762656347,
+      "loss": 0.7621,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010573378340181614,
+      "loss": 0.7032,
+      "step": 2583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010567152694748045,
+      "loss": 0.7176,
+      "step": 2584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001056092682877649,
+      "loss": 0.8246,
+      "step": 2585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010554700744687885,
+      "loss": 0.8488,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010548474444903247,
+      "loss": 0.7797,
+      "step": 2587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010542247931843682,
+      "loss": 0.6893,
+      "step": 2588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001053602120793038,
+      "loss": 0.7453,
+      "step": 2589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010529794275584605,
+      "loss": 0.7593,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010523567137227707,
+      "loss": 0.7316,
+      "step": 2591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010517339795281118,
+      "loss": 0.8323,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010511112252166349,
+      "loss": 0.7162,
+      "step": 2593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010504884510304982,
+      "loss": 0.7043,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010498656572118686,
+      "loss": 0.9726,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010492428440029199,
+      "loss": 0.7423,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010486200116458342,
+      "loss": 0.7186,
+      "step": 2597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010479971603828,
+      "loss": 0.8036,
+      "step": 2598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010473742904560143,
+      "loss": 0.7353,
+      "step": 2599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010467514021076805,
+      "loss": 0.7464,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010461284955800094,
+      "loss": 0.829,
+      "step": 2601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010455055711152193,
+      "loss": 0.809,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010448826289555345,
+      "loss": 0.8222,
+      "step": 2603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010442596693431872,
+      "loss": 0.832,
+      "step": 2604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010436366925204159,
+      "loss": 0.6619,
+      "step": 2605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001043013698729466,
+      "loss": 0.7312,
+      "step": 2606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010423906882125889,
+      "loss": 0.8084,
+      "step": 2607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010417676612120437,
+      "loss": 0.7188,
+      "step": 2608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010411446179700943,
+      "loss": 0.745,
+      "step": 2609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010405215587290124,
+      "loss": 0.6513,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010398984837310755,
+      "loss": 0.7699,
+      "step": 2611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010392753932185666,
+      "loss": 0.7242,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010386522874337752,
+      "loss": 0.7843,
+      "step": 2613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010380291666189967,
+      "loss": 0.8438,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010374060310165323,
+      "loss": 0.7244,
+      "step": 2615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010367828808686893,
+      "loss": 0.6421,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010361597164177801,
+      "loss": 0.7635,
+      "step": 2617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010355365379061231,
+      "loss": 0.8681,
+      "step": 2618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001034913345576042,
+      "loss": 0.7478,
+      "step": 2619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010342901396698659,
+      "loss": 0.8397,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010336669204299286,
+      "loss": 0.772,
+      "step": 2621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010330436880985703,
+      "loss": 0.7737,
+      "step": 2622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010324204429181354,
+      "loss": 0.8541,
+      "step": 2623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010317971851309735,
+      "loss": 0.9417,
+      "step": 2624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010311739149794392,
+      "loss": 0.748,
+      "step": 2625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010305506327058916,
+      "loss": 0.8443,
+      "step": 2626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010299273385526949,
+      "loss": 0.7638,
+      "step": 2627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010293040327622175,
+      "loss": 0.795,
+      "step": 2628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001028680715576833,
+      "loss": 0.8194,
+      "step": 2629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010280573872389187,
+      "loss": 0.7621,
+      "step": 2630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010274340479908568,
+      "loss": 0.728,
+      "step": 2631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010268106980750334,
+      "loss": 0.6429,
+      "step": 2632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010261873377338386,
+      "loss": 0.7573,
+      "step": 2633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010255639672096673,
+      "loss": 0.7107,
+      "step": 2634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010249405867449175,
+      "loss": 0.7856,
+      "step": 2635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010243171965819913,
+      "loss": 0.711,
+      "step": 2636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010236937969632952,
+      "loss": 0.8354,
+      "step": 2637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001023070388131238,
+      "loss": 0.7934,
+      "step": 2638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010224469703282335,
+      "loss": 0.8949,
+      "step": 2639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001021823543796699,
+      "loss": 0.6739,
+      "step": 2640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010212001087790537,
+      "loss": 0.6848,
+      "step": 2641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010205766655177215,
+      "loss": 0.8314,
+      "step": 2642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010199532142551289,
+      "loss": 0.7485,
+      "step": 2643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010193297552337055,
+      "loss": 0.7939,
+      "step": 2644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010187062886958844,
+      "loss": 0.7971,
+      "step": 2645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010180828148841013,
+      "loss": 0.6955,
+      "step": 2646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010174593340407945,
+      "loss": 0.7347,
+      "step": 2647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010168358464084051,
+      "loss": 0.7276,
+      "step": 2648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010162123522293773,
+      "loss": 0.7171,
+      "step": 2649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010155888517461574,
+      "loss": 0.717,
+      "step": 2650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010149653452011943,
+      "loss": 0.7366,
+      "step": 2651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010143418328369395,
+      "loss": 0.6799,
+      "step": 2652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010137183148958463,
+      "loss": 0.7696,
+      "step": 2653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010130947916203701,
+      "loss": 0.6438,
+      "step": 2654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010124712632529689,
+      "loss": 0.781,
+      "step": 2655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010118477300361027,
+      "loss": 0.7172,
+      "step": 2656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010112241922122327,
+      "loss": 0.8043,
+      "step": 2657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010106006500238226,
+      "loss": 0.6894,
+      "step": 2658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010099771037133372,
+      "loss": 0.6182,
+      "step": 2659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010093535535232434,
+      "loss": 0.8151,
+      "step": 2660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010087299996960096,
+      "loss": 0.7652,
+      "step": 2661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010081064424741051,
+      "loss": 0.6584,
+      "step": 2662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010074828821000011,
+      "loss": 0.7898,
+      "step": 2663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010068593188161697,
+      "loss": 0.8094,
+      "step": 2664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010062357528650841,
+      "loss": 0.7606,
+      "step": 2665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010056121844892188,
+      "loss": 0.7683,
+      "step": 2666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010049886139310489,
+      "loss": 0.724,
+      "step": 2667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010043650414330508,
+      "loss": 0.9111,
+      "step": 2668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010037414672377012,
+      "loss": 0.7701,
+      "step": 2669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010031178915874774,
+      "loss": 0.7626,
+      "step": 2670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010024943147248583,
+      "loss": 0.9791,
+      "step": 2671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010018707368923217,
+      "loss": 0.7453,
+      "step": 2672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010012471583323468,
+      "loss": 0.8493,
+      "step": 2673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010006235792874132,
+      "loss": 0.8081,
+      "step": 2674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001,
+      "loss": 0.7902,
+      "step": 2675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993764207125867e-05,
+      "loss": 0.8393,
+      "step": 2676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987528416676534e-05,
+      "loss": 0.8289,
+      "step": 2677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.981292631076784e-05,
+      "loss": 0.7036,
+      "step": 2678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97505685275142e-05,
+      "loss": 0.7389,
+      "step": 2679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.968821084125227e-05,
+      "loss": 0.7773,
+      "step": 2680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96258532762299e-05,
+      "loss": 0.7342,
+      "step": 2681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956349585669496e-05,
+      "loss": 0.6872,
+      "step": 2682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.950113860689512e-05,
+      "loss": 0.7897,
+      "step": 2683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943878155107816e-05,
+      "loss": 0.6599,
+      "step": 2684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937642471349161e-05,
+      "loss": 0.7498,
+      "step": 2685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931406811838308e-05,
+      "loss": 0.811,
+      "step": 2686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925171178999991e-05,
+      "loss": 0.7155,
+      "step": 2687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.918935575258948e-05,
+      "loss": 0.6763,
+      "step": 2688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912700003039908e-05,
+      "loss": 0.7626,
+      "step": 2689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.906464464767567e-05,
+      "loss": 0.7557,
+      "step": 2690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90022896286663e-05,
+      "loss": 0.7526,
+      "step": 2691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.893993499761775e-05,
+      "loss": 0.6743,
+      "step": 2692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887758077877672e-05,
+      "loss": 0.8069,
+      "step": 2693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.881522699638975e-05,
+      "loss": 0.7646,
+      "step": 2694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87528736747031e-05,
+      "loss": 0.7199,
+      "step": 2695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869052083796303e-05,
+      "loss": 0.7702,
+      "step": 2696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862816851041541e-05,
+      "loss": 0.7088,
+      "step": 2697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856581671630609e-05,
+      "loss": 0.7239,
+      "step": 2698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850346547988058e-05,
+      "loss": 0.791,
+      "step": 2699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844111482538427e-05,
+      "loss": 0.7517,
+      "step": 2700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837876477706229e-05,
+      "loss": 0.7171,
+      "step": 2701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.831641535915951e-05,
+      "loss": 0.7318,
+      "step": 2702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.825406659592059e-05,
+      "loss": 0.6213,
+      "step": 2703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.819171851158989e-05,
+      "loss": 0.8155,
+      "step": 2704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.812937113041155e-05,
+      "loss": 0.8442,
+      "step": 2705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806702447662947e-05,
+      "loss": 0.6328,
+      "step": 2706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800467857448713e-05,
+      "loss": 0.8173,
+      "step": 2707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79423334482279e-05,
+      "loss": 0.7873,
+      "step": 2708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.787998912209466e-05,
+      "loss": 0.7106,
+      "step": 2709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.781764562033011e-05,
+      "loss": 0.7258,
+      "step": 2710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775530296717666e-05,
+      "loss": 0.7407,
+      "step": 2711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.769296118687623e-05,
+      "loss": 0.7398,
+      "step": 2712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763062030367054e-05,
+      "loss": 0.8376,
+      "step": 2713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.756828034180088e-05,
+      "loss": 0.8223,
+      "step": 2714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75059413255083e-05,
+      "loss": 0.6887,
+      "step": 2715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.744360327903329e-05,
+      "loss": 0.9347,
+      "step": 2716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738126622661613e-05,
+      "loss": 0.7076,
+      "step": 2717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73189301924967e-05,
+      "loss": 0.7138,
+      "step": 2718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725659520091433e-05,
+      "loss": 0.663,
+      "step": 2719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.719426127610816e-05,
+      "loss": 0.7002,
+      "step": 2720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.713192844231674e-05,
+      "loss": 0.673,
+      "step": 2721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.706959672377826e-05,
+      "loss": 0.7872,
+      "step": 2722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700726614473055e-05,
+      "loss": 0.7483,
+      "step": 2723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.694493672941085e-05,
+      "loss": 0.6986,
+      "step": 2724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.688260850205613e-05,
+      "loss": 0.7485,
+      "step": 2725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.682028148690266e-05,
+      "loss": 0.8204,
+      "step": 2726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.675795570818645e-05,
+      "loss": 0.7068,
+      "step": 2727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6695631190143e-05,
+      "loss": 0.6675,
+      "step": 2728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663330795700714e-05,
+      "loss": 0.7294,
+      "step": 2729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.657098603301346e-05,
+      "loss": 0.7302,
+      "step": 2730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.650866544239581e-05,
+      "loss": 0.7046,
+      "step": 2731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64463462093877e-05,
+      "loss": 0.8611,
+      "step": 2732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638402835822202e-05,
+      "loss": 0.7351,
+      "step": 2733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.632171191313108e-05,
+      "loss": 0.748,
+      "step": 2734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62593968983468e-05,
+      "loss": 0.7079,
+      "step": 2735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619708333810036e-05,
+      "loss": 0.7386,
+      "step": 2736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.613477125662253e-05,
+      "loss": 0.8598,
+      "step": 2737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.607246067814338e-05,
+      "loss": 0.7402,
+      "step": 2738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.601015162689245e-05,
+      "loss": 0.7515,
+      "step": 2739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594784412709877e-05,
+      "loss": 0.884,
+      "step": 2740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588553820299056e-05,
+      "loss": 0.7679,
+      "step": 2741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.582323387879568e-05,
+      "loss": 0.9042,
+      "step": 2742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576093117874112e-05,
+      "loss": 0.7543,
+      "step": 2743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.569863012705345e-05,
+      "loss": 0.7028,
+      "step": 2744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563633074795843e-05,
+      "loss": 0.7491,
+      "step": 2745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557403306568127e-05,
+      "loss": 0.7068,
+      "step": 2746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551173710444659e-05,
+      "loss": 0.8443,
+      "step": 2747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54494428884781e-05,
+      "loss": 0.7862,
+      "step": 2748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53871504419991e-05,
+      "loss": 0.7736,
+      "step": 2749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.532485978923198e-05,
+      "loss": 0.7106,
+      "step": 2750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.526257095439858e-05,
+      "loss": 0.6818,
+      "step": 2751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520028396172003e-05,
+      "loss": 0.7871,
+      "step": 2752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51379988354166e-05,
+      "loss": 0.6459,
+      "step": 2753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507571559970802e-05,
+      "loss": 0.7317,
+      "step": 2754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.501343427881317e-05,
+      "loss": 0.731,
+      "step": 2755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49511548969502e-05,
+      "loss": 0.7724,
+      "step": 2756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.488887747833655e-05,
+      "loss": 0.7516,
+      "step": 2757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.482660204718881e-05,
+      "loss": 0.7522,
+      "step": 2758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.476432862772296e-05,
+      "loss": 0.7097,
+      "step": 2759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.470205724415397e-05,
+      "loss": 0.7204,
+      "step": 2760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.463978792069624e-05,
+      "loss": 0.7389,
+      "step": 2761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.457752068156319e-05,
+      "loss": 0.7851,
+      "step": 2762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451525555096753e-05,
+      "loss": 0.7931,
+      "step": 2763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.445299255312116e-05,
+      "loss": 0.7798,
+      "step": 2764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.439073171223511e-05,
+      "loss": 0.7516,
+      "step": 2765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.432847305251957e-05,
+      "loss": 0.7057,
+      "step": 2766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.426621659818388e-05,
+      "loss": 0.6547,
+      "step": 2767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.420396237343654e-05,
+      "loss": 0.8478,
+      "step": 2768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.414171040248523e-05,
+      "loss": 0.7636,
+      "step": 2769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407946070953662e-05,
+      "loss": 0.7607,
+      "step": 2770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.401721331879665e-05,
+      "loss": 0.7603,
+      "step": 2771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395496825447022e-05,
+      "loss": 0.731,
+      "step": 2772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.389272554076141e-05,
+      "loss": 0.9425,
+      "step": 2773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383048520187344e-05,
+      "loss": 0.7203,
+      "step": 2774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.376824726200846e-05,
+      "loss": 0.7408,
+      "step": 2775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.370601174536781e-05,
+      "loss": 0.7535,
+      "step": 2776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.364377867615181e-05,
+      "loss": 0.7558,
+      "step": 2777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358154807855991e-05,
+      "loss": 0.7267,
+      "step": 2778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.351931997679049e-05,
+      "loss": 0.8245,
+      "step": 2779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.345709439504104e-05,
+      "loss": 0.7601,
+      "step": 2780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.339487135750811e-05,
+      "loss": 0.7381,
+      "step": 2781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.333265088838711e-05,
+      "loss": 0.7179,
+      "step": 2782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.327043301187262e-05,
+      "loss": 0.7529,
+      "step": 2783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.320821775215807e-05,
+      "loss": 0.7504,
+      "step": 2784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.314600513343595e-05,
+      "loss": 0.7605,
+      "step": 2785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308379517989773e-05,
+      "loss": 0.8203,
+      "step": 2786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.302158791573383e-05,
+      "loss": 0.7459,
+      "step": 2787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.295938336513361e-05,
+      "loss": 0.7522,
+      "step": 2788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.289718155228533e-05,
+      "loss": 0.7612,
+      "step": 2789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.283498250137636e-05,
+      "loss": 0.6803,
+      "step": 2790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277278623659273e-05,
+      "loss": 0.7631,
+      "step": 2791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.271059278211959e-05,
+      "loss": 0.6961,
+      "step": 2792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.264840216214099e-05,
+      "loss": 0.8027,
+      "step": 2793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.258621440083973e-05,
+      "loss": 0.7784,
+      "step": 2794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252402952239767e-05,
+      "loss": 0.7623,
+      "step": 2795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246184755099545e-05,
+      "loss": 0.7677,
+      "step": 2796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.239966851081255e-05,
+      "loss": 0.7251,
+      "step": 2797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.233749242602745e-05,
+      "loss": 0.7004,
+      "step": 2798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.227531932081733e-05,
+      "loss": 0.7897,
+      "step": 2799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.221314921935833e-05,
+      "loss": 0.7766,
+      "step": 2800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21509821458253e-05,
+      "loss": 0.7282,
+      "step": 2801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.208881812439198e-05,
+      "loss": 0.6775,
+      "step": 2802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.202665717923103e-05,
+      "loss": 0.8094,
+      "step": 2803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.196449933451367e-05,
+      "loss": 0.6505,
+      "step": 2804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.190234461441016e-05,
+      "loss": 0.7027,
+      "step": 2805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184019304308933e-05,
+      "loss": 0.7521,
+      "step": 2806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.177804464471898e-05,
+      "loss": 0.6391,
+      "step": 2807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171589944346552e-05,
+      "loss": 0.6863,
+      "step": 2808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.165375746349424e-05,
+      "loss": 0.6841,
+      "step": 2809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15916187289691e-05,
+      "loss": 0.8119,
+      "step": 2810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15294832640528e-05,
+      "loss": 0.8425,
+      "step": 2811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.146735109290683e-05,
+      "loss": 0.8277,
+      "step": 2812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.140522223969131e-05,
+      "loss": 0.835,
+      "step": 2813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134309672856514e-05,
+      "loss": 0.7337,
+      "step": 2814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.128097458368594e-05,
+      "loss": 0.8322,
+      "step": 2815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121885582920988e-05,
+      "loss": 0.6938,
+      "step": 2816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115674048929199e-05,
+      "loss": 0.7618,
+      "step": 2817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.109462858808586e-05,
+      "loss": 0.8097,
+      "step": 2818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10325201497438e-05,
+      "loss": 0.7183,
+      "step": 2819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09704151984167e-05,
+      "loss": 0.6806,
+      "step": 2820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090831375825416e-05,
+      "loss": 0.853,
+      "step": 2821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.084621585340448e-05,
+      "loss": 0.7144,
+      "step": 2822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078412150801437e-05,
+      "loss": 0.6742,
+      "step": 2823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07220307462294e-05,
+      "loss": 0.6827,
+      "step": 2824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.065994359219356e-05,
+      "loss": 0.7135,
+      "step": 2825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.059786007004954e-05,
+      "loss": 0.7783,
+      "step": 2826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.053578020393862e-05,
+      "loss": 0.7703,
+      "step": 2827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.047370401800055e-05,
+      "loss": 0.7059,
+      "step": 2828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041163153637381e-05,
+      "loss": 0.6735,
+      "step": 2829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.034956278319529e-05,
+      "loss": 0.6894,
+      "step": 2830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.028749778260054e-05,
+      "loss": 0.7335,
+      "step": 2831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022543655872363e-05,
+      "loss": 0.7468,
+      "step": 2832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016337913569706e-05,
+      "loss": 0.6844,
+      "step": 2833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.010132553765203e-05,
+      "loss": 0.6669,
+      "step": 2834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.003927578871807e-05,
+      "loss": 0.6517,
+      "step": 2835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.997722991302335e-05,
+      "loss": 0.6874,
+      "step": 2836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.991518793469443e-05,
+      "loss": 0.6347,
+      "step": 2837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.985314987785643e-05,
+      "loss": 0.8066,
+      "step": 2838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.979111576663296e-05,
+      "loss": 0.8273,
+      "step": 2839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.972908562514598e-05,
+      "loss": 0.9081,
+      "step": 2840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.966705947751609e-05,
+      "loss": 0.6871,
+      "step": 2841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96050373478621e-05,
+      "loss": 0.8323,
+      "step": 2842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954301926030146e-05,
+      "loss": 0.7342,
+      "step": 2843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.948100523895003e-05,
+      "loss": 0.781,
+      "step": 2844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94189953079219e-05,
+      "loss": 0.8106,
+      "step": 2845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.935698949132983e-05,
+      "loss": 0.6771,
+      "step": 2846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.929498781328477e-05,
+      "loss": 0.8144,
+      "step": 2847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923299029789617e-05,
+      "loss": 0.782,
+      "step": 2848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917099696927188e-05,
+      "loss": 0.813,
+      "step": 2849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.910900785151798e-05,
+      "loss": 0.6431,
+      "step": 2850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.904702296873912e-05,
+      "loss": 0.7303,
+      "step": 2851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.898504234503812e-05,
+      "loss": 0.7512,
+      "step": 2852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.892306600451628e-05,
+      "loss": 0.7073,
+      "step": 2853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886109397127308e-05,
+      "loss": 0.701,
+      "step": 2854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.879912626940647e-05,
+      "loss": 0.8253,
+      "step": 2855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.873716292301273e-05,
+      "loss": 0.7553,
+      "step": 2856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867520395618626e-05,
+      "loss": 0.6902,
+      "step": 2857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.861324939301998e-05,
+      "loss": 0.7332,
+      "step": 2858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.855129925760488e-05,
+      "loss": 0.6542,
+      "step": 2859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848935357403042e-05,
+      "loss": 0.6857,
+      "step": 2860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.842741236638427e-05,
+      "loss": 0.7005,
+      "step": 2861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836547565875227e-05,
+      "loss": 0.8073,
+      "step": 2862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.830354347521867e-05,
+      "loss": 0.8494,
+      "step": 2863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.824161583986578e-05,
+      "loss": 0.7832,
+      "step": 2864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817969277677433e-05,
+      "loss": 0.8155,
+      "step": 2865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811777431002307e-05,
+      "loss": 0.7499,
+      "step": 2866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80558604636891e-05,
+      "loss": 0.7056,
+      "step": 2867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.799395126184777e-05,
+      "loss": 0.7758,
+      "step": 2868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.793204672857244e-05,
+      "loss": 0.758,
+      "step": 2869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.787014688793482e-05,
+      "loss": 0.7775,
+      "step": 2870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78082517640047e-05,
+      "loss": 0.8302,
+      "step": 2871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.774636138085011e-05,
+      "loss": 0.7206,
+      "step": 2872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76844757625372e-05,
+      "loss": 0.7357,
+      "step": 2873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76225949331302e-05,
+      "loss": 0.7546,
+      "step": 2874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.756071891669164e-05,
+      "loss": 0.7657,
+      "step": 2875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.749884773728198e-05,
+      "loss": 0.7737,
+      "step": 2876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.743698141895994e-05,
+      "loss": 0.695,
+      "step": 2877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.737511998578236e-05,
+      "loss": 0.6434,
+      "step": 2878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.731326346180405e-05,
+      "loss": 0.7327,
+      "step": 2879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.725141187107808e-05,
+      "loss": 0.7668,
+      "step": 2880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718956523765541e-05,
+      "loss": 0.858,
+      "step": 2881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.712772358558526e-05,
+      "loss": 0.8501,
+      "step": 2882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.706588693891475e-05,
+      "loss": 0.6905,
+      "step": 2883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70040553216892e-05,
+      "loss": 0.8744,
+      "step": 2884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.694222875795192e-05,
+      "loss": 0.8403,
+      "step": 2885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.688040727174418e-05,
+      "loss": 0.7361,
+      "step": 2886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68185908871054e-05,
+      "loss": 0.8013,
+      "step": 2887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675677962807284e-05,
+      "loss": 0.7868,
+      "step": 2888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669497351868199e-05,
+      "loss": 0.6588,
+      "step": 2889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663317258296624e-05,
+      "loss": 0.7428,
+      "step": 2890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.657137684495686e-05,
+      "loss": 0.6746,
+      "step": 2891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.650958632868325e-05,
+      "loss": 0.7393,
+      "step": 2892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.644780105817271e-05,
+      "loss": 0.7501,
+      "step": 2893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.638602105745058e-05,
+      "loss": 0.6854,
+      "step": 2894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.632424635053997e-05,
+      "loss": 0.7071,
+      "step": 2895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62624769614621e-05,
+      "loss": 0.6023,
+      "step": 2896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.620071291423613e-05,
+      "loss": 0.7458,
+      "step": 2897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613895423287897e-05,
+      "loss": 0.7256,
+      "step": 2898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607720094140567e-05,
+      "loss": 0.6878,
+      "step": 2899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.601545306382895e-05,
+      "loss": 0.7513,
+      "step": 2900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.595371062415962e-05,
+      "loss": 0.8262,
+      "step": 2901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58919736464063e-05,
+      "loss": 0.7275,
+      "step": 2902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.583024215457548e-05,
+      "loss": 0.8441,
+      "step": 2903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57685161726715e-05,
+      "loss": 0.6663,
+      "step": 2904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570679572469658e-05,
+      "loss": 0.728,
+      "step": 2905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.564508083465079e-05,
+      "loss": 0.6396,
+      "step": 2906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.558337152653211e-05,
+      "loss": 0.7271,
+      "step": 2907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.552166782433617e-05,
+      "loss": 0.6134,
+      "step": 2908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.545996975205659e-05,
+      "loss": 0.7703,
+      "step": 2909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53982773336847e-05,
+      "loss": 0.752,
+      "step": 2910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533659059320971e-05,
+      "loss": 0.681,
+      "step": 2911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.527490955461852e-05,
+      "loss": 0.7238,
+      "step": 2912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.521323424189593e-05,
+      "loss": 0.7749,
+      "step": 2913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.515156467902444e-05,
+      "loss": 0.7466,
+      "step": 2914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.508990088998431e-05,
+      "loss": 0.6694,
+      "step": 2915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502824289875365e-05,
+      "loss": 0.7042,
+      "step": 2916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496659072930813e-05,
+      "loss": 0.8644,
+      "step": 2917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.490494440562135e-05,
+      "loss": 0.7347,
+      "step": 2918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.484330395166455e-05,
+      "loss": 0.7006,
+      "step": 2919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.478166939140664e-05,
+      "loss": 0.761,
+      "step": 2920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472004074881437e-05,
+      "loss": 0.7287,
+      "step": 2921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.465841804785202e-05,
+      "loss": 0.6841,
+      "step": 2922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.459680131248169e-05,
+      "loss": 0.7809,
+      "step": 2923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.453519056666315e-05,
+      "loss": 0.7775,
+      "step": 2924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.447358583435377e-05,
+      "loss": 0.7626,
+      "step": 2925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.441198713950865e-05,
+      "loss": 0.7305,
+      "step": 2926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.435039450608046e-05,
+      "loss": 0.567,
+      "step": 2927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.428880795801965e-05,
+      "loss": 0.6624,
+      "step": 2928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.422722751927415e-05,
+      "loss": 0.8058,
+      "step": 2929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.416565321378961e-05,
+      "loss": 0.8178,
+      "step": 2930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41040850655093e-05,
+      "loss": 0.7879,
+      "step": 2931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4042523098374e-05,
+      "loss": 0.6553,
+      "step": 2932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.398096733632223e-05,
+      "loss": 0.7112,
+      "step": 2933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.391941780328995e-05,
+      "loss": 0.6804,
+      "step": 2934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.385787452321081e-05,
+      "loss": 0.7661,
+      "step": 2935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.379633752001597e-05,
+      "loss": 0.8017,
+      "step": 2936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.373480681763417e-05,
+      "loss": 0.7207,
+      "step": 2937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.367328243999173e-05,
+      "loss": 0.6726,
+      "step": 2938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36117644110124e-05,
+      "loss": 0.7209,
+      "step": 2939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355025275461764e-05,
+      "loss": 0.8091,
+      "step": 2940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348874749472619e-05,
+      "loss": 0.7498,
+      "step": 2941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.342724865525455e-05,
+      "loss": 0.7363,
+      "step": 2942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33657562601166e-05,
+      "loss": 0.7733,
+      "step": 2943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33042703332237e-05,
+      "loss": 0.7209,
+      "step": 2944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.324279089848475e-05,
+      "loss": 0.7211,
+      "step": 2945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.318131797980606e-05,
+      "loss": 0.6785,
+      "step": 2946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.311985160109149e-05,
+      "loss": 0.6765,
+      "step": 2947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.305839178624233e-05,
+      "loss": 0.6717,
+      "step": 2948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299693855915724e-05,
+      "loss": 0.6372,
+      "step": 2949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.293549194373243e-05,
+      "loss": 0.6356,
+      "step": 2950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.287405196386144e-05,
+      "loss": 0.6234,
+      "step": 2951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28126186434353e-05,
+      "loss": 0.6529,
+      "step": 2952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.275119200634247e-05,
+      "loss": 0.7801,
+      "step": 2953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26897720764687e-05,
+      "loss": 0.7629,
+      "step": 2954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262835887769725e-05,
+      "loss": 0.795,
+      "step": 2955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256695243390869e-05,
+      "loss": 0.7631,
+      "step": 2956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2505552768981e-05,
+      "loss": 0.7766,
+      "step": 2957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.244415990678948e-05,
+      "loss": 0.8622,
+      "step": 2958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238277387120685e-05,
+      "loss": 0.81,
+      "step": 2959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.232139468610316e-05,
+      "loss": 0.7104,
+      "step": 2960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226002237534572e-05,
+      "loss": 0.672,
+      "step": 2961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.219865696279931e-05,
+      "loss": 0.7709,
+      "step": 2962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213729847232582e-05,
+      "loss": 0.692,
+      "step": 2963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.207594692778464e-05,
+      "loss": 0.8582,
+      "step": 2964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.201460235303245e-05,
+      "loss": 0.7487,
+      "step": 2965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.195326477192305e-05,
+      "loss": 0.6931,
+      "step": 2966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.189193420830767e-05,
+      "loss": 0.7713,
+      "step": 2967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183061068603479e-05,
+      "loss": 0.7304,
+      "step": 2968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17692942289501e-05,
+      "loss": 0.7488,
+      "step": 2969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.170798486089666e-05,
+      "loss": 0.7405,
+      "step": 2970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.164668260571456e-05,
+      "loss": 0.7988,
+      "step": 2971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.158538748724139e-05,
+      "loss": 0.771,
+      "step": 2972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.152409952931169e-05,
+      "loss": 0.7731,
+      "step": 2973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.146281875575747e-05,
+      "loss": 0.7613,
+      "step": 2974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.140154519040774e-05,
+      "loss": 0.681,
+      "step": 2975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134027885708883e-05,
+      "loss": 0.6996,
+      "step": 2976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.127901977962427e-05,
+      "loss": 0.7314,
+      "step": 2977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121776798183467e-05,
+      "loss": 0.788,
+      "step": 2978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115652348753791e-05,
+      "loss": 0.7489,
+      "step": 2979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.109528632054893e-05,
+      "loss": 0.6936,
+      "step": 2980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103405650467988e-05,
+      "loss": 0.7737,
+      "step": 2981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.097283406374011e-05,
+      "loss": 0.8931,
+      "step": 2982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.091161902153595e-05,
+      "loss": 0.7148,
+      "step": 2983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.085041140187101e-05,
+      "loss": 0.7266,
+      "step": 2984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.078921122854589e-05,
+      "loss": 0.7424,
+      "step": 2985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07280185253584e-05,
+      "loss": 0.6107,
+      "step": 2986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.066683331610333e-05,
+      "loss": 0.7365,
+      "step": 2987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.060565562457269e-05,
+      "loss": 0.8401,
+      "step": 2988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.054448547455545e-05,
+      "loss": 0.7531,
+      "step": 2989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.048332288983769e-05,
+      "loss": 0.7368,
+      "step": 2990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042216789420263e-05,
+      "loss": 0.7071,
+      "step": 2991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.036102051143035e-05,
+      "loss": 0.7782,
+      "step": 2992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.029988076529814e-05,
+      "loss": 0.7197,
+      "step": 2993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.023874867958027e-05,
+      "loss": 0.6973,
+      "step": 2994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.017762427804799e-05,
+      "loss": 0.5931,
+      "step": 2995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.011650758446966e-05,
+      "loss": 0.7017,
+      "step": 2996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005539862261047e-05,
+      "loss": 0.718,
+      "step": 2997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999429741623278e-05,
+      "loss": 0.7093,
+      "step": 2998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.993320398909587e-05,
+      "loss": 0.7125,
+      "step": 2999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987211836495596e-05,
+      "loss": 0.7487,
+      "step": 3000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981104056756633e-05,
+      "loss": 0.7002,
+      "step": 3001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.974997062067706e-05,
+      "loss": 0.8264,
+      "step": 3002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.968890854803536e-05,
+      "loss": 0.734,
+      "step": 3003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.962785437338519e-05,
+      "loss": 0.7152,
+      "step": 3004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95668081204676e-05,
+      "loss": 0.7283,
+      "step": 3005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.950576981302053e-05,
+      "loss": 0.7958,
+      "step": 3006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.944473947477871e-05,
+      "loss": 0.7773,
+      "step": 3007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.938371712947392e-05,
+      "loss": 0.7404,
+      "step": 3008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932270280083474e-05,
+      "loss": 0.8159,
+      "step": 3009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926169651258669e-05,
+      "loss": 0.8019,
+      "step": 3010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.920069828845211e-05,
+      "loss": 0.6516,
+      "step": 3011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.913970815215024e-05,
+      "loss": 0.6738,
+      "step": 3012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90787261273972e-05,
+      "loss": 0.7827,
+      "step": 3013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.901775223790584e-05,
+      "loss": 0.6955,
+      "step": 3014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.895678650738603e-05,
+      "loss": 0.8019,
+      "step": 3015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.889582895954427e-05,
+      "loss": 0.7084,
+      "step": 3016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.883487961808402e-05,
+      "loss": 0.6526,
+      "step": 3017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87739385067055e-05,
+      "loss": 0.7959,
+      "step": 3018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871300564910575e-05,
+      "loss": 0.7099,
+      "step": 3019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.865208106897856e-05,
+      "loss": 0.719,
+      "step": 3020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85911647900145e-05,
+      "loss": 0.6652,
+      "step": 3021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.853025683590096e-05,
+      "loss": 0.7298,
+      "step": 3022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.846935723032213e-05,
+      "loss": 0.7028,
+      "step": 3023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.840846599695878e-05,
+      "loss": 0.7245,
+      "step": 3024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.834758315948864e-05,
+      "loss": 0.7634,
+      "step": 3025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.828670874158597e-05,
+      "loss": 0.699,
+      "step": 3026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822584276692191e-05,
+      "loss": 0.7231,
+      "step": 3027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.816498525916434e-05,
+      "loss": 0.8874,
+      "step": 3028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.810413624197763e-05,
+      "loss": 0.6935,
+      "step": 3029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80432957390231e-05,
+      "loss": 0.8185,
+      "step": 3030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.798246377395858e-05,
+      "loss": 0.7638,
+      "step": 3031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.792164037043872e-05,
+      "loss": 0.7639,
+      "step": 3032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.786082555211471e-05,
+      "loss": 0.7169,
+      "step": 3033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.780001934263448e-05,
+      "loss": 0.7549,
+      "step": 3034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.773922176564263e-05,
+      "loss": 0.8267,
+      "step": 3035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.767843284478032e-05,
+      "loss": 0.7028,
+      "step": 3036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.761765260368546e-05,
+      "loss": 0.8791,
+      "step": 3037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.755688106599241e-05,
+      "loss": 0.7336,
+      "step": 3038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.749611825533232e-05,
+      "loss": 0.8075,
+      "step": 3039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.743536419533289e-05,
+      "loss": 0.7163,
+      "step": 3040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73746189096184e-05,
+      "loss": 0.6325,
+      "step": 3041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.731388242180972e-05,
+      "loss": 0.6553,
+      "step": 3042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.725315475552427e-05,
+      "loss": 0.8059,
+      "step": 3043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.719243593437611e-05,
+      "loss": 0.6772,
+      "step": 3044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713172598197586e-05,
+      "loss": 0.7787,
+      "step": 3045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707102492193056e-05,
+      "loss": 0.75,
+      "step": 3046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.701033277784401e-05,
+      "loss": 0.7705,
+      "step": 3047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.694964957331629e-05,
+      "loss": 0.9248,
+      "step": 3048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688897533194424e-05,
+      "loss": 0.7417,
+      "step": 3049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.682831007732106e-05,
+      "loss": 0.749,
+      "step": 3050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.676765383303649e-05,
+      "loss": 0.7385,
+      "step": 3051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670700662267682e-05,
+      "loss": 0.6059,
+      "step": 3052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664636846982476e-05,
+      "loss": 0.6659,
+      "step": 3053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.658573939805957e-05,
+      "loss": 0.6786,
+      "step": 3054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.652511943095687e-05,
+      "loss": 0.7605,
+      "step": 3055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.646450859208883e-05,
+      "loss": 0.6728,
+      "step": 3056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.640390690502411e-05,
+      "loss": 0.8072,
+      "step": 3057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.634331439332765e-05,
+      "loss": 0.7428,
+      "step": 3058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.628273108056098e-05,
+      "loss": 0.5986,
+      "step": 3059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.622215699028196e-05,
+      "loss": 0.7773,
+      "step": 3060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.61615921460449e-05,
+      "loss": 0.7401,
+      "step": 3061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.610103657140053e-05,
+      "loss": 0.7638,
+      "step": 3062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.60404902898959e-05,
+      "loss": 0.6363,
+      "step": 3063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597995332507462e-05,
+      "loss": 0.6856,
+      "step": 3064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591942570047643e-05,
+      "loss": 0.7555,
+      "step": 3065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585890743963768e-05,
+      "loss": 0.7113,
+      "step": 3066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.579839856609084e-05,
+      "loss": 0.7307,
+      "step": 3067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.573789910336494e-05,
+      "loss": 0.7393,
+      "step": 3068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567740907498529e-05,
+      "loss": 0.7112,
+      "step": 3069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.561692850447342e-05,
+      "loss": 0.5622,
+      "step": 3070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.555645741534736e-05,
+      "loss": 0.6144,
+      "step": 3071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.549599583112131e-05,
+      "loss": 0.7772,
+      "step": 3072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.543554377530582e-05,
+      "loss": 0.7504,
+      "step": 3073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.537510127140778e-05,
+      "loss": 0.7503,
+      "step": 3074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.531466834293028e-05,
+      "loss": 0.7303,
+      "step": 3075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525424501337284e-05,
+      "loss": 0.7407,
+      "step": 3076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5193831306231e-05,
+      "loss": 0.748,
+      "step": 3077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513342724499681e-05,
+      "loss": 0.7369,
+      "step": 3078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.507303285315839e-05,
+      "loss": 0.8284,
+      "step": 3079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.501264815420019e-05,
+      "loss": 0.6548,
+      "step": 3080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.495227317160292e-05,
+      "loss": 0.6727,
+      "step": 3081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489190792884338e-05,
+      "loss": 0.707,
+      "step": 3082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48315524493947e-05,
+      "loss": 0.6436,
+      "step": 3083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.47712067567262e-05,
+      "loss": 0.7425,
+      "step": 3084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.471087087430334e-05,
+      "loss": 0.6916,
+      "step": 3085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.465054482558787e-05,
+      "loss": 0.6472,
+      "step": 3086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.459022863403753e-05,
+      "loss": 0.6939,
+      "step": 3087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.452992232310644e-05,
+      "loss": 0.7847,
+      "step": 3088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446962591624472e-05,
+      "loss": 0.8384,
+      "step": 3089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.440933943689874e-05,
+      "loss": 0.8171,
+      "step": 3090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.434906290851091e-05,
+      "loss": 0.728,
+      "step": 3091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428879635451986e-05,
+      "loss": 0.7289,
+      "step": 3092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.422853979836034e-05,
+      "loss": 0.8926,
+      "step": 3093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.416829326346313e-05,
+      "loss": 0.7431,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.410805677325523e-05,
+      "loss": 0.7164,
+      "step": 3095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.404783035115961e-05,
+      "loss": 0.7625,
+      "step": 3096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.398761402059539e-05,
+      "loss": 0.7448,
+      "step": 3097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.392740780497781e-05,
+      "loss": 0.7743,
+      "step": 3098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386721172771807e-05,
+      "loss": 0.6696,
+      "step": 3099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.380702581222353e-05,
+      "loss": 0.7297,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.37468500818975e-05,
+      "loss": 0.682,
+      "step": 3101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.368668456013941e-05,
+      "loss": 0.7797,
+      "step": 3102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362652927034473e-05,
+      "loss": 0.917,
+      "step": 3103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356638423590485e-05,
+      "loss": 0.7357,
+      "step": 3104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.350624948020728e-05,
+      "loss": 0.7414,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.344612502663548e-05,
+      "loss": 0.7586,
+      "step": 3106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.338601089856893e-05,
+      "loss": 0.7208,
+      "step": 3107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.332590711938303e-05,
+      "loss": 0.7964,
+      "step": 3108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.326581371244923e-05,
+      "loss": 0.8155,
+      "step": 3109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.320573070113496e-05,
+      "loss": 0.8328,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.31456581088035e-05,
+      "loss": 0.7216,
+      "step": 3111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.308559595881424e-05,
+      "loss": 0.7845,
+      "step": 3112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.30255442745223e-05,
+      "loss": 0.6895,
+      "step": 3113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.296550307927892e-05,
+      "loss": 0.6721,
+      "step": 3114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.290547239643117e-05,
+      "loss": 0.8446,
+      "step": 3115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.284545224932207e-05,
+      "loss": 0.6929,
+      "step": 3116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.278544266129053e-05,
+      "loss": 0.7949,
+      "step": 3117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.272544365567132e-05,
+      "loss": 0.7633,
+      "step": 3118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.26654552557951e-05,
+      "loss": 0.7801,
+      "step": 3119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.26054774849885e-05,
+      "loss": 0.7102,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.254551036657389e-05,
+      "loss": 0.7677,
+      "step": 3121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24855539238696e-05,
+      "loss": 0.7636,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24256081801897e-05,
+      "loss": 0.7183,
+      "step": 3123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236567315884422e-05,
+      "loss": 0.7244,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.23057488831389e-05,
+      "loss": 0.7133,
+      "step": 3125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.224583537637544e-05,
+      "loss": 0.6894,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.218593266185125e-05,
+      "loss": 0.6414,
+      "step": 3127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.212604076285951e-05,
+      "loss": 0.7789,
+      "step": 3128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.206615970268937e-05,
+      "loss": 0.7432,
+      "step": 3129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.200628950462556e-05,
+      "loss": 0.7124,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.194643019194867e-05,
+      "loss": 0.74,
+      "step": 3131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.188658178793515e-05,
+      "loss": 0.6778,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.182674431585704e-05,
+      "loss": 0.7425,
+      "step": 3133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.176691779898223e-05,
+      "loss": 0.7565,
+      "step": 3134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.170710226057438e-05,
+      "loss": 0.7042,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.164729772389275e-05,
+      "loss": 0.7506,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158750421219244e-05,
+      "loss": 0.7355,
+      "step": 3137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.152772174872424e-05,
+      "loss": 0.7734,
+      "step": 3138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.146795035673466e-05,
+      "loss": 0.715,
+      "step": 3139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.14081900594658e-05,
+      "loss": 0.8082,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.13484408801556e-05,
+      "loss": 0.7104,
+      "step": 3141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.12887028420375e-05,
+      "loss": 0.7774,
+      "step": 3142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.122897596834077e-05,
+      "loss": 0.6864,
+      "step": 3143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.116926028229032e-05,
+      "loss": 0.6832,
+      "step": 3144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.110955580710656e-05,
+      "loss": 0.7845,
+      "step": 3145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.104986256600573e-05,
+      "loss": 0.7689,
+      "step": 3146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.099018058219953e-05,
+      "loss": 0.7306,
+      "step": 3147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.093050987889547e-05,
+      "loss": 0.6859,
+      "step": 3148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.087085047929652e-05,
+      "loss": 0.7778,
+      "step": 3149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.081120240660127e-05,
+      "loss": 0.7358,
+      "step": 3150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.075156568400405e-05,
+      "loss": 0.8447,
+      "step": 3151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.069194033469456e-05,
+      "loss": 0.7187,
+      "step": 3152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.063232638185825e-05,
+      "loss": 0.7164,
+      "step": 3153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.057272384867603e-05,
+      "loss": 0.7019,
+      "step": 3154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.051313275832443e-05,
+      "loss": 0.8409,
+      "step": 3155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.045355313397555e-05,
+      "loss": 0.7605,
+      "step": 3156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.039398499879697e-05,
+      "loss": 0.7801,
+      "step": 3157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.033442837595182e-05,
+      "loss": 0.7185,
+      "step": 3158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027488328859876e-05,
+      "loss": 0.7061,
+      "step": 3159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.021534975989197e-05,
+      "loss": 0.7134,
+      "step": 3160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.015582781298118e-05,
+      "loss": 0.7571,
+      "step": 3161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00963174710115e-05,
+      "loss": 0.6416,
+      "step": 3162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.003681875712366e-05,
+      "loss": 0.7148,
+      "step": 3163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.997733169445372e-05,
+      "loss": 0.7161,
+      "step": 3164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.991785630613334e-05,
+      "loss": 0.7131,
+      "step": 3165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.985839261528965e-05,
+      "loss": 0.6592,
+      "step": 3166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.979894064504508e-05,
+      "loss": 0.636,
+      "step": 3167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.973950041851765e-05,
+      "loss": 0.9232,
+      "step": 3168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.968007195882073e-05,
+      "loss": 0.79,
+      "step": 3169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.96206552890632e-05,
+      "loss": 0.7678,
+      "step": 3170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.956125043234925e-05,
+      "loss": 0.6481,
+      "step": 3171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.950185741177852e-05,
+      "loss": 0.7108,
+      "step": 3172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.944247625044615e-05,
+      "loss": 0.6578,
+      "step": 3173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.938310697144244e-05,
+      "loss": 0.6905,
+      "step": 3174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.932374959785333e-05,
+      "loss": 0.6958,
+      "step": 3175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.926440415275988e-05,
+      "loss": 0.7285,
+      "step": 3176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.920507065923871e-05,
+      "loss": 0.736,
+      "step": 3177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.914574914036171e-05,
+      "loss": 0.6364,
+      "step": 3178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.908643961919614e-05,
+      "loss": 0.7598,
+      "step": 3179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.902714211880455e-05,
+      "loss": 0.7099,
+      "step": 3180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.896785666224481e-05,
+      "loss": 0.6612,
+      "step": 3181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.890858327257021e-05,
+      "loss": 0.7538,
+      "step": 3182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.884932197282923e-05,
+      "loss": 0.6584,
+      "step": 3183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879007278606568e-05,
+      "loss": 0.6909,
+      "step": 3184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.873083573531876e-05,
+      "loss": 0.8767,
+      "step": 3185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.867161084362275e-05,
+      "loss": 0.6944,
+      "step": 3186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.861239813400741e-05,
+      "loss": 0.6672,
+      "step": 3187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.855319762949764e-05,
+      "loss": 0.7107,
+      "step": 3188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.849400935311359e-05,
+      "loss": 0.9109,
+      "step": 3189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.843483332787075e-05,
+      "loss": 0.6653,
+      "step": 3190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.837566957677975e-05,
+      "loss": 0.6685,
+      "step": 3191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.831651812284652e-05,
+      "loss": 0.7815,
+      "step": 3192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.825737898907209e-05,
+      "loss": 0.684,
+      "step": 3193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.819825219845286e-05,
+      "loss": 0.6084,
+      "step": 3194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.813913777398037e-05,
+      "loss": 0.6871,
+      "step": 3195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808003573864124e-05,
+      "loss": 0.6854,
+      "step": 3196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.802094611541745e-05,
+      "loss": 0.7447,
+      "step": 3197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7961868927286e-05,
+      "loss": 0.6185,
+      "step": 3198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.790280419721918e-05,
+      "loss": 0.8349,
+      "step": 3199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.784375194818434e-05,
+      "loss": 0.7851,
+      "step": 3200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.778471220314403e-05,
+      "loss": 0.7205,
+      "step": 3201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.772568498505596e-05,
+      "loss": 0.6327,
+      "step": 3202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.766667031687286e-05,
+      "loss": 0.6931,
+      "step": 3203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760766822154275e-05,
+      "loss": 0.7471,
+      "step": 3204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.754867872200854e-05,
+      "loss": 0.8281,
+      "step": 3205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.748970184120845e-05,
+      "loss": 0.7771,
+      "step": 3206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.743073760207573e-05,
+      "loss": 0.753,
+      "step": 3207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.737178602753859e-05,
+      "loss": 0.6966,
+      "step": 3208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.73128471405205e-05,
+      "loss": 0.8076,
+      "step": 3209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.72539209639399e-05,
+      "loss": 0.7607,
+      "step": 3210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719500752071033e-05,
+      "loss": 0.7819,
+      "step": 3211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.713610683374028e-05,
+      "loss": 0.6344,
+      "step": 3212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.70772189259334e-05,
+      "loss": 0.7386,
+      "step": 3213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.701834382018832e-05,
+      "loss": 0.7366,
+      "step": 3214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695948153939868e-05,
+      "loss": 0.7217,
+      "step": 3215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.690063210645317e-05,
+      "loss": 0.7813,
+      "step": 3216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.684179554423542e-05,
+      "loss": 0.743,
+      "step": 3217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.67829718756241e-05,
+      "loss": 0.7672,
+      "step": 3218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.672416112349293e-05,
+      "loss": 0.6896,
+      "step": 3219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.666536331071045e-05,
+      "loss": 0.801,
+      "step": 3220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660657846014031e-05,
+      "loss": 0.8003,
+      "step": 3221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.654780659464105e-05,
+      "loss": 0.7709,
+      "step": 3222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.648904773706616e-05,
+      "loss": 0.7673,
+      "step": 3223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.643030191026418e-05,
+      "loss": 0.8138,
+      "step": 3224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.637156913707839e-05,
+      "loss": 0.7954,
+      "step": 3225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.631284944034716e-05,
+      "loss": 0.7444,
+      "step": 3226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.625414284290365e-05,
+      "loss": 0.7459,
+      "step": 3227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.619544936757607e-05,
+      "loss": 0.7704,
+      "step": 3228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.613676903718736e-05,
+      "loss": 0.7622,
+      "step": 3229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.607810187455546e-05,
+      "loss": 0.7314,
+      "step": 3230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601944790249322e-05,
+      "loss": 0.6109,
+      "step": 3231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596080714380825e-05,
+      "loss": 0.6808,
+      "step": 3232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.590217962130312e-05,
+      "loss": 0.6931,
+      "step": 3233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.584356535777516e-05,
+      "loss": 0.6857,
+      "step": 3234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.578496437601662e-05,
+      "loss": 0.7555,
+      "step": 3235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.572637669881458e-05,
+      "loss": 0.8348,
+      "step": 3236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.566780234895089e-05,
+      "loss": 0.6437,
+      "step": 3237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56092413492023e-05,
+      "loss": 0.7398,
+      "step": 3238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.555069372234025e-05,
+      "loss": 0.7191,
+      "step": 3239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.54921594911311e-05,
+      "loss": 0.7333,
+      "step": 3240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.543363867833596e-05,
+      "loss": 0.7639,
+      "step": 3241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.537513130671072e-05,
+      "loss": 0.6703,
+      "step": 3242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.531663739900604e-05,
+      "loss": 0.7794,
+      "step": 3243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.525815697796729e-05,
+      "loss": 0.7414,
+      "step": 3244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519969006633473e-05,
+      "loss": 0.7007,
+      "step": 3245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.514123668684321e-05,
+      "loss": 0.6543,
+      "step": 3246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.508279686222243e-05,
+      "loss": 0.6559,
+      "step": 3247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.502437061519684e-05,
+      "loss": 0.7967,
+      "step": 3248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.496595796848545e-05,
+      "loss": 0.789,
+      "step": 3249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.490755894480217e-05,
+      "loss": 0.7243,
+      "step": 3250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484917356685546e-05,
+      "loss": 0.6826,
+      "step": 3251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479080185734858e-05,
+      "loss": 0.753,
+      "step": 3252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473244383897945e-05,
+      "loss": 0.7425,
+      "step": 3253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467409953444061e-05,
+      "loss": 0.7973,
+      "step": 3254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.461576896641939e-05,
+      "loss": 0.7714,
+      "step": 3255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.45574521575976e-05,
+      "loss": 0.7013,
+      "step": 3256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44991491306519e-05,
+      "loss": 0.8815,
+      "step": 3257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.444085990825338e-05,
+      "loss": 0.7491,
+      "step": 3258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.438258451306793e-05,
+      "loss": 0.7335,
+      "step": 3259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.432432296775606e-05,
+      "loss": 0.6806,
+      "step": 3260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.426607529497273e-05,
+      "loss": 0.7242,
+      "step": 3261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.42078415173677e-05,
+      "loss": 0.7456,
+      "step": 3262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.414962165758516e-05,
+      "loss": 0.7829,
+      "step": 3263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.409141573826407e-05,
+      "loss": 0.7022,
+      "step": 3264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.403322378203782e-05,
+      "loss": 0.7192,
+      "step": 3265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.39750458115344e-05,
+      "loss": 0.6657,
+      "step": 3266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.391688184937643e-05,
+      "loss": 0.7154,
+      "step": 3267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3858731918181e-05,
+      "loss": 0.7123,
+      "step": 3268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.380059604055974e-05,
+      "loss": 0.7367,
+      "step": 3269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.374247423911896e-05,
+      "loss": 0.6341,
+      "step": 3270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.36843665364593e-05,
+      "loss": 0.8146,
+      "step": 3271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.362627295517606e-05,
+      "loss": 0.6223,
+      "step": 3272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.356819351785896e-05,
+      "loss": 0.7317,
+      "step": 3273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.351012824709229e-05,
+      "loss": 0.7585,
+      "step": 3274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.345207716545473e-05,
+      "loss": 0.7222,
+      "step": 3275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.339404029551959e-05,
+      "loss": 0.7069,
+      "step": 3276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333601765985457e-05,
+      "loss": 0.723,
+      "step": 3277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.327800928102178e-05,
+      "loss": 0.7627,
+      "step": 3278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.322001518157791e-05,
+      "loss": 0.6806,
+      "step": 3279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.316203538407397e-05,
+      "loss": 0.7458,
+      "step": 3280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.310406991105551e-05,
+      "loss": 0.6283,
+      "step": 3281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.304611878506251e-05,
+      "loss": 0.7247,
+      "step": 3282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.298818202862924e-05,
+      "loss": 0.7096,
+      "step": 3283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.293025966428455e-05,
+      "loss": 0.7078,
+      "step": 3284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287235171455157e-05,
+      "loss": 0.7213,
+      "step": 3285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.281445820194794e-05,
+      "loss": 0.7549,
+      "step": 3286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.275657914898554e-05,
+      "loss": 0.7197,
+      "step": 3287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.269871457817074e-05,
+      "loss": 0.7437,
+      "step": 3288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26408645120043e-05,
+      "loss": 0.7624,
+      "step": 3289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.258302897298118e-05,
+      "loss": 0.7498,
+      "step": 3290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252520798359092e-05,
+      "loss": 0.6878,
+      "step": 3291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.246740156631714e-05,
+      "loss": 0.8759,
+      "step": 3292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240960974363803e-05,
+      "loss": 0.6415,
+      "step": 3293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.235183253802599e-05,
+      "loss": 0.6875,
+      "step": 3294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.229406997194775e-05,
+      "loss": 0.7445,
+      "step": 3295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.223632206786434e-05,
+      "loss": 0.5654,
+      "step": 3296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.21785888482311e-05,
+      "loss": 0.7018,
+      "step": 3297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.212087033549765e-05,
+      "loss": 0.8239,
+      "step": 3298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206316655210798e-05,
+      "loss": 0.7417,
+      "step": 3299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.200547752050014e-05,
+      "loss": 0.655,
+      "step": 3300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.194780326310671e-05,
+      "loss": 0.6937,
+      "step": 3301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18901438023543e-05,
+      "loss": 0.6612,
+      "step": 3302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18324991606639e-05,
+      "loss": 0.745,
+      "step": 3303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177486936045068e-05,
+      "loss": 0.6757,
+      "step": 3304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.171725442412404e-05,
+      "loss": 0.7782,
+      "step": 3305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.165965437408766e-05,
+      "loss": 0.7813,
+      "step": 3306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.160206923273934e-05,
+      "loss": 0.7888,
+      "step": 3307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.154449902247119e-05,
+      "loss": 0.8265,
+      "step": 3308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.148694376566937e-05,
+      "loss": 0.6429,
+      "step": 3309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.142940348471436e-05,
+      "loss": 0.7389,
+      "step": 3310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13718782019808e-05,
+      "loss": 0.8054,
+      "step": 3311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.131436793983739e-05,
+      "loss": 0.7642,
+      "step": 3312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.125687272064713e-05,
+      "loss": 0.684,
+      "step": 3313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.119939256676701e-05,
+      "loss": 0.7161,
+      "step": 3314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.114192750054832e-05,
+      "loss": 0.7051,
+      "step": 3315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.108447754433643e-05,
+      "loss": 0.7396,
+      "step": 3316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1027042720470794e-05,
+      "loss": 0.708,
+      "step": 3317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.096962305128506e-05,
+      "loss": 0.692,
+      "step": 3318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.091221855910685e-05,
+      "loss": 0.7688,
+      "step": 3319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.085482926625803e-05,
+      "loss": 0.7246,
+      "step": 3320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0797455195054454e-05,
+      "loss": 0.6638,
+      "step": 3321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.074009636780611e-05,
+      "loss": 0.7145,
+      "step": 3322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0682752806817087e-05,
+      "loss": 0.595,
+      "step": 3323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0625424534385425e-05,
+      "loss": 0.7351,
+      "step": 3324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056811157280333e-05,
+      "loss": 0.7474,
+      "step": 3325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0510813944357025e-05,
+      "loss": 0.805,
+      "step": 3326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0453531671326714e-05,
+      "loss": 0.7857,
+      "step": 3327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.039626477598672e-05,
+      "loss": 0.7387,
+      "step": 3328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0339013280605314e-05,
+      "loss": 0.6864,
+      "step": 3329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0281777207444853e-05,
+      "loss": 0.767,
+      "step": 3330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.022455657876158e-05,
+      "loss": 0.6774,
+      "step": 3331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0167351416805874e-05,
+      "loss": 0.7257,
+      "step": 3332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0110161743821955e-05,
+      "loss": 0.6822,
+      "step": 3333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0052987582048136e-05,
+      "loss": 0.7063,
+      "step": 3334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9995828953716695e-05,
+      "loss": 0.7545,
+      "step": 3335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.993868588105375e-05,
+      "loss": 0.7966,
+      "step": 3336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9881558386279514e-05,
+      "loss": 0.6739,
+      "step": 3337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.982444649160803e-05,
+      "loss": 0.7205,
+      "step": 3338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.976735021924735e-05,
+      "loss": 0.6608,
+      "step": 3339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.971026959139947e-05,
+      "loss": 0.7228,
+      "step": 3340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9653204630260185e-05,
+      "loss": 0.7977,
+      "step": 3341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.959615535801933e-05,
+      "loss": 0.8164,
+      "step": 3342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.953912179686052e-05,
+      "loss": 0.6966,
+      "step": 3343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.948210396896137e-05,
+      "loss": 0.7415,
+      "step": 3344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9425101896493326e-05,
+      "loss": 0.6655,
+      "step": 3345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.936811560162169e-05,
+      "loss": 0.6991,
+      "step": 3346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9311145106505664e-05,
+      "loss": 0.6832,
+      "step": 3347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.925419043329826e-05,
+      "loss": 0.6759,
+      "step": 3348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.919725160414644e-05,
+      "loss": 0.7449,
+      "step": 3349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.914032864119086e-05,
+      "loss": 0.6559,
+      "step": 3350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9083421566566096e-05,
+      "loss": 0.7184,
+      "step": 3351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.902653040240058e-05,
+      "loss": 0.7234,
+      "step": 3352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8969655170816455e-05,
+      "loss": 0.7442,
+      "step": 3353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.891279589392975e-05,
+      "loss": 0.5986,
+      "step": 3354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8855952593850214e-05,
+      "loss": 0.7183,
+      "step": 3355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.879912529268148e-05,
+      "loss": 0.6817,
+      "step": 3356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87423140125209e-05,
+      "loss": 0.7449,
+      "step": 3357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8685518775459614e-05,
+      "loss": 0.7022,
+      "step": 3358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.862873960358251e-05,
+      "loss": 0.7176,
+      "step": 3359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8571976518968185e-05,
+      "loss": 0.6862,
+      "step": 3360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.851522954368909e-05,
+      "loss": 0.7396,
+      "step": 3361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845849869981137e-05,
+      "loss": 0.6364,
+      "step": 3362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.840178400939481e-05,
+      "loss": 0.6448,
+      "step": 3363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.834508549449305e-05,
+      "loss": 0.7698,
+      "step": 3364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8288403177153306e-05,
+      "loss": 0.7788,
+      "step": 3365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8231737079416645e-05,
+      "loss": 0.733,
+      "step": 3366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.817508722331767e-05,
+      "loss": 0.6628,
+      "step": 3367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.811845363088477e-05,
+      "loss": 0.7191,
+      "step": 3368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8061836324140026e-05,
+      "loss": 0.6614,
+      "step": 3369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.800523532509907e-05,
+      "loss": 0.7485,
+      "step": 3370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.794865065577134e-05,
+      "loss": 0.7149,
+      "step": 3371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.789208233815977e-05,
+      "loss": 0.6769,
+      "step": 3372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7835530394261075e-05,
+      "loss": 0.8453,
+      "step": 3373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.777899484606555e-05,
+      "loss": 0.6833,
+      "step": 3374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7722475715557056e-05,
+      "loss": 0.7737,
+      "step": 3375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7665973024713195e-05,
+      "loss": 0.763,
+      "step": 3376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.760948679550502e-05,
+      "loss": 0.6669,
+      "step": 3377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755301704989735e-05,
+      "loss": 0.7623,
+      "step": 3378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.749656380984844e-05,
+      "loss": 0.7005,
+      "step": 3379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7440127097310214e-05,
+      "loss": 0.7456,
+      "step": 3380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.738370693422822e-05,
+      "loss": 0.718,
+      "step": 3381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7327303342541415e-05,
+      "loss": 0.6658,
+      "step": 3382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.727091634418244e-05,
+      "loss": 0.7629,
+      "step": 3383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.721454596107747e-05,
+      "loss": 0.7924,
+      "step": 3384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7158192215146154e-05,
+      "loss": 0.7847,
+      "step": 3385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.710185512830171e-05,
+      "loss": 0.7428,
+      "step": 3386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.70455347224509e-05,
+      "loss": 0.6837,
+      "step": 3387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6989231019494014e-05,
+      "loss": 0.6741,
+      "step": 3388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6932944041324744e-05,
+      "loss": 0.7724,
+      "step": 3389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.687667380983037e-05,
+      "loss": 0.7945,
+      "step": 3390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.682042034689168e-05,
+      "loss": 0.7279,
+      "step": 3391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.676418367438283e-05,
+      "loss": 0.863,
+      "step": 3392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.670796381417159e-05,
+      "loss": 0.7174,
+      "step": 3393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6651760788119044e-05,
+      "loss": 0.6353,
+      "step": 3394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6595574618079873e-05,
+      "loss": 0.8395,
+      "step": 3395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653940532590205e-05,
+      "loss": 0.7874,
+      "step": 3396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6483252933427134e-05,
+      "loss": 0.7693,
+      "step": 3397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.642711746249005e-05,
+      "loss": 0.6741,
+      "step": 3398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.637099893491911e-05,
+      "loss": 0.6985,
+      "step": 3399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.631489737253611e-05,
+      "loss": 0.7104,
+      "step": 3400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.625881279715615e-05,
+      "loss": 0.6856,
+      "step": 3401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6202745230587806e-05,
+      "loss": 0.6321,
+      "step": 3402
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6146694694633064e-05,
+      "loss": 0.5933,
+      "step": 3403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.609066121108717e-05,
+      "loss": 0.8123,
+      "step": 3404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603464480173884e-05,
+      "loss": 0.7394,
+      "step": 3405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.597864548837011e-05,
+      "loss": 0.6233,
+      "step": 3406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.592266329275642e-05,
+      "loss": 0.7459,
+      "step": 3407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.586669823666646e-05,
+      "loss": 0.6897,
+      "step": 3408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5810750341862315e-05,
+      "loss": 0.7923,
+      "step": 3409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.575481963009945e-05,
+      "loss": 0.6675,
+      "step": 3410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.569890612312649e-05,
+      "loss": 0.6378,
+      "step": 3411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5643009842685554e-05,
+      "loss": 0.7302,
+      "step": 3412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.558713081051192e-05,
+      "loss": 0.8081,
+      "step": 3413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.553126904833421e-05,
+      "loss": 0.7636,
+      "step": 3414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.547542457787441e-05,
+      "loss": 0.7084,
+      "step": 3415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5419597420847614e-05,
+      "loss": 0.7674,
+      "step": 3416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.536378759896237e-05,
+      "loss": 0.7798,
+      "step": 3417
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5307995133920286e-05,
+      "loss": 0.6913,
+      "step": 3418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.525222004741639e-05,
+      "loss": 0.7081,
+      "step": 3419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.519646236113891e-05,
+      "loss": 0.7076,
+      "step": 3420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.514072209676924e-05,
+      "loss": 0.7517,
+      "step": 3421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.508499927598209e-05,
+      "loss": 0.7356,
+      "step": 3422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.502929392044528e-05,
+      "loss": 0.74,
+      "step": 3423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497360605181998e-05,
+      "loss": 0.6576,
+      "step": 3424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.491793569176041e-05,
+      "loss": 0.7117,
+      "step": 3425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486228286191409e-05,
+      "loss": 0.7678,
+      "step": 3426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.480664758392169e-05,
+      "loss": 0.6785,
+      "step": 3427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.475102987941705e-05,
+      "loss": 0.6005,
+      "step": 3428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.469542977002722e-05,
+      "loss": 0.7075,
+      "step": 3429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.463984727737227e-05,
+      "loss": 0.7823,
+      "step": 3430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.458428242306559e-05,
+      "loss": 0.8194,
+      "step": 3431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.452873522871365e-05,
+      "loss": 0.794,
+      "step": 3432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.447320571591597e-05,
+      "loss": 0.8313,
+      "step": 3433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4417693906265365e-05,
+      "loss": 0.8908,
+      "step": 3434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.436219982134756e-05,
+      "loss": 0.6918,
+      "step": 3435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.430672348274155e-05,
+      "loss": 0.8451,
+      "step": 3436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.42512649120194e-05,
+      "loss": 0.7497,
+      "step": 3437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4195824130746176e-05,
+      "loss": 0.7094,
+      "step": 3438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.414040116048016e-05,
+      "loss": 0.658,
+      "step": 3439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.408499602277256e-05,
+      "loss": 0.6863,
+      "step": 3440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.402960873916781e-05,
+      "loss": 0.695,
+      "step": 3441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.397423933120325e-05,
+      "loss": 0.7048,
+      "step": 3442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.391888782040936e-05,
+      "loss": 0.6184,
+      "step": 3443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3863554228309685e-05,
+      "loss": 0.7395,
+      "step": 3444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.380823857642069e-05,
+      "loss": 0.7578,
+      "step": 3445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3752940886251956e-05,
+      "loss": 0.8091,
+      "step": 3446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36976611793061e-05,
+      "loss": 0.6877,
+      "step": 3447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3642399477078606e-05,
+      "loss": 0.7112,
+      "step": 3448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.358715580105813e-05,
+      "loss": 0.7511,
+      "step": 3449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3531930172726197e-05,
+      "loss": 0.7528,
+      "step": 3450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.347672261355742e-05,
+      "loss": 0.729,
+      "step": 3451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.342153314501923e-05,
+      "loss": 0.7205,
+      "step": 3452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3366361788572215e-05,
+      "loss": 0.8246,
+      "step": 3453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.331120856566971e-05,
+      "loss": 0.6135,
+      "step": 3454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3256073497758186e-05,
+      "loss": 0.6841,
+      "step": 3455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3200956606277006e-05,
+      "loss": 0.8058,
+      "step": 3456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.314585791265835e-05,
+      "loss": 0.6791,
+      "step": 3457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.309077743832749e-05,
+      "loss": 0.7262,
+      "step": 3458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.303571520470245e-05,
+      "loss": 0.7386,
+      "step": 3459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29806712331943e-05,
+      "loss": 0.7826,
+      "step": 3460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.292564554520698e-05,
+      "loss": 0.7224,
+      "step": 3461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2870638162137196e-05,
+      "loss": 0.6517,
+      "step": 3462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.281564910537475e-05,
+      "loss": 0.834,
+      "step": 3463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.27606783963021e-05,
+      "loss": 0.7442,
+      "step": 3464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2705726056294715e-05,
+      "loss": 0.8093,
+      "step": 3465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2650792106720905e-05,
+      "loss": 0.7871,
+      "step": 3466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.259587656894174e-05,
+      "loss": 0.6857,
+      "step": 3467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.254097946431121e-05,
+      "loss": 0.7049,
+      "step": 3468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.248610081417618e-05,
+      "loss": 0.6908,
+      "step": 3469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2431240639876176e-05,
+      "loss": 0.7173,
+      "step": 3470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.237639896274369e-05,
+      "loss": 0.672,
+      "step": 3471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.232157580410397e-05,
+      "loss": 0.7951,
+      "step": 3472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226677118527512e-05,
+      "loss": 0.7369,
+      "step": 3473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2211985127567885e-05,
+      "loss": 0.7449,
+      "step": 3474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.215721765228596e-05,
+      "loss": 0.6539,
+      "step": 3475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.210246878072568e-05,
+      "loss": 0.6441,
+      "step": 3476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2047738534176236e-05,
+      "loss": 0.7327,
+      "step": 3477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.199302693391959e-05,
+      "loss": 0.6795,
+      "step": 3478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1938334001230336e-05,
+      "loss": 0.779,
+      "step": 3479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.188365975737595e-05,
+      "loss": 0.7206,
+      "step": 3480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.182900422361652e-05,
+      "loss": 0.7335,
+      "step": 3481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1774367421204975e-05,
+      "loss": 0.7518,
+      "step": 3482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171974937138682e-05,
+      "loss": 0.8075,
+      "step": 3483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16651500954004e-05,
+      "loss": 0.6546,
+      "step": 3484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.161056961447674e-05,
+      "loss": 0.8787,
+      "step": 3485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1556007949839435e-05,
+      "loss": 0.7072,
+      "step": 3486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.150146512270494e-05,
+      "loss": 0.779,
+      "step": 3487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1446941154282216e-05,
+      "loss": 0.7722,
+      "step": 3488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.139243606577302e-05,
+      "loss": 0.6584,
+      "step": 3489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1337949878371705e-05,
+      "loss": 0.8032,
+      "step": 3490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12834826132653e-05,
+      "loss": 0.7599,
+      "step": 3491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1229034291633505e-05,
+      "loss": 0.7535,
+      "step": 3492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.117460493464852e-05,
+      "loss": 0.7425,
+      "step": 3493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.112019456347531e-05,
+      "loss": 0.6379,
+      "step": 3494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1065803199271456e-05,
+      "loss": 0.7593,
+      "step": 3495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1011430863187024e-05,
+      "loss": 0.746,
+      "step": 3496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0957077576364834e-05,
+      "loss": 0.6895,
+      "step": 3497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090274335994014e-05,
+      "loss": 0.8189,
+      "step": 3498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.084842823504096e-05,
+      "loss": 0.6767,
+      "step": 3499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0794132222787707e-05,
+      "loss": 0.8013,
+      "step": 3500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0739855344293484e-05,
+      "loss": 0.7255,
+      "step": 3501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.068559762066396e-05,
+      "loss": 0.7943,
+      "step": 3502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.063135907299725e-05,
+      "loss": 0.7343,
+      "step": 3503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0577139722384135e-05,
+      "loss": 0.6927,
+      "step": 3504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.052293958990779e-05,
+      "loss": 0.7253,
+      "step": 3505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.046875869664407e-05,
+      "loss": 0.6766,
+      "step": 3506
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.04145970636613e-05,
+      "loss": 0.7173,
+      "step": 3507
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.03604547120202e-05,
+      "loss": 0.6331,
+      "step": 3508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0306331662774185e-05,
+      "loss": 0.727,
+      "step": 3509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0252227936968975e-05,
+      "loss": 0.7863,
+      "step": 3510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.019814355564292e-05,
+      "loss": 0.7745,
+      "step": 3511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.014407853982679e-05,
+      "loss": 0.6826,
+      "step": 3512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0090032910543814e-05,
+      "loss": 0.6485,
+      "step": 3513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.003600668880976e-05,
+      "loss": 0.7784,
+      "step": 3514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.998199989563269e-05,
+      "loss": 0.6729,
+      "step": 3515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.99280125520133e-05,
+      "loss": 0.7682,
+      "step": 3516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9874044678944555e-05,
+      "loss": 0.7288,
+      "step": 3517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9820096297411947e-05,
+      "loss": 0.6804,
+      "step": 3518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.976616742839342e-05,
+      "loss": 0.7718,
+      "step": 3519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.97122580928592e-05,
+      "loss": 0.7041,
+      "step": 3520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965836831177206e-05,
+      "loss": 0.7806,
+      "step": 3521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.960449810608705e-05,
+      "loss": 0.7768,
+      "step": 3522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9550647496751676e-05,
+      "loss": 0.7343,
+      "step": 3523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.949681650470587e-05,
+      "loss": 0.718,
+      "step": 3524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9443005150881784e-05,
+      "loss": 0.7324,
+      "step": 3525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9389213456204106e-05,
+      "loss": 0.7503,
+      "step": 3526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9335441441589725e-05,
+      "loss": 0.6604,
+      "step": 3527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.928168912794803e-05,
+      "loss": 0.7501,
+      "step": 3528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.922795653618059e-05,
+      "loss": 0.699,
+      "step": 3529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917424368718141e-05,
+      "loss": 0.7078,
+      "step": 3530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.912055060183687e-05,
+      "loss": 0.6473,
+      "step": 3531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9066877301025474e-05,
+      "loss": 0.7394,
+      "step": 3532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.90132238056182e-05,
+      "loss": 0.6568,
+      "step": 3533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.895959013647828e-05,
+      "loss": 0.6726,
+      "step": 3534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8905976314461236e-05,
+      "loss": 0.6529,
+      "step": 3535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8852382360414875e-05,
+      "loss": 0.7268,
+      "step": 3536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879880829517921e-05,
+      "loss": 0.6333,
+      "step": 3537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.874525413958665e-05,
+      "loss": 0.8594,
+      "step": 3538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.869171991446172e-05,
+      "loss": 0.5744,
+      "step": 3539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.86382056406213e-05,
+      "loss": 0.7179,
+      "step": 3540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.858471133887451e-05,
+      "loss": 0.6357,
+      "step": 3541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853123703002262e-05,
+      "loss": 0.6395,
+      "step": 3542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.847778273485921e-05,
+      "loss": 0.8265,
+      "step": 3543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8424348474170014e-05,
+      "loss": 0.7885,
+      "step": 3544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.837093426873306e-05,
+      "loss": 0.6864,
+      "step": 3545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.831754013931844e-05,
+      "loss": 0.6902,
+      "step": 3546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826416610668858e-05,
+      "loss": 0.7296,
+      "step": 3547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.821081219159806e-05,
+      "loss": 0.6462,
+      "step": 3548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.815747841479355e-05,
+      "loss": 0.6948,
+      "step": 3549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.810416479701402e-05,
+      "loss": 0.6195,
+      "step": 3550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8050871358990444e-05,
+      "loss": 0.6854,
+      "step": 3551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.79975981214461e-05,
+      "loss": 0.8146,
+      "step": 3552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.794434510509632e-05,
+      "loss": 0.6951,
+      "step": 3553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.789111233064867e-05,
+      "loss": 0.7256,
+      "step": 3554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.783789981880267e-05,
+      "loss": 0.7166,
+      "step": 3555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.778470759025012e-05,
+      "loss": 0.6404,
+      "step": 3556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773153566567489e-05,
+      "loss": 0.793,
+      "step": 3557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.767838406575297e-05,
+      "loss": 0.5872,
+      "step": 3558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.762525281115234e-05,
+      "loss": 0.7978,
+      "step": 3559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.757214192253324e-05,
+      "loss": 0.7745,
+      "step": 3560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7519051420547836e-05,
+      "loss": 0.7304,
+      "step": 3561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.746598132584047e-05,
+      "loss": 0.7485,
+      "step": 3562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.741293165904747e-05,
+      "loss": 0.7782,
+      "step": 3563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7359902440797266e-05,
+      "loss": 0.6738,
+      "step": 3564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7306893691710395e-05,
+      "loss": 0.7347,
+      "step": 3565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.725390543239929e-05,
+      "loss": 0.6704,
+      "step": 3566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.720093768346857e-05,
+      "loss": 0.6601,
+      "step": 3567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7147990465514735e-05,
+      "loss": 0.727,
+      "step": 3568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7095063799126405e-05,
+      "loss": 0.6576,
+      "step": 3569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7042157704884214e-05,
+      "loss": 0.799,
+      "step": 3570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.698927220336068e-05,
+      "loss": 0.7887,
+      "step": 3571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.693640731512047e-05,
+      "loss": 0.7871,
+      "step": 3572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.688356306072009e-05,
+      "loss": 0.5909,
+      "step": 3573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.683073946070812e-05,
+      "loss": 0.6757,
+      "step": 3574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677793653562507e-05,
+      "loss": 0.7294,
+      "step": 3575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.672515430600347e-05,
+      "loss": 0.8022,
+      "step": 3576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667239279236768e-05,
+      "loss": 0.6532,
+      "step": 3577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6619652015234097e-05,
+      "loss": 0.7501,
+      "step": 3578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6566931995111086e-05,
+      "loss": 0.7584,
+      "step": 3579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65142327524988e-05,
+      "loss": 0.7125,
+      "step": 3580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6461554307889446e-05,
+      "loss": 0.6585,
+      "step": 3581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6408896681767134e-05,
+      "loss": 0.6312,
+      "step": 3582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6356259894607776e-05,
+      "loss": 0.6456,
+      "step": 3583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.630364396687932e-05,
+      "loss": 0.6676,
+      "step": 3584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.625104891904144e-05,
+      "loss": 0.6667,
+      "step": 3585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619847477154584e-05,
+      "loss": 0.8712,
+      "step": 3586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6145921544836066e-05,
+      "loss": 0.7072,
+      "step": 3587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.609338925934743e-05,
+      "loss": 0.7554,
+      "step": 3588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6040877935507245e-05,
+      "loss": 0.667,
+      "step": 3589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.598838759373453e-05,
+      "loss": 0.6863,
+      "step": 3590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.593591825444028e-05,
+      "loss": 0.7077,
+      "step": 3591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5883469938027194e-05,
+      "loss": 0.8151,
+      "step": 3592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5831042664889886e-05,
+      "loss": 0.8042,
+      "step": 3593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.577863645541483e-05,
+      "loss": 0.7774,
+      "step": 3594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.572625132998013e-05,
+      "loss": 0.8088,
+      "step": 3595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567388730895585e-05,
+      "loss": 0.7127,
+      "step": 3596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.562154441270382e-05,
+      "loss": 0.758,
+      "step": 3597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5569222661577615e-05,
+      "loss": 0.7046,
+      "step": 3598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551692207592265e-05,
+      "loss": 0.7501,
+      "step": 3599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.546464267607601e-05,
+      "loss": 0.6413,
+      "step": 3600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5412384482366665e-05,
+      "loss": 0.758,
+      "step": 3601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536014751511518e-05,
+      "loss": 0.7012,
+      "step": 3602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530793179463407e-05,
+      "loss": 0.8436,
+      "step": 3603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.525573734122738e-05,
+      "loss": 0.7784,
+      "step": 3604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5203564175191035e-05,
+      "loss": 0.6785,
+      "step": 3605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.515141231681266e-05,
+      "loss": 0.6975,
+      "step": 3606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5099281786371495e-05,
+      "loss": 0.6871,
+      "step": 3607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504717260413862e-05,
+      "loss": 0.7697,
+      "step": 3608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.49950847903767e-05,
+      "loss": 0.6766,
+      "step": 3609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.494301836534016e-05,
+      "loss": 0.7534,
+      "step": 3610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.489097334927515e-05,
+      "loss": 0.7075,
+      "step": 3611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4838949762419325e-05,
+      "loss": 0.7071,
+      "step": 3612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.478694762500222e-05,
+      "loss": 0.712,
+      "step": 3613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.473496695724484e-05,
+      "loss": 0.8449,
+      "step": 3614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.468300777935997e-05,
+      "loss": 0.7073,
+      "step": 3615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463107011155204e-05,
+      "loss": 0.7208,
+      "step": 3616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.457915397401698e-05,
+      "loss": 0.6644,
+      "step": 3617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.452725938694251e-05,
+      "loss": 0.7492,
+      "step": 3618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4475386370507864e-05,
+      "loss": 0.7367,
+      "step": 3619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.442353494488398e-05,
+      "loss": 0.6331,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4371705130233275e-05,
+      "loss": 0.7179,
+      "step": 3621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.431989694670987e-05,
+      "loss": 0.7827,
+      "step": 3622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.426811041445946e-05,
+      "loss": 0.6532,
+      "step": 3623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.421634555361923e-05,
+      "loss": 0.7836,
+      "step": 3624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4164602384318085e-05,
+      "loss": 0.675,
+      "step": 3625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.411288092667636e-05,
+      "loss": 0.7784,
+      "step": 3626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4061181200806e-05,
+      "loss": 0.6847,
+      "step": 3627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4009503226810576e-05,
+      "loss": 0.7282,
+      "step": 3628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3957847024785034e-05,
+      "loss": 0.6724,
+      "step": 3629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3906212614816035e-05,
+      "loss": 0.8692,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385460001698159e-05,
+      "loss": 0.6749,
+      "step": 3631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.380300925135138e-05,
+      "loss": 0.7461,
+      "step": 3632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.375144033798655e-05,
+      "loss": 0.658,
+      "step": 3633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3699893296939664e-05,
+      "loss": 0.7008,
+      "step": 3634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.364836814825492e-05,
+      "loss": 0.6353,
+      "step": 3635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359686491196788e-05,
+      "loss": 0.746,
+      "step": 3636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.354538360810564e-05,
+      "loss": 0.6406,
+      "step": 3637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349392425668682e-05,
+      "loss": 0.7617,
+      "step": 3638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3442486877721376e-05,
+      "loss": 0.7918,
+      "step": 3639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.339107149121083e-05,
+      "loss": 0.7899,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333967811714811e-05,
+      "loss": 0.7242,
+      "step": 3641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.328830677551763e-05,
+      "loss": 0.7711,
+      "step": 3642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3236957486295115e-05,
+      "loss": 0.7437,
+      "step": 3643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318563026944784e-05,
+      "loss": 0.6856,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3134325144934486e-05,
+      "loss": 0.6608,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.308304213270504e-05,
+      "loss": 0.7082,
+      "step": 3646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.303178125270104e-05,
+      "loss": 0.7206,
+      "step": 3647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.298054252485525e-05,
+      "loss": 0.7694,
+      "step": 3648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292932596909199e-05,
+      "loss": 0.6174,
+      "step": 3649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.287813160532682e-05,
+      "loss": 0.7211,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282695945346674e-05,
+      "loss": 0.594,
+      "step": 3651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277580953341015e-05,
+      "loss": 0.7165,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272468186504668e-05,
+      "loss": 0.7397,
+      "step": 3653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.267357646825746e-05,
+      "loss": 0.6174,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2622493362914805e-05,
+      "loss": 0.642,
+      "step": 3655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2571432568882484e-05,
+      "loss": 0.5863,
+      "step": 3656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.252039410601557e-05,
+      "loss": 0.7105,
+      "step": 3657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.246937799416036e-05,
+      "loss": 0.6843,
+      "step": 3658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241838425315458e-05,
+      "loss": 0.7544,
+      "step": 3659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.236741290282719e-05,
+      "loss": 0.616,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.23164639629985e-05,
+      "loss": 0.6239,
+      "step": 3661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2265537453479985e-05,
+      "loss": 0.7396,
+      "step": 3662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2214633394074534e-05,
+      "loss": 0.7312,
+      "step": 3663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.216375180457626e-05,
+      "loss": 0.7105,
+      "step": 3664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.211289270477047e-05,
+      "loss": 0.8107,
+      "step": 3665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.206205611443386e-05,
+      "loss": 0.8536,
+      "step": 3666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.201124205333421e-05,
+      "loss": 0.6293,
+      "step": 3667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.196045054123068e-05,
+      "loss": 0.6879,
+      "step": 3668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.190968159787364e-05,
+      "loss": 0.7405,
+      "step": 3669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.185893524300457e-05,
+      "loss": 0.7346,
+      "step": 3670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.180821149635634e-05,
+      "loss": 0.6939,
+      "step": 3671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1757510377652853e-05,
+      "loss": 0.7665,
+      "step": 3672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.170683190660933e-05,
+      "loss": 0.7243,
+      "step": 3673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.16561761029322e-05,
+      "loss": 0.6772,
+      "step": 3674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.160554298631896e-05,
+      "loss": 0.7506,
+      "step": 3675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1554932576458415e-05,
+      "loss": 0.7101,
+      "step": 3676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1504344893030437e-05,
+      "loss": 0.7549,
+      "step": 3677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1453779955706104e-05,
+      "loss": 0.7387,
+      "step": 3678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.140323778414772e-05,
+      "loss": 0.7766,
+      "step": 3679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1352718398008575e-05,
+      "loss": 0.8583,
+      "step": 3680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130222181693323e-05,
+      "loss": 0.7707,
+      "step": 3681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.125174806055735e-05,
+      "loss": 0.6756,
+      "step": 3682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.120129714850773e-05,
+      "loss": 0.7455,
+      "step": 3683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1150869100402225e-05,
+      "loss": 0.8022,
+      "step": 3684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.110046393584985e-05,
+      "loss": 0.7183,
+      "step": 3685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.105008167445076e-05,
+      "loss": 0.7089,
+      "step": 3686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0999722335796075e-05,
+      "loss": 0.6671,
+      "step": 3687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0949385939468166e-05,
+      "loss": 0.6936,
+      "step": 3688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0899072505040325e-05,
+      "loss": 0.7631,
+      "step": 3689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0848782052077014e-05,
+      "loss": 0.6861,
+      "step": 3690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.079851460013378e-05,
+      "loss": 0.7002,
+      "step": 3691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074827016875711e-05,
+      "loss": 0.7129,
+      "step": 3692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069804877748468e-05,
+      "loss": 0.7278,
+      "step": 3693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0647850445845046e-05,
+      "loss": 0.8148,
+      "step": 3694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0597675193357995e-05,
+      "loss": 0.6951,
+      "step": 3695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054752303953413e-05,
+      "loss": 0.7406,
+      "step": 3696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.049739400387523e-05,
+      "loss": 0.6683,
+      "step": 3697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.044728810587406e-05,
+      "loss": 0.6715,
+      "step": 3698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039720536501428e-05,
+      "loss": 0.7337,
+      "step": 3699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034714580077066e-05,
+      "loss": 0.7996,
+      "step": 3700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.029710943260895e-05,
+      "loss": 0.7239,
+      "step": 3701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0247096279985784e-05,
+      "loss": 0.7475,
+      "step": 3702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019710636234888e-05,
+      "loss": 0.6951,
+      "step": 3703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.014713969913686e-05,
+      "loss": 0.7036,
+      "step": 3704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009719630977937e-05,
+      "loss": 0.6377,
+      "step": 3705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.004727621369687e-05,
+      "loss": 0.7549,
+      "step": 3706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999737943030088e-05,
+      "loss": 0.7486,
+      "step": 3707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994750597899388e-05,
+      "loss": 0.8043,
+      "step": 3708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989765587916914e-05,
+      "loss": 0.7495,
+      "step": 3709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984782915021098e-05,
+      "loss": 0.7462,
+      "step": 3710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979802581149453e-05,
+      "loss": 0.7377,
+      "step": 3711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.974824588238595e-05,
+      "loss": 0.7087,
+      "step": 3712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9698489382242147e-05,
+      "loss": 0.7731,
+      "step": 3713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.964875633041102e-05,
+      "loss": 0.6649,
+      "step": 3714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.959904674623137e-05,
+      "loss": 0.6971,
+      "step": 3715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9549360649032763e-05,
+      "loss": 0.697,
+      "step": 3716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9499698058135746e-05,
+      "loss": 0.6592,
+      "step": 3717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945005899285162e-05,
+      "loss": 0.6615,
+      "step": 3718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940044347248263e-05,
+      "loss": 0.814,
+      "step": 3719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935085151632185e-05,
+      "loss": 0.6793,
+      "step": 3720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9301283143653125e-05,
+      "loss": 0.8108,
+      "step": 3721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925173837375118e-05,
+      "loss": 0.7264,
+      "step": 3722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920221722588161e-05,
+      "loss": 0.6459,
+      "step": 3723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.915271971930069e-05,
+      "loss": 0.7545,
+      "step": 3724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9103245873255636e-05,
+      "loss": 0.714,
+      "step": 3725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9053795706984386e-05,
+      "loss": 0.7072,
+      "step": 3726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9004369239715756e-05,
+      "loss": 0.6495,
+      "step": 3727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895496649066919e-05,
+      "loss": 0.7379,
+      "step": 3728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8905587479055096e-05,
+      "loss": 0.7887,
+      "step": 3729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.885623222407447e-05,
+      "loss": 0.6974,
+      "step": 3730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8806900744919205e-05,
+      "loss": 0.599,
+      "step": 3731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8757593060771955e-05,
+      "loss": 0.7183,
+      "step": 3732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.870830919080598e-05,
+      "loss": 0.7723,
+      "step": 3733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8659049154185446e-05,
+      "loss": 0.6673,
+      "step": 3734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8609812970065096e-05,
+      "loss": 0.6446,
+      "step": 3735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.856060065759053e-05,
+      "loss": 0.694,
+      "step": 3736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851141223589805e-05,
+      "loss": 0.7726,
+      "step": 3737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846224772411454e-05,
+      "loss": 0.8126,
+      "step": 3738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.841310714135776e-05,
+      "loss": 0.7747,
+      "step": 3739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836399050673602e-05,
+      "loss": 0.7229,
+      "step": 3740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.831489783934843e-05,
+      "loss": 0.8524,
+      "step": 3741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.826582915828468e-05,
+      "loss": 0.6884,
+      "step": 3742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.821678448262521e-05,
+      "loss": 0.7327,
+      "step": 3743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8167763831441106e-05,
+      "loss": 0.6581,
+      "step": 3744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81187672237941e-05,
+      "loss": 0.7532,
+      "step": 3745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806979467873662e-05,
+      "loss": 0.7722,
+      "step": 3746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8020846215311614e-05,
+      "loss": 0.7097,
+      "step": 3747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79719218525528e-05,
+      "loss": 0.7148,
+      "step": 3748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79230216094845e-05,
+      "loss": 0.714,
+      "step": 3749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.787414550512156e-05,
+      "loss": 0.7485,
+      "step": 3750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.782529355846959e-05,
+      "loss": 0.6609,
+      "step": 3751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.777646578852464e-05,
+      "loss": 0.6457,
+      "step": 3752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7727662214273495e-05,
+      "loss": 0.6833,
+      "step": 3753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7678882854693507e-05,
+      "loss": 0.6775,
+      "step": 3754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7630127728752526e-05,
+      "loss": 0.7787,
+      "step": 3755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.75813968554091e-05,
+      "loss": 0.7562,
+      "step": 3756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.753269025361222e-05,
+      "loss": 0.7134,
+      "step": 3757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.748400794230158e-05,
+      "loss": 0.7131,
+      "step": 3758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.743534994040727e-05,
+      "loss": 0.7132,
+      "step": 3759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.738671626685006e-05,
+      "loss": 0.7747,
+      "step": 3760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.733810694054124e-05,
+      "loss": 0.7626,
+      "step": 3761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.728952198038254e-05,
+      "loss": 0.7044,
+      "step": 3762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.724096140526633e-05,
+      "loss": 0.7822,
+      "step": 3763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.719242523407539e-05,
+      "loss": 0.5726,
+      "step": 3764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.71439134856831e-05,
+      "loss": 0.6872,
+      "step": 3765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7095426178953317e-05,
+      "loss": 0.7039,
+      "step": 3766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.704696333274037e-05,
+      "loss": 0.7138,
+      "step": 3767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.699852496588914e-05,
+      "loss": 0.613,
+      "step": 3768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.695011109723486e-05,
+      "loss": 0.6855,
+      "step": 3769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.690172174560341e-05,
+      "loss": 0.77,
+      "step": 3770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6853356929810965e-05,
+      "loss": 0.6433,
+      "step": 3771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.680501666866428e-05,
+      "loss": 0.6632,
+      "step": 3772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6756700980960545e-05,
+      "loss": 0.6713,
+      "step": 3773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.67084098854873e-05,
+      "loss": 0.6892,
+      "step": 3774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.666014340102268e-05,
+      "loss": 0.745,
+      "step": 3775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.661190154633509e-05,
+      "loss": 0.7695,
+      "step": 3776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6563684340183454e-05,
+      "loss": 0.867,
+      "step": 3777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.651549180131714e-05,
+      "loss": 0.6303,
+      "step": 3778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.646732394847579e-05,
+      "loss": 0.6715,
+      "step": 3779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6419180800389605e-05,
+      "loss": 0.6044,
+      "step": 3780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.637106237577903e-05,
+      "loss": 0.658,
+      "step": 3781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.632296869335501e-05,
+      "loss": 0.7645,
+      "step": 3782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.627489977181887e-05,
+      "loss": 0.8172,
+      "step": 3783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.622685562986219e-05,
+      "loss": 0.5802,
+      "step": 3784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.617883628616705e-05,
+      "loss": 0.7193,
+      "step": 3785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.613084175940578e-05,
+      "loss": 0.8338,
+      "step": 3786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6082872068241124e-05,
+      "loss": 0.6922,
+      "step": 3787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.603492723132615e-05,
+      "loss": 0.6228,
+      "step": 3788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.59870072673043e-05,
+      "loss": 0.6851,
+      "step": 3789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5939112194809286e-05,
+      "loss": 0.6515,
+      "step": 3790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.589124203246515e-05,
+      "loss": 0.7115,
+      "step": 3791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.584339679888627e-05,
+      "loss": 0.6022,
+      "step": 3792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.579557651267731e-05,
+      "loss": 0.6531,
+      "step": 3793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.574778119243325e-05,
+      "loss": 0.6783,
+      "step": 3794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5700010856739383e-05,
+      "loss": 0.7956,
+      "step": 3795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.56522655241712e-05,
+      "loss": 0.7082,
+      "step": 3796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5604545213294616e-05,
+      "loss": 0.6697,
+      "step": 3797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.555684994266563e-05,
+      "loss": 0.7474,
+      "step": 3798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.550917973083069e-05,
+      "loss": 0.6481,
+      "step": 3799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546153459632634e-05,
+      "loss": 0.6941,
+      "step": 3800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.541391455767949e-05,
+      "loss": 0.7902,
+      "step": 3801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.536631963340725e-05,
+      "loss": 0.626,
+      "step": 3802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.531874984201694e-05,
+      "loss": 0.7168,
+      "step": 3803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5271205202006166e-05,
+      "loss": 0.7827,
+      "step": 3804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5223685731862655e-05,
+      "loss": 0.6924,
+      "step": 3805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5176191450064444e-05,
+      "loss": 0.6643,
+      "step": 3806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5128722375079736e-05,
+      "loss": 0.7521,
+      "step": 3807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.508127852536698e-05,
+      "loss": 0.7031,
+      "step": 3808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.503385991937469e-05,
+      "loss": 0.725,
+      "step": 3809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4986466575541684e-05,
+      "loss": 0.697,
+      "step": 3810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.493909851229693e-05,
+      "loss": 0.7336,
+      "step": 3811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.489175574805959e-05,
+      "loss": 0.689,
+      "step": 3812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.484443830123887e-05,
+      "loss": 0.7071,
+      "step": 3813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479714619023429e-05,
+      "loss": 0.7285,
+      "step": 3814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4749879433435365e-05,
+      "loss": 0.7111,
+      "step": 3815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4702638049221904e-05,
+      "loss": 0.8085,
+      "step": 3816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.465542205596372e-05,
+      "loss": 0.8067,
+      "step": 3817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4608231472020816e-05,
+      "loss": 0.8381,
+      "step": 3818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.456106631574336e-05,
+      "loss": 0.6917,
+      "step": 3819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.45139266054715e-05,
+      "loss": 0.6896,
+      "step": 3820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4466812359535636e-05,
+      "loss": 0.7211,
+      "step": 3821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4419723596256126e-05,
+      "loss": 0.6919,
+      "step": 3822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.437266033394354e-05,
+      "loss": 0.6967,
+      "step": 3823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.43256225908985e-05,
+      "loss": 0.6247,
+      "step": 3824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4278610385411614e-05,
+      "loss": 0.7186,
+      "step": 3825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.423162373576372e-05,
+      "loss": 0.6576,
+      "step": 3826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4184662660225564e-05,
+      "loss": 0.81,
+      "step": 3827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.413772717705802e-05,
+      "loss": 0.5543,
+      "step": 3828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4090817304512047e-05,
+      "loss": 0.6167,
+      "step": 3829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4043933060828605e-05,
+      "loss": 0.6916,
+      "step": 3830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.399707446423862e-05,
+      "loss": 0.8214,
+      "step": 3831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.395024153296317e-05,
+      "loss": 0.6865,
+      "step": 3832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.390343428521331e-05,
+      "loss": 0.8389,
+      "step": 3833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.385665273919003e-05,
+      "loss": 0.7062,
+      "step": 3834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.380989691308441e-05,
+      "loss": 0.7231,
+      "step": 3835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.376316682507757e-05,
+      "loss": 0.7484,
+      "step": 3836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3716462493340473e-05,
+      "loss": 0.7363,
+      "step": 3837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.366978393603422e-05,
+      "loss": 0.7481,
+      "step": 3838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.362313117130975e-05,
+      "loss": 0.7667,
+      "step": 3839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.357650421730809e-05,
+      "loss": 0.6229,
+      "step": 3840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.352990309216022e-05,
+      "loss": 0.6567,
+      "step": 3841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3483327813986965e-05,
+      "loss": 0.7871,
+      "step": 3842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3436778400899235e-05,
+      "loss": 0.6158,
+      "step": 3843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.339025487099779e-05,
+      "loss": 0.6426,
+      "step": 3844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3343757242373396e-05,
+      "loss": 0.7996,
+      "step": 3845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.329728553310665e-05,
+      "loss": 0.6695,
+      "step": 3846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.325083976126817e-05,
+      "loss": 0.7255,
+      "step": 3847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3204419944918476e-05,
+      "loss": 0.6795,
+      "step": 3848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3158026102107905e-05,
+      "loss": 0.7168,
+      "step": 3849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.311165825087679e-05,
+      "loss": 0.7228,
+      "step": 3850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.306531640925532e-05,
+      "loss": 0.7001,
+      "step": 3851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3019000595263574e-05,
+      "loss": 0.7757,
+      "step": 3852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.297271082691155e-05,
+      "loss": 0.6248,
+      "step": 3853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2926447122199e-05,
+      "loss": 0.6796,
+      "step": 3854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.288020949911569e-05,
+      "loss": 0.7473,
+      "step": 3855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.283399797564112e-05,
+      "loss": 0.7204,
+      "step": 3856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2787812569744715e-05,
+      "loss": 0.7181,
+      "step": 3857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2741653299385764e-05,
+      "loss": 0.8134,
+      "step": 3858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2695520182513286e-05,
+      "loss": 0.7409,
+      "step": 3859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.264941323706625e-05,
+      "loss": 0.6727,
+      "step": 3860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.260333248097335e-05,
+      "loss": 0.8332,
+      "step": 3861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.25572779321532e-05,
+      "loss": 0.7378,
+      "step": 3862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.251124960851408e-05,
+      "loss": 0.7305,
+      "step": 3863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2465247527954224e-05,
+      "loss": 0.7651,
+      "step": 3864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.241927170836161e-05,
+      "loss": 0.6843,
+      "step": 3865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.237332216761392e-05,
+      "loss": 0.7011,
+      "step": 3866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.232739892357876e-05,
+      "loss": 0.8754,
+      "step": 3867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.228150199411336e-05,
+      "loss": 0.7241,
+      "step": 3868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.223563139706484e-05,
+      "loss": 0.6717,
+      "step": 3869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2189787150270054e-05,
+      "loss": 0.6553,
+      "step": 3870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.214396927155555e-05,
+      "loss": 0.8054,
+      "step": 3871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.209817777873767e-05,
+      "loss": 0.7386,
+      "step": 3872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2052412689622504e-05,
+      "loss": 0.741,
+      "step": 3873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.200667402200586e-05,
+      "loss": 0.657,
+      "step": 3874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.196096179367328e-05,
+      "loss": 0.6415,
+      "step": 3875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.19152760224e-05,
+      "loss": 0.7446,
+      "step": 3876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.186961672595101e-05,
+      "loss": 0.6517,
+      "step": 3877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1823983922080926e-05,
+      "loss": 0.6899,
+      "step": 3878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1778377628534196e-05,
+      "loss": 0.7262,
+      "step": 3879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.173279786304479e-05,
+      "loss": 0.7585,
+      "step": 3880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.168724464333651e-05,
+      "loss": 0.7352,
+      "step": 3881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.16417179871228e-05,
+      "loss": 0.874,
+      "step": 3882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.159621791210669e-05,
+      "loss": 0.7108,
+      "step": 3883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1550744435980993e-05,
+      "loss": 0.7201,
+      "step": 3884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1505297576428075e-05,
+      "loss": 0.8008,
+      "step": 3885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.145987735112003e-05,
+      "loss": 0.7762,
+      "step": 3886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141448377771858e-05,
+      "loss": 0.7291,
+      "step": 3887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.136911687387502e-05,
+      "loss": 0.7723,
+      "step": 3888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.132377665723039e-05,
+      "loss": 0.6399,
+      "step": 3889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.12784631454152e-05,
+      "loss": 0.736,
+      "step": 3890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.123317635604973e-05,
+      "loss": 0.6134,
+      "step": 3891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1187916306743784e-05,
+      "loss": 0.6753,
+      "step": 3892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.11426830150968e-05,
+      "loss": 0.7071,
+      "step": 3893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.109747649869774e-05,
+      "loss": 0.6325,
+      "step": 3894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105229677512527e-05,
+      "loss": 0.6976,
+      "step": 3895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.100714386194757e-05,
+      "loss": 0.6815,
+      "step": 3896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.096201777672236e-05,
+      "loss": 0.6948,
+      "step": 3897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091691853699702e-05,
+      "loss": 0.7854,
+      "step": 3898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0871846160308436e-05,
+      "loss": 0.7216,
+      "step": 3899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.082680066418301e-05,
+      "loss": 0.6823,
+      "step": 3900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.078178206613681e-05,
+      "loss": 0.8322,
+      "step": 3901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.073679038367528e-05,
+      "loss": 0.7344,
+      "step": 3902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0691825634293546e-05,
+      "loss": 0.6846,
+      "step": 3903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0646887835476224e-05,
+      "loss": 0.7527,
+      "step": 3904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.060197700469736e-05,
+      "loss": 0.7601,
+      "step": 3905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055709315942066e-05,
+      "loss": 0.7076,
+      "step": 3906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0512236317099175e-05,
+      "loss": 0.7323,
+      "step": 3907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0467406495175633e-05,
+      "loss": 0.6228,
+      "step": 3908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0422603711082076e-05,
+      "loss": 0.7467,
+      "step": 3909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0377827982240147e-05,
+      "loss": 0.6148,
+      "step": 3910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0333079326060998e-05,
+      "loss": 0.696,
+      "step": 3911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0288357759945118e-05,
+      "loss": 0.7278,
+      "step": 3912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0243663301282564e-05,
+      "loss": 0.8234,
+      "step": 3913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0198995967452836e-05,
+      "loss": 0.8412,
+      "step": 3914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0154355775824918e-05,
+      "loss": 0.776,
+      "step": 3915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.010974274375712e-05,
+      "loss": 0.6444,
+      "step": 3916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0065156888597312e-05,
+      "loss": 0.8175,
+      "step": 3917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0020598227682795e-05,
+      "loss": 0.7957,
+      "step": 3918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9976066778340184e-05,
+      "loss": 0.7301,
+      "step": 3919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.993156255788565e-05,
+      "loss": 0.7897,
+      "step": 3920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9887085583624653e-05,
+      "loss": 0.709,
+      "step": 3921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.984263587285213e-05,
+      "loss": 0.7444,
+      "step": 3922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.979821344285246e-05,
+      "loss": 0.744,
+      "step": 3923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.975381831089927e-05,
+      "loss": 0.7867,
+      "step": 3924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9709450494255743e-05,
+      "loss": 0.6889,
+      "step": 3925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.966511001017428e-05,
+      "loss": 0.7167,
+      "step": 3926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9620796875896762e-05,
+      "loss": 0.6737,
+      "step": 3927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9576511108654436e-05,
+      "loss": 0.6596,
+      "step": 3928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.953225272566782e-05,
+      "loss": 0.6897,
+      "step": 3929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9488021744146878e-05,
+      "loss": 0.7177,
+      "step": 3930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9443818181290838e-05,
+      "loss": 0.7035,
+      "step": 3931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9399642054288334e-05,
+      "loss": 0.8673,
+      "step": 3932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9355493380317334e-05,
+      "loss": 0.7768,
+      "step": 3933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9311372176545026e-05,
+      "loss": 0.7687,
+      "step": 3934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9267278460128046e-05,
+      "loss": 0.6008,
+      "step": 3935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9223212248212273e-05,
+      "loss": 0.7792,
+      "step": 3936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9179173557932938e-05,
+      "loss": 0.7707,
+      "step": 3937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9135162406414473e-05,
+      "loss": 0.699,
+      "step": 3938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.909117881077069e-05,
+      "loss": 0.7129,
+      "step": 3939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.904722278810471e-05,
+      "loss": 0.7515,
+      "step": 3940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9003294355508816e-05,
+      "loss": 0.6934,
+      "step": 3941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8959393530064694e-05,
+      "loss": 0.8133,
+      "step": 3942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.891552032884315e-05,
+      "loss": 0.7227,
+      "step": 3943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.887167476890438e-05,
+      "loss": 0.6227,
+      "step": 3944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.882785686729781e-05,
+      "loss": 0.6564,
+      "step": 3945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8784066641062014e-05,
+      "loss": 0.6514,
+      "step": 3946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8740304107224936e-05,
+      "loss": 0.7763,
+      "step": 3947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.869656928280362e-05,
+      "loss": 0.7164,
+      "step": 3948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8652862184804453e-05,
+      "loss": 0.6594,
+      "step": 3949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8609182830223014e-05,
+      "loss": 0.6487,
+      "step": 3950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8565531236043997e-05,
+      "loss": 0.7779,
+      "step": 3951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8521907419241445e-05,
+      "loss": 0.686,
+      "step": 3952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8478311396778477e-05,
+      "loss": 0.7005,
+      "step": 3953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8434743185607482e-05,
+      "loss": 0.7715,
+      "step": 3954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8391202802670026e-05,
+      "loss": 0.6219,
+      "step": 3955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8347690264896797e-05,
+      "loss": 0.7582,
+      "step": 3956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8304205589207722e-05,
+      "loss": 0.7576,
+      "step": 3957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.826074879251186e-05,
+      "loss": 0.6153,
+      "step": 3958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.821731989170747e-05,
+      "loss": 0.6478,
+      "step": 3959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8173918903681874e-05,
+      "loss": 0.6972,
+      "step": 3960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8130545845311628e-05,
+      "loss": 0.8236,
+      "step": 3961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8087200733462425e-05,
+      "loss": 0.7793,
+      "step": 3962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8043883584988995e-05,
+      "loss": 0.7818,
+      "step": 3963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8000594416735326e-05,
+      "loss": 0.7052,
+      "step": 3964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7957333245534412e-05,
+      "loss": 0.7716,
+      "step": 3965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.791410008820845e-05,
+      "loss": 0.6765,
+      "step": 3966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7870894961568662e-05,
+      "loss": 0.6647,
+      "step": 3967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.782771788241543e-05,
+      "loss": 0.7362,
+      "step": 3968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.778456886753824e-05,
+      "loss": 0.7522,
+      "step": 3969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7741447933715592e-05,
+      "loss": 0.7871,
+      "step": 3970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7698355097715167e-05,
+      "loss": 0.7106,
+      "step": 3971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7655290376293586e-05,
+      "loss": 0.78,
+      "step": 3972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7612253786196664e-05,
+      "loss": 0.8234,
+      "step": 3973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7569245344159254e-05,
+      "loss": 0.7966,
+      "step": 3974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7526265066905176e-05,
+      "loss": 0.657,
+      "step": 3975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7483312971147402e-05,
+      "loss": 0.7432,
+      "step": 3976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7440389073587924e-05,
+      "loss": 0.7719,
+      "step": 3977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7397493390917693e-05,
+      "loss": 0.78,
+      "step": 3978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.735462593981677e-05,
+      "loss": 0.693,
+      "step": 3979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7311786736954226e-05,
+      "loss": 0.6873,
+      "step": 3980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.726897579898815e-05,
+      "loss": 0.729,
+      "step": 3981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.722619314256557e-05,
+      "loss": 0.6968,
+      "step": 3982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7183438784322645e-05,
+      "loss": 0.814,
+      "step": 3983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7140712740884376e-05,
+      "loss": 0.7693,
+      "step": 3984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7098015028864875e-05,
+      "loss": 0.7194,
+      "step": 3985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7055345664867228e-05,
+      "loss": 0.7102,
+      "step": 3986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7012704665483413e-05,
+      "loss": 0.7125,
+      "step": 3987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.697009204729447e-05,
+      "loss": 0.6892,
+      "step": 3988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6927507826870336e-05,
+      "loss": 0.7798,
+      "step": 3989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.688495202076994e-05,
+      "loss": 0.7007,
+      "step": 3990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.68424246455412e-05,
+      "loss": 0.7312,
+      "step": 3991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6799925717720865e-05,
+      "loss": 0.6883,
+      "step": 3992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6757455253834764e-05,
+      "loss": 0.6456,
+      "step": 3993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6715013270397514e-05,
+      "loss": 0.7453,
+      "step": 3994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.667259978391281e-05,
+      "loss": 0.7471,
+      "step": 3995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.66302148108731e-05,
+      "loss": 0.7491,
+      "step": 3996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6587858367759878e-05,
+      "loss": 0.6928,
+      "step": 3997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.654553047104349e-05,
+      "loss": 0.6809,
+      "step": 3998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.650323113718319e-05,
+      "loss": 0.7152,
+      "step": 3999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6460960382627154e-05,
+      "loss": 0.7256,
+      "step": 4000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6418718223812357e-05,
+      "loss": 0.6814,
+      "step": 4001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6376504677164726e-05,
+      "loss": 0.8163,
+      "step": 4002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.633431975909909e-05,
+      "loss": 0.6768,
+      "step": 4003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.629216348601905e-05,
+      "loss": 0.6527,
+      "step": 4004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.625003587431717e-05,
+      "loss": 0.8385,
+      "step": 4005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6207936940374767e-05,
+      "loss": 0.6605,
+      "step": 4006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6165866700562082e-05,
+      "loss": 0.8062,
+      "step": 4007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6123825171238214e-05,
+      "loss": 0.6747,
+      "step": 4008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6081812368751014e-05,
+      "loss": 0.6935,
+      "step": 4009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.603982830943724e-05,
+      "loss": 0.7397,
+      "step": 4010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.599787300962241e-05,
+      "loss": 0.697,
+      "step": 4011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.595594648562093e-05,
+      "loss": 0.6784,
+      "step": 4012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.591404875373593e-05,
+      "loss": 0.7067,
+      "step": 4013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5872179830259413e-05,
+      "loss": 0.752,
+      "step": 4014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5830339731472207e-05,
+      "loss": 0.7005,
+      "step": 4015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5788528473643804e-05,
+      "loss": 0.6749,
+      "step": 4016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5746746073032625e-05,
+      "loss": 0.7505,
+      "step": 4017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.570499254588575e-05,
+      "loss": 0.7294,
+      "step": 4018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5663267908439114e-05,
+      "loss": 0.7991,
+      "step": 4019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5621572176917385e-05,
+      "loss": 0.645,
+      "step": 4020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.557990536753401e-05,
+      "loss": 0.8104,
+      "step": 4021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5538267496491197e-05,
+      "loss": 0.7353,
+      "step": 4022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5496658579979815e-05,
+      "loss": 0.7137,
+      "step": 4023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5455078634179584e-05,
+      "loss": 0.7476,
+      "step": 4024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5413527675258942e-05,
+      "loss": 0.7173,
+      "step": 4025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.537200571937496e-05,
+      "loss": 0.7056,
+      "step": 4026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5330512782673555e-05,
+      "loss": 0.6672,
+      "step": 4027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5289048881289256e-05,
+      "loss": 0.7676,
+      "step": 4028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5247614031345413e-05,
+      "loss": 0.6821,
+      "step": 4029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.520620824895393e-05,
+      "loss": 0.6705,
+      "step": 4030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.516483155021555e-05,
+      "loss": 0.8073,
+      "step": 4031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.512348395121967e-05,
+      "loss": 0.6321,
+      "step": 4032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5082165468044304e-05,
+      "loss": 0.6747,
+      "step": 4033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5040876116756228e-05,
+      "loss": 0.5991,
+      "step": 4034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4999615913410824e-05,
+      "loss": 0.8348,
+      "step": 4035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4958384874052164e-05,
+      "loss": 0.6937,
+      "step": 4036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4917183014713053e-05,
+      "loss": 0.7163,
+      "step": 4037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.487601035141479e-05,
+      "loss": 0.6715,
+      "step": 4038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4834866900167475e-05,
+      "loss": 0.7086,
+      "step": 4039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4793752676969794e-05,
+      "loss": 0.732,
+      "step": 4040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4752667697809006e-05,
+      "loss": 0.6765,
+      "step": 4041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4711611978661077e-05,
+      "loss": 0.6606,
+      "step": 4042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4670585535490587e-05,
+      "loss": 0.7106,
+      "step": 4043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4629588384250736e-05,
+      "loss": 0.8036,
+      "step": 4044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4588620540883255e-05,
+      "loss": 0.6468,
+      "step": 4045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4547682021318596e-05,
+      "loss": 0.7788,
+      "step": 4046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.450677284147569e-05,
+      "loss": 0.7923,
+      "step": 4047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4465893017262155e-05,
+      "loss": 0.6513,
+      "step": 4048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4425042564574184e-05,
+      "loss": 0.7243,
+      "step": 4049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4384221499296466e-05,
+      "loss": 0.6576,
+      "step": 4050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4343429837302366e-05,
+      "loss": 0.655,
+      "step": 4051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4302667594453733e-05,
+      "loss": 0.8009,
+      "step": 4052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4261934786601024e-05,
+      "loss": 0.6559,
+      "step": 4053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.422123142958328e-05,
+      "loss": 0.6755,
+      "step": 4054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4180557539227987e-05,
+      "loss": 0.827,
+      "step": 4055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4139913131351288e-05,
+      "loss": 0.8568,
+      "step": 4056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4099298221757748e-05,
+      "loss": 0.5961,
+      "step": 4057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4058712826240592e-05,
+      "loss": 0.6405,
+      "step": 4058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4018156960581427e-05,
+      "loss": 0.6926,
+      "step": 4059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3977630640550497e-05,
+      "loss": 0.7452,
+      "step": 4060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.393713388190648e-05,
+      "loss": 0.6603,
+      "step": 4061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.389666670039664e-05,
+      "loss": 0.6155,
+      "step": 4062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3856229111756622e-05,
+      "loss": 0.6896,
+      "step": 4063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.381582113171066e-05,
+      "loss": 0.6946,
+      "step": 4064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3775442775971444e-05,
+      "loss": 0.6558,
+      "step": 4065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.373509406024017e-05,
+      "loss": 0.6141,
+      "step": 4066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3694775000206426e-05,
+      "loss": 0.7311,
+      "step": 4067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3654485611548394e-05,
+      "loss": 0.6715,
+      "step": 4068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3614225909932574e-05,
+      "loss": 0.6899,
+      "step": 4069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3573995911014036e-05,
+      "loss": 0.7068,
+      "step": 4070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3533795630436293e-05,
+      "loss": 0.7726,
+      "step": 4071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3493625083831217e-05,
+      "loss": 0.7769,
+      "step": 4072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3453484286819215e-05,
+      "loss": 0.7825,
+      "step": 4073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.341337325500904e-05,
+      "loss": 0.7143,
+      "step": 4074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.337329200399796e-05,
+      "loss": 0.6752,
+      "step": 4075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.333324054937157e-05,
+      "loss": 0.7364,
+      "step": 4076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.329321890670394e-05,
+      "loss": 0.7033,
+      "step": 4077
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3253227091557574e-05,
+      "loss": 0.7124,
+      "step": 4078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.321326511948326e-05,
+      "loss": 0.7729,
+      "step": 4079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.317333300602034e-05,
+      "loss": 0.7448,
+      "step": 4080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3133430766696385e-05,
+      "loss": 0.7572,
+      "step": 4081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3093558417027473e-05,
+      "loss": 0.6593,
+      "step": 4082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3053715972518e-05,
+      "loss": 0.6424,
+      "step": 4083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3013903448660746e-05,
+      "loss": 0.7885,
+      "step": 4084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2974120860936887e-05,
+      "loss": 0.6387,
+      "step": 4085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2934368224815894e-05,
+      "loss": 0.6284,
+      "step": 4086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2894645555755646e-05,
+      "loss": 0.7221,
+      "step": 4087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2854952869202316e-05,
+      "loss": 0.5993,
+      "step": 4088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2815290180590488e-05,
+      "loss": 0.8169,
+      "step": 4089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.277565750534305e-05,
+      "loss": 0.7412,
+      "step": 4090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.273605485887116e-05,
+      "loss": 0.6999,
+      "step": 4091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2696482256574424e-05,
+      "loss": 0.7823,
+      "step": 4092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.265693971384063e-05,
+      "loss": 0.84,
+      "step": 4093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2617427246045973e-05,
+      "loss": 0.6675,
+      "step": 4094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2577944868554946e-05,
+      "loss": 0.678,
+      "step": 4095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2538492596720274e-05,
+      "loss": 0.6839,
+      "step": 4096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2499070445883054e-05,
+      "loss": 0.6824,
+      "step": 4097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.245967843137259e-05,
+      "loss": 0.7533,
+      "step": 4098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.242031656850654e-05,
+      "loss": 0.6162,
+      "step": 4099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.238098487259084e-05,
+      "loss": 0.7732,
+      "step": 4100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2341683358919607e-05,
+      "loss": 0.7339,
+      "step": 4101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2302412042775345e-05,
+      "loss": 0.72,
+      "step": 4102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.226317093942868e-05,
+      "loss": 0.7568,
+      "step": 4103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2223960064138605e-05,
+      "loss": 0.7702,
+      "step": 4104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.218477943215229e-05,
+      "loss": 0.8118,
+      "step": 4105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2145629058705207e-05,
+      "loss": 0.6868,
+      "step": 4106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.210650895902102e-05,
+      "loss": 0.7915,
+      "step": 4107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2067419148311586e-05,
+      "loss": 0.7654,
+      "step": 4108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2028359641777062e-05,
+      "loss": 0.7417,
+      "step": 4109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.198933045460575e-05,
+      "loss": 0.6194,
+      "step": 4110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1950331601974207e-05,
+      "loss": 0.5914,
+      "step": 4111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1911363099047222e-05,
+      "loss": 0.6871,
+      "step": 4112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1872424960977666e-05,
+      "loss": 0.7877,
+      "step": 4113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1833517202906763e-05,
+      "loss": 0.6758,
+      "step": 4114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.179463983996376e-05,
+      "loss": 0.589,
+      "step": 4115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1755792887266234e-05,
+      "loss": 0.7132,
+      "step": 4116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1716976359919817e-05,
+      "loss": 0.7632,
+      "step": 4117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.167819027301837e-05,
+      "loss": 0.596,
+      "step": 4118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1639434641643952e-05,
+      "loss": 0.7654,
+      "step": 4119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1600709480866665e-05,
+      "loss": 0.7057,
+      "step": 4120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1562014805744912e-05,
+      "loss": 0.678,
+      "step": 4121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.152335063132508e-05,
+      "loss": 0.6205,
+      "step": 4122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1484716972641826e-05,
+      "loss": 0.7595,
+      "step": 4123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1446113844717918e-05,
+      "loss": 0.7618,
+      "step": 4124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1407541262564157e-05,
+      "loss": 0.7319,
+      "step": 4125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1368999241179576e-05,
+      "loss": 0.7208,
+      "step": 4126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.133048779555129e-05,
+      "loss": 0.7588,
+      "step": 4127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1292006940654518e-05,
+      "loss": 0.6855,
+      "step": 4128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1253556691452604e-05,
+      "loss": 0.7481,
+      "step": 4129
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1215137062896917e-05,
+      "loss": 0.7122,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1176748069927033e-05,
+      "loss": 0.7783,
+      "step": 4131
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.113838972747051e-05,
+      "loss": 0.7422,
+      "step": 4132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.110006205044307e-05,
+      "loss": 0.7705,
+      "step": 4133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1061765053748438e-05,
+      "loss": 0.7741,
+      "step": 4134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.102349875227847e-05,
+      "loss": 0.6443,
+      "step": 4135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.098526316091308e-05,
+      "loss": 0.8123,
+      "step": 4136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0947058294520183e-05,
+      "loss": 0.7837,
+      "step": 4137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0908884167955824e-05,
+      "loss": 0.7117,
+      "step": 4138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0870740796064013e-05,
+      "loss": 0.7106,
+      "step": 4139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0832628193676863e-05,
+      "loss": 0.7807,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0794546375614543e-05,
+      "loss": 0.7568,
+      "step": 4141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0756495356685147e-05,
+      "loss": 0.7346,
+      "step": 4142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0718475151684912e-05,
+      "loss": 0.7672,
+      "step": 4143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0680485775397984e-05,
+      "loss": 0.7433,
+      "step": 4144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.064252724259662e-05,
+      "loss": 0.8183,
+      "step": 4145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0604599568041026e-05,
+      "loss": 0.7471,
+      "step": 4146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.056670276647945e-05,
+      "loss": 0.7595,
+      "step": 4147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0528836852648058e-05,
+      "loss": 0.6895,
+      "step": 4148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0491001841271074e-05,
+      "loss": 0.7314,
+      "step": 4149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0453197747060726e-05,
+      "loss": 0.7997,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0415424584717113e-05,
+      "loss": 0.7224,
+      "step": 4151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.037768236892841e-05,
+      "loss": 0.671,
+      "step": 4152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0339971114370738e-05,
+      "loss": 0.7774,
+      "step": 4153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0302290835708116e-05,
+      "loss": 0.69,
+      "step": 4154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0264641547592612e-05,
+      "loss": 0.7363,
+      "step": 4155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0227023264664156e-05,
+      "loss": 0.7113,
+      "step": 4156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0189436001550665e-05,
+      "loss": 0.6515,
+      "step": 4157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0151879772868042e-05,
+      "loss": 0.6897,
+      "step": 4158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.011435459322001e-05,
+      "loss": 0.7281,
+      "step": 4159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0076860477198313e-05,
+      "loss": 0.819,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0039397439382544e-05,
+      "loss": 0.7709,
+      "step": 4161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0001965494340304e-05,
+      "loss": 0.8167,
+      "step": 4162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9964564656626995e-05,
+      "loss": 0.6518,
+      "step": 4163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9927194940785998e-05,
+      "loss": 0.7319,
+      "step": 4164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9889856361348603e-05,
+      "loss": 0.7293,
+      "step": 4165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9852548932833893e-05,
+      "loss": 0.7196,
+      "step": 4166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.981527266974895e-05,
+      "loss": 0.7644,
+      "step": 4167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9778027586588686e-05,
+      "loss": 0.7117,
+      "step": 4168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9740813697835915e-05,
+      "loss": 0.6938,
+      "step": 4169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9703631017961243e-05,
+      "loss": 0.7333,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9666479561423244e-05,
+      "loss": 0.6888,
+      "step": 4171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962935934266833e-05,
+      "loss": 0.7162,
+      "step": 4172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9592270376130672e-05,
+      "loss": 0.7624,
+      "step": 4173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9555212676232394e-05,
+      "loss": 0.6077,
+      "step": 4174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9518186257383454e-05,
+      "loss": 0.6979,
+      "step": 4175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.948119113398156e-05,
+      "loss": 0.7734,
+      "step": 4176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9444227320412357e-05,
+      "loss": 0.5744,
+      "step": 4177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.940729483104923e-05,
+      "loss": 0.716,
+      "step": 4178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9370393680253462e-05,
+      "loss": 0.6938,
+      "step": 4179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9333523882374062e-05,
+      "loss": 0.6383,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.929668545174791e-05,
+      "loss": 0.6833,
+      "step": 4181
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9259878402699705e-05,
+      "loss": 0.6631,
+      "step": 4182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9223102749541855e-05,
+      "loss": 0.5917,
+      "step": 4183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9186358506574686e-05,
+      "loss": 0.7793,
+      "step": 4184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914964568808616e-05,
+      "loss": 0.7133,
+      "step": 4185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9112964308352142e-05,
+      "loss": 0.7246,
+      "step": 4186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.907631438163626e-05,
+      "loss": 0.7593,
+      "step": 4187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9039695922189814e-05,
+      "loss": 0.7157,
+      "step": 4188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.900310894425197e-05,
+      "loss": 0.7167,
+      "step": 4189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.896655346204961e-05,
+      "loss": 0.7605,
+      "step": 4190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8930029489797408e-05,
+      "loss": 0.7093,
+      "step": 4191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8893537041697707e-05,
+      "loss": 0.6035,
+      "step": 4192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8857076131940642e-05,
+      "loss": 0.8843,
+      "step": 4193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8820646774704133e-05,
+      "loss": 0.7022,
+      "step": 4194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8784248984153707e-05,
+      "loss": 0.6326,
+      "step": 4195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.874788277444275e-05,
+      "loss": 0.6928,
+      "step": 4196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8711548159712243e-05,
+      "loss": 0.809,
+      "step": 4197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.867524515409097e-05,
+      "loss": 0.7326,
+      "step": 4198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8638973771695443e-05,
+      "loss": 0.7089,
+      "step": 4199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8602734026629754e-05,
+      "loss": 0.6433,
+      "step": 4200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8566525932985833e-05,
+      "loss": 0.7284,
+      "step": 4201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8530349504843192e-05,
+      "loss": 0.7196,
+      "step": 4202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8494204756269095e-05,
+      "loss": 0.7064,
+      "step": 4203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8458091701318504e-05,
+      "loss": 0.6924,
+      "step": 4204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.842201035403396e-05,
+      "loss": 0.7018,
+      "step": 4205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8385960728445794e-05,
+      "loss": 0.7387,
+      "step": 4206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8349942838571898e-05,
+      "loss": 0.6288,
+      "step": 4207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.831395669841789e-05,
+      "loss": 0.7134,
+      "step": 4208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.827800232197705e-05,
+      "loss": 0.8032,
+      "step": 4209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.824207972323023e-05,
+      "loss": 0.6791,
+      "step": 4210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8206188916145993e-05,
+      "loss": 0.6402,
+      "step": 4211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.817032991468053e-05,
+      "loss": 0.6902,
+      "step": 4212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.813450273277767e-05,
+      "loss": 0.7664,
+      "step": 4213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.809870738436882e-05,
+      "loss": 0.769,
+      "step": 4214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.806294388337305e-05,
+      "loss": 0.6528,
+      "step": 4215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.802721224369708e-05,
+      "loss": 0.692,
+      "step": 4216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.799151247923514e-05,
+      "loss": 0.6859,
+      "step": 4217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7955844603869177e-05,
+      "loss": 0.7175,
+      "step": 4218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.792020863146864e-05,
+      "loss": 0.7257,
+      "step": 4219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.788460457589064e-05,
+      "loss": 0.7825,
+      "step": 4220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7849032450979876e-05,
+      "loss": 0.6915,
+      "step": 4221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7813492270568567e-05,
+      "loss": 0.727,
+      "step": 4222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7777984048476605e-05,
+      "loss": 0.7185,
+      "step": 4223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7742507798511343e-05,
+      "loss": 0.7088,
+      "step": 4224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7707063534467816e-05,
+      "loss": 0.6516,
+      "step": 4225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7671651270128532e-05,
+      "loss": 0.6847,
+      "step": 4226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7636271019263596e-05,
+      "loss": 0.6749,
+      "step": 4227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.760092279563069e-05,
+      "loss": 0.7439,
+      "step": 4228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7565606612974982e-05,
+      "loss": 0.7319,
+      "step": 4229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7530322485029206e-05,
+      "loss": 0.8684,
+      "step": 4230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7495070425513648e-05,
+      "loss": 0.7428,
+      "step": 4231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.745985044813615e-05,
+      "loss": 0.6932,
+      "step": 4232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7424662566591965e-05,
+      "loss": 0.809,
+      "step": 4233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7389506794563993e-05,
+      "loss": 0.7763,
+      "step": 4234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7354383145722608e-05,
+      "loss": 0.825,
+      "step": 4235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.731929163372563e-05,
+      "loss": 0.6764,
+      "step": 4236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7284232272218504e-05,
+      "loss": 0.7341,
+      "step": 4237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7249205074834017e-05,
+      "loss": 0.6848,
+      "step": 4238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7214210055192602e-05,
+      "loss": 0.7544,
+      "step": 4239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7179247226902107e-05,
+      "loss": 0.7534,
+      "step": 4240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.714431660355783e-05,
+      "loss": 0.6672,
+      "step": 4241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7109418198742642e-05,
+      "loss": 0.6837,
+      "step": 4242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7074552026026757e-05,
+      "loss": 0.7148,
+      "step": 4243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703971809896798e-05,
+      "loss": 0.7063,
+      "step": 4244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7004916431111528e-05,
+      "loss": 0.8128,
+      "step": 4245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6970147035990024e-05,
+      "loss": 0.6876,
+      "step": 4246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6935409927123648e-05,
+      "loss": 0.688,
+      "step": 4247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.69007051180199e-05,
+      "loss": 0.6857,
+      "step": 4248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6866032622173823e-05,
+      "loss": 0.7721,
+      "step": 4249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6831392453067885e-05,
+      "loss": 0.7414,
+      "step": 4250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6796784624171892e-05,
+      "loss": 0.7367,
+      "step": 4251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6762209148943163e-05,
+      "loss": 0.6389,
+      "step": 4252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.672766604082643e-05,
+      "loss": 0.6921,
+      "step": 4253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.669315531325383e-05,
+      "loss": 0.7007,
+      "step": 4254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.665867697964486e-05,
+      "loss": 0.7916,
+      "step": 4255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6624231053406468e-05,
+      "loss": 0.6637,
+      "step": 4256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.658981754793304e-05,
+      "loss": 0.7229,
+      "step": 4257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6555436476606245e-05,
+      "loss": 0.5804,
+      "step": 4258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.652108785279526e-05,
+      "loss": 0.7455,
+      "step": 4259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6486771689856528e-05,
+      "loss": 0.7414,
+      "step": 4260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.645248800113396e-05,
+      "loss": 0.6083,
+      "step": 4261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.641823679995883e-05,
+      "loss": 0.6445,
+      "step": 4262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6384018099649713e-05,
+      "loss": 0.7598,
+      "step": 4263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6349831913512625e-05,
+      "loss": 0.742,
+      "step": 4264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6315678254840884e-05,
+      "loss": 0.7018,
+      "step": 4265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.628155713691517e-05,
+      "loss": 0.6743,
+      "step": 4266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6247468573003564e-05,
+      "loss": 0.7834,
+      "step": 4267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6213412576361396e-05,
+      "loss": 0.8779,
+      "step": 4268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.617938916023143e-05,
+      "loss": 0.7773,
+      "step": 4269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6145398337843652e-05,
+      "loss": 0.6589,
+      "step": 4270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6111440122415488e-05,
+      "loss": 0.7928,
+      "step": 4271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6077514527151573e-05,
+      "loss": 0.726,
+      "step": 4272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6043621565243948e-05,
+      "loss": 0.764,
+      "step": 4273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6009761249871935e-05,
+      "loss": 0.7463,
+      "step": 4274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.597593359420214e-05,
+      "loss": 0.6557,
+      "step": 4275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5942138611388535e-05,
+      "loss": 0.8115,
+      "step": 4276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.590837631457226e-05,
+      "loss": 0.7159,
+      "step": 4277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587464671688187e-05,
+      "loss": 0.7357,
+      "step": 4278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5840949831433172e-05,
+      "loss": 0.6784,
+      "step": 4279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5807285671329196e-05,
+      "loss": 0.7903,
+      "step": 4280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.577365424966034e-05,
+      "loss": 0.7997,
+      "step": 4281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5740055579504176e-05,
+      "loss": 0.7551,
+      "step": 4282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5706489673925616e-05,
+      "loss": 0.6254,
+      "step": 4283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.567295654597677e-05,
+      "loss": 0.7335,
+      "step": 4284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5639456208697058e-05,
+      "loss": 0.6401,
+      "step": 4285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5605988675113136e-05,
+      "loss": 0.6932,
+      "step": 4286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5572553958238868e-05,
+      "loss": 0.706,
+      "step": 4287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5539152071075403e-05,
+      "loss": 0.7251,
+      "step": 4288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5505783026611075e-05,
+      "loss": 0.756,
+      "step": 4289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5472446837821485e-05,
+      "loss": 0.7178,
+      "step": 4290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5439143517669495e-05,
+      "loss": 0.7211,
+      "step": 4291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.540587307910508e-05,
+      "loss": 0.7702,
+      "step": 4292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5372635535065503e-05,
+      "loss": 0.7424,
+      "step": 4293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.533943089847526e-05,
+      "loss": 0.734,
+      "step": 4294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5306259182245963e-05,
+      "loss": 0.6798,
+      "step": 4295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5273120399276487e-05,
+      "loss": 0.6587,
+      "step": 4296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5240014562452887e-05,
+      "loss": 0.6674,
+      "step": 4297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5206941684648446e-05,
+      "loss": 0.6772,
+      "step": 4298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5173901778723531e-05,
+      "loss": 0.7758,
+      "step": 4299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.514089485752579e-05,
+      "loss": 0.6066,
+      "step": 4300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5107920933889974e-05,
+      "loss": 0.7288,
+      "step": 4301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.507498002063803e-05,
+      "loss": 0.6973,
+      "step": 4302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.504207213057912e-05,
+      "loss": 0.6538,
+      "step": 4303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5009197276509446e-05,
+      "loss": 0.7638,
+      "step": 4304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.49763554712125e-05,
+      "loss": 0.7752,
+      "step": 4305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4943546727458801e-05,
+      "loss": 0.6859,
+      "step": 4306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4910771058006079e-05,
+      "loss": 0.663,
+      "step": 4307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4878028475599237e-05,
+      "loss": 0.7498,
+      "step": 4308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4845318992970203e-05,
+      "loss": 0.7001,
+      "step": 4309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4812642622838147e-05,
+      "loss": 0.6404,
+      "step": 4310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.477999937790927e-05,
+      "loss": 0.7018,
+      "step": 4311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4747389270876976e-05,
+      "loss": 0.7214,
+      "step": 4312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4714812314421711e-05,
+      "loss": 0.694,
+      "step": 4313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4682268521211073e-05,
+      "loss": 0.722,
+      "step": 4314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4649757903899752e-05,
+      "loss": 0.6824,
+      "step": 4315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.461728047512958e-05,
+      "loss": 0.6531,
+      "step": 4316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4584836247529377e-05,
+      "loss": 0.7862,
+      "step": 4317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.455242523371515e-05,
+      "loss": 0.6401,
+      "step": 4318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4520047446289964e-05,
+      "loss": 0.6049,
+      "step": 4319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.448770289784397e-05,
+      "loss": 0.7272,
+      "step": 4320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.445539160095436e-05,
+      "loss": 0.7242,
+      "step": 4321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4423113568185432e-05,
+      "loss": 0.7001,
+      "step": 4322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.439086881208852e-05,
+      "loss": 0.7209,
+      "step": 4323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4358657345202043e-05,
+      "loss": 0.6967,
+      "step": 4324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.43264791800515e-05,
+      "loss": 0.906,
+      "step": 4325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.429433432914935e-05,
+      "loss": 0.7187,
+      "step": 4326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4262222804995218e-05,
+      "loss": 0.7992,
+      "step": 4327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4230144620075648e-05,
+      "loss": 0.7253,
+      "step": 4328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4198099786864327e-05,
+      "loss": 0.6273,
+      "step": 4329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.416608831782188e-05,
+      "loss": 0.737,
+      "step": 4330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4134110225396024e-05,
+      "loss": 0.6888,
+      "step": 4331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4102165522021516e-05,
+      "loss": 0.7073,
+      "step": 4332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4070254220120038e-05,
+      "loss": 0.7254,
+      "step": 4333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4038376332100379e-05,
+      "loss": 0.6424,
+      "step": 4334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4006531870358241e-05,
+      "loss": 0.6841,
+      "step": 4335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3974720847276412e-05,
+      "loss": 0.7883,
+      "step": 4336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3942943275224641e-05,
+      "loss": 0.7059,
+      "step": 4337
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3911199166559675e-05,
+      "loss": 0.7376,
+      "step": 4338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3879488533625285e-05,
+      "loss": 0.7679,
+      "step": 4339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3847811388752119e-05,
+      "loss": 0.6341,
+      "step": 4340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3816167744257901e-05,
+      "loss": 0.7561,
+      "step": 4341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3784557612447324e-05,
+      "loss": 0.6866,
+      "step": 4342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3752981005611986e-05,
+      "loss": 0.7663,
+      "step": 4343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3721437936030523e-05,
+      "loss": 0.69,
+      "step": 4344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3689928415968455e-05,
+      "loss": 0.732,
+      "step": 4345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3658452457678351e-05,
+      "loss": 0.7553,
+      "step": 4346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3627010073399604e-05,
+      "loss": 0.6852,
+      "step": 4347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.359560127535865e-05,
+      "loss": 0.824,
+      "step": 4348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3564226075768882e-05,
+      "loss": 0.7237,
+      "step": 4349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3532884486830511e-05,
+      "loss": 0.6444,
+      "step": 4350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3501576520730808e-05,
+      "loss": 0.7735,
+      "step": 4351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.347030218964387e-05,
+      "loss": 0.7299,
+      "step": 4352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3439061505730777e-05,
+      "loss": 0.7348,
+      "step": 4353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3407854481139515e-05,
+      "loss": 0.6128,
+      "step": 4354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3376681128004942e-05,
+      "loss": 0.7419,
+      "step": 4355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3345541458448895e-05,
+      "loss": 0.7509,
+      "step": 4356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3314435484580034e-05,
+      "loss": 0.6665,
+      "step": 4357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.328336321849396e-05,
+      "loss": 0.766,
+      "step": 4358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3252324672273165e-05,
+      "loss": 0.6995,
+      "step": 4359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3221319857987025e-05,
+      "loss": 0.6692,
+      "step": 4360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3190348787691809e-05,
+      "loss": 0.7022,
+      "step": 4361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3159411473430638e-05,
+      "loss": 0.7363,
+      "step": 4362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3128507927233536e-05,
+      "loss": 0.6255,
+      "step": 4363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3097638161117364e-05,
+      "loss": 0.7413,
+      "step": 4364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.306680218708587e-05,
+      "loss": 0.7544,
+      "step": 4365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.303600001712969e-05,
+      "loss": 0.7929,
+      "step": 4366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3005231663226236e-05,
+      "loss": 0.6789,
+      "step": 4367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.297449713733987e-05,
+      "loss": 0.7014,
+      "step": 4368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2943796451421686e-05,
+      "loss": 0.639,
+      "step": 4369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2913129617409724e-05,
+      "loss": 0.6737,
+      "step": 4370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2882496647228825e-05,
+      "loss": 0.7145,
+      "step": 4371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2851897552790627e-05,
+      "loss": 0.6719,
+      "step": 4372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2821332345993653e-05,
+      "loss": 0.7694,
+      "step": 4373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2790801038723177e-05,
+      "loss": 0.6883,
+      "step": 4374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2760303642851378e-05,
+      "loss": 0.6097,
+      "step": 4375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2729840170237172e-05,
+      "loss": 0.5996,
+      "step": 4376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2699410632726327e-05,
+      "loss": 0.8025,
+      "step": 4377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2669015042151422e-05,
+      "loss": 0.6827,
+      "step": 4378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.263865341033178e-05,
+      "loss": 0.7057,
+      "step": 4379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2608325749073591e-05,
+      "loss": 0.7404,
+      "step": 4380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2578032070169776e-05,
+      "loss": 0.6075,
+      "step": 4381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2547772385400081e-05,
+      "loss": 0.677,
+      "step": 4382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2517546706531046e-05,
+      "loss": 0.759,
+      "step": 4383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2487355045315907e-05,
+      "loss": 0.6363,
+      "step": 4384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.245719741349477e-05,
+      "loss": 0.6626,
+      "step": 4385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2427073822794443e-05,
+      "loss": 0.716,
+      "step": 4386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2396984284928515e-05,
+      "loss": 0.7993,
+      "step": 4387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2366928811597378e-05,
+      "loss": 0.7402,
+      "step": 4388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2336907414488074e-05,
+      "loss": 0.6812,
+      "step": 4389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2306920105274522e-05,
+      "loss": 0.6843,
+      "step": 4390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.227696689561727e-05,
+      "loss": 0.7958,
+      "step": 4391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2247047797163701e-05,
+      "loss": 0.6826,
+      "step": 4392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2217162821547857e-05,
+      "loss": 0.7214,
+      "step": 4393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.218731198039056e-05,
+      "loss": 0.7214,
+      "step": 4394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2157495285299358e-05,
+      "loss": 0.6673,
+      "step": 4395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.212771274786847e-05,
+      "loss": 0.6729,
+      "step": 4396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2097964379678927e-05,
+      "loss": 0.7229,
+      "step": 4397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2068250192298359e-05,
+      "loss": 0.803,
+      "step": 4398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2038570197281185e-05,
+      "loss": 0.7364,
+      "step": 4399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2008924406168509e-05,
+      "loss": 0.7286,
+      "step": 4400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1979312830488143e-05,
+      "loss": 0.7395,
+      "step": 4401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1949735481754565e-05,
+      "loss": 0.7272,
+      "step": 4402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1920192371468952e-05,
+      "loss": 0.7539,
+      "step": 4403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1890683511119228e-05,
+      "loss": 0.6495,
+      "step": 4404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1861208912179877e-05,
+      "loss": 0.6528,
+      "step": 4405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.183176858611218e-05,
+      "loss": 0.7284,
+      "step": 4406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1802362544364031e-05,
+      "loss": 0.6942,
+      "step": 4407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1772990798369988e-05,
+      "loss": 0.7071,
+      "step": 4408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.174365335955132e-05,
+      "loss": 0.6604,
+      "step": 4409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1714350239315897e-05,
+      "loss": 0.6925,
+      "step": 4410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1685081449058267e-05,
+      "loss": 0.5885,
+      "step": 4411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.165584700015967e-05,
+      "loss": 0.736,
+      "step": 4412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1626646903987904e-05,
+      "loss": 0.7673,
+      "step": 4413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1597481171897517e-05,
+      "loss": 0.777,
+      "step": 4414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1568349815229584e-05,
+      "loss": 0.6444,
+      "step": 4415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1539252845311888e-05,
+      "loss": 0.6941,
+      "step": 4416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1510190273458854e-05,
+      "loss": 0.7044,
+      "step": 4417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1481162110971444e-05,
+      "loss": 0.7374,
+      "step": 4418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1452168369137329e-05,
+      "loss": 0.6864,
+      "step": 4419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1423209059230721e-05,
+      "loss": 0.6883,
+      "step": 4420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1394284192512517e-05,
+      "loss": 0.6888,
+      "step": 4421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1365393780230171e-05,
+      "loss": 0.6518,
+      "step": 4422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1336537833617777e-05,
+      "loss": 0.7837,
+      "step": 4423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.130771636389596e-05,
+      "loss": 0.6796,
+      "step": 4424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1278929382272007e-05,
+      "loss": 0.7368,
+      "step": 4425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1250176899939791e-05,
+      "loss": 0.7762,
+      "step": 4426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1221458928079698e-05,
+      "loss": 0.7934,
+      "step": 4427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1192775477858774e-05,
+      "loss": 0.6855,
+      "step": 4428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1164126560430632e-05,
+      "loss": 0.7401,
+      "step": 4429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1135512186935392e-05,
+      "loss": 0.6204,
+      "step": 4430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1106932368499823e-05,
+      "loss": 0.7284,
+      "step": 4431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.10783871162372e-05,
+      "loss": 0.6832,
+      "step": 4432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1049876441247397e-05,
+      "loss": 0.6308,
+      "step": 4433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1021400354616785e-05,
+      "loss": 0.6869,
+      "step": 4434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0992958867418357e-05,
+      "loss": 0.6748,
+      "step": 4435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0964551990711625e-05,
+      "loss": 0.7774,
+      "step": 4436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0936179735542606e-05,
+      "loss": 0.6897,
+      "step": 4437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0907842112943923e-05,
+      "loss": 0.7691,
+      "step": 4438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0879539133934647e-05,
+      "loss": 0.8021,
+      "step": 4439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0851270809520442e-05,
+      "loss": 0.7908,
+      "step": 4440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0823037150693517e-05,
+      "loss": 0.7262,
+      "step": 4441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0794838168432497e-05,
+      "loss": 0.7197,
+      "step": 4442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0766673873702638e-05,
+      "loss": 0.7534,
+      "step": 4443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0738544277455631e-05,
+      "loss": 0.731,
+      "step": 4444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0710449390629717e-05,
+      "loss": 0.7238,
+      "step": 4445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0682389224149647e-05,
+      "loss": 0.6368,
+      "step": 4446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0654363788926614e-05,
+      "loss": 0.803,
+      "step": 4447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0626373095858366e-05,
+      "loss": 0.6902,
+      "step": 4448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0598417155829076e-05,
+      "loss": 0.6975,
+      "step": 4449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0570495979709504e-05,
+      "loss": 0.6606,
+      "step": 4450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0542609578356777e-05,
+      "loss": 0.6859,
+      "step": 4451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0514757962614574e-05,
+      "loss": 0.7342,
+      "step": 4452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0486941143313045e-05,
+      "loss": 0.6663,
+      "step": 4453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.045915913126877e-05,
+      "loss": 0.6463,
+      "step": 4454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0431411937284841e-05,
+      "loss": 0.8073,
+      "step": 4455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.040369957215076e-05,
+      "loss": 0.7473,
+      "step": 4456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.037602204664252e-05,
+      "loss": 0.676,
+      "step": 4457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0348379371522576e-05,
+      "loss": 0.6661,
+      "step": 4458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0320771557539788e-05,
+      "loss": 0.715,
+      "step": 4459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0293198615429523e-05,
+      "loss": 0.7028,
+      "step": 4460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0265660555913503e-05,
+      "loss": 0.6456,
+      "step": 4461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0238157389699954e-05,
+      "loss": 0.661,
+      "step": 4462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0210689127483553e-05,
+      "loss": 0.7103,
+      "step": 4463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0183255779945312e-05,
+      "loss": 0.6864,
+      "step": 4464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.015585735775274e-05,
+      "loss": 0.7731,
+      "step": 4465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.012849387155974e-05,
+      "loss": 0.6802,
+      "step": 4466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0101165332006668e-05,
+      "loss": 0.78,
+      "step": 4467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0073871749720221e-05,
+      "loss": 0.6911,
+      "step": 4468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0046613135313542e-05,
+      "loss": 0.6779,
+      "step": 4469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0019389499386212e-05,
+      "loss": 0.5889,
+      "step": 4470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.992200852524125e-06,
+      "loss": 0.7473,
+      "step": 4471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.965047205299648e-06,
+      "loss": 0.7203,
+      "step": 4472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.937928568271492e-06,
+      "loss": 0.7667,
+      "step": 4473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.910844951984789e-06,
+      "loss": 0.7183,
+      "step": 4474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.883796366971021e-06,
+      "loss": 0.7097,
+      "step": 4475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.856782823748067e-06,
+      "loss": 0.7245,
+      "step": 4476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.82980433282018e-06,
+      "loss": 0.6504,
+      "step": 4477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.802860904677957e-06,
+      "loss": 0.7368,
+      "step": 4478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.775952549798406e-06,
+      "loss": 0.7065,
+      "step": 4479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.749079278644846e-06,
+      "loss": 0.7547,
+      "step": 4480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.722241101667007e-06,
+      "loss": 0.8309,
+      "step": 4481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.69543802930094e-06,
+      "loss": 0.7,
+      "step": 4482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.668670071969033e-06,
+      "loss": 0.573,
+      "step": 4483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.641937240080045e-06,
+      "loss": 0.7132,
+      "step": 4484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.615239544029064e-06,
+      "loss": 0.8048,
+      "step": 4485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.588576994197562e-06,
+      "loss": 0.6019,
+      "step": 4486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.561949600953247e-06,
+      "loss": 0.6673,
+      "step": 4487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.535357374650234e-06,
+      "loss": 0.7003,
+      "step": 4488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.508800325628953e-06,
+      "loss": 0.6457,
+      "step": 4489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.482278464216121e-06,
+      "loss": 0.7696,
+      "step": 4490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.455791800724801e-06,
+      "loss": 0.6607,
+      "step": 4491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.429340345454396e-06,
+      "loss": 0.7039,
+      "step": 4492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.402924108690536e-06,
+      "loss": 0.7342,
+      "step": 4493
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.376543100705238e-06,
+      "loss": 0.7902,
+      "step": 4494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.35019733175676e-06,
+      "loss": 0.73,
+      "step": 4495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.323886812089722e-06,
+      "loss": 0.6531,
+      "step": 4496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.29761155193496e-06,
+      "loss": 0.7918,
+      "step": 4497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.271371561509657e-06,
+      "loss": 0.7135,
+      "step": 4498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.245166851017285e-06,
+      "loss": 0.6326,
+      "step": 4499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.218997430647547e-06,
+      "loss": 0.6951,
+      "step": 4500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.192863310576472e-06,
+      "loss": 0.6773,
+      "step": 4501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.166764500966319e-06,
+      "loss": 0.7307,
+      "step": 4502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.140701011965646e-06,
+      "loss": 0.6808,
+      "step": 4503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.114672853709304e-06,
+      "loss": 0.6761,
+      "step": 4504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.088680036318342e-06,
+      "loss": 0.5612,
+      "step": 4505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.062722569900083e-06,
+      "loss": 0.6442,
+      "step": 4506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.036800464548157e-06,
+      "loss": 0.6931,
+      "step": 4507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.010913730342396e-06,
+      "loss": 0.6615,
+      "step": 4508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.985062377348862e-06,
+      "loss": 0.624,
+      "step": 4509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.959246415619904e-06,
+      "loss": 0.738,
+      "step": 4510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.9334658551941e-06,
+      "loss": 0.6919,
+      "step": 4511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.907720706096224e-06,
+      "loss": 0.6761,
+      "step": 4512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.88201097833733e-06,
+      "loss": 0.6736,
+      "step": 4513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.856336681914645e-06,
+      "loss": 0.6091,
+      "step": 4514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.830697826811674e-06,
+      "loss": 0.7381,
+      "step": 4515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.805094422998117e-06,
+      "loss": 0.7067,
+      "step": 4516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.779526480429867e-06,
+      "loss": 0.6744,
+      "step": 4517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.753994009049083e-06,
+      "loss": 0.7859,
+      "step": 4518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.728497018784065e-06,
+      "loss": 0.7309,
+      "step": 4519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.70303551954934e-06,
+      "loss": 0.6256,
+      "step": 4520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.67760952124569e-06,
+      "loss": 0.6979,
+      "step": 4521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.652219033759989e-06,
+      "loss": 0.666,
+      "step": 4522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.626864066965402e-06,
+      "loss": 0.6166,
+      "step": 4523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.601544630721203e-06,
+      "loss": 0.5743,
+      "step": 4524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.57626073487291e-06,
+      "loss": 0.6343,
+      "step": 4525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.551012389252178e-06,
+      "loss": 0.7327,
+      "step": 4526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.525799603676842e-06,
+      "loss": 0.7553,
+      "step": 4527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.500622387950963e-06,
+      "loss": 0.8017,
+      "step": 4528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.475480751864695e-06,
+      "loss": 0.6941,
+      "step": 4529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.450374705194431e-06,
+      "loss": 0.663,
+      "step": 4530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.425304257702649e-06,
+      "loss": 0.6931,
+      "step": 4531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.40026941913803e-06,
+      "loss": 0.7231,
+      "step": 4532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.37527019923543e-06,
+      "loss": 0.7334,
+      "step": 4533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.350306607715774e-06,
+      "loss": 0.7525,
+      "step": 4534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.325378654286231e-06,
+      "loss": 0.6596,
+      "step": 4535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.300486348640035e-06,
+      "loss": 0.6842,
+      "step": 4536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.275629700456588e-06,
+      "loss": 0.8339,
+      "step": 4537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.25080871940146e-06,
+      "loss": 0.6661,
+      "step": 4538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.226023415126272e-06,
+      "loss": 0.6513,
+      "step": 4539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.20127379726886e-06,
+      "loss": 0.6837,
+      "step": 4540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.176559875453105e-06,
+      "loss": 0.6881,
+      "step": 4541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.151881659289084e-06,
+      "loss": 0.737,
+      "step": 4542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.127239158372902e-06,
+      "loss": 0.6912,
+      "step": 4543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.102632382286857e-06,
+      "loss": 0.718,
+      "step": 4544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.07806134059933e-06,
+      "loss": 0.7524,
+      "step": 4545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.053526042864768e-06,
+      "loss": 0.73,
+      "step": 4546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.029026498623771e-06,
+      "loss": 0.8184,
+      "step": 4547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.004562717403019e-06,
+      "loss": 0.6904,
+      "step": 4548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.98013470871527e-06,
+      "loss": 0.6559,
+      "step": 4549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.95574248205938e-06,
+      "loss": 0.6335,
+      "step": 4550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.931386046920308e-06,
+      "loss": 0.7242,
+      "step": 4551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.907065412769088e-06,
+      "loss": 0.7296,
+      "step": 4552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.882780589062799e-06,
+      "loss": 0.7716,
+      "step": 4553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.858531585244654e-06,
+      "loss": 0.6499,
+      "step": 4554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.83431841074389e-06,
+      "loss": 0.7364,
+      "step": 4555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.810141074975818e-06,
+      "loss": 0.6109,
+      "step": 4556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.785999587341853e-06,
+      "loss": 0.6931,
+      "step": 4557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.761893957229405e-06,
+      "loss": 0.7288,
+      "step": 4558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.737824194012022e-06,
+      "loss": 0.6579,
+      "step": 4559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.713790307049206e-06,
+      "loss": 0.7714,
+      "step": 4560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.689792305686583e-06,
+      "loss": 0.7472,
+      "step": 4561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.665830199255842e-06,
+      "loss": 0.6079,
+      "step": 4562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.641903997074607e-06,
+      "loss": 0.7112,
+      "step": 4563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.61801370844667e-06,
+      "loss": 0.6939,
+      "step": 4564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.594159342661767e-06,
+      "loss": 0.7262,
+      "step": 4565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.570340908995688e-06,
+      "loss": 0.7048,
+      "step": 4566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.546558416710292e-06,
+      "loss": 0.7038,
+      "step": 4567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.522811875053393e-06,
+      "loss": 0.6123,
+      "step": 4568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4991012932588815e-06,
+      "loss": 0.6797,
+      "step": 4569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.47542668054666e-06,
+      "loss": 0.5977,
+      "step": 4570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.451788046122587e-06,
+      "loss": 0.7294,
+      "step": 4571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4281853991786e-06,
+      "loss": 0.775,
+      "step": 4572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.404618748892611e-06,
+      "loss": 0.6365,
+      "step": 4573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3810881044285576e-06,
+      "loss": 0.7489,
+      "step": 4574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.35759347493632e-06,
+      "loss": 0.6351,
+      "step": 4575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.334134869551856e-06,
+      "loss": 0.7357,
+      "step": 4576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3107122973970376e-06,
+      "loss": 0.714,
+      "step": 4577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.287325767579756e-06,
+      "loss": 0.7326,
+      "step": 4578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2639752891939365e-06,
+      "loss": 0.7347,
+      "step": 4579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.240660871319382e-06,
+      "loss": 0.7273,
+      "step": 4580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.217382523021987e-06,
+      "loss": 0.6888,
+      "step": 4581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.19414025335351e-06,
+      "loss": 0.699,
+      "step": 4582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.170934071351754e-06,
+      "loss": 0.7948,
+      "step": 4583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.147763986040501e-06,
+      "loss": 0.6672,
+      "step": 4584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.124630006429422e-06,
+      "loss": 0.6451,
+      "step": 4585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.101532141514233e-06,
+      "loss": 0.7284,
+      "step": 4586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.078470400276516e-06,
+      "loss": 0.7811,
+      "step": 4587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.055444791683907e-06,
+      "loss": 0.6469,
+      "step": 4588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.032455324689902e-06,
+      "loss": 0.7141,
+      "step": 4589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.009502008233992e-06,
+      "loss": 0.6516,
+      "step": 4590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.986584851241606e-06,
+      "loss": 0.6008,
+      "step": 4591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.963703862624105e-06,
+      "loss": 0.6572,
+      "step": 4592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.940859051278814e-06,
+      "loss": 0.6957,
+      "step": 4593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.9180504260889175e-06,
+      "loss": 0.5993,
+      "step": 4594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.895277995923599e-06,
+      "loss": 0.6547,
+      "step": 4595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.872541769637963e-06,
+      "loss": 0.6037,
+      "step": 4596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.849841756072983e-06,
+      "loss": 0.7321,
+      "step": 4597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.82717796405562e-06,
+      "loss": 0.6354,
+      "step": 4598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.804550402398679e-06,
+      "loss": 0.6885,
+      "step": 4599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.781959079900957e-06,
+      "loss": 0.7976,
+      "step": 4600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.75940400534707e-06,
+      "loss": 0.7078,
+      "step": 4601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.736885187507602e-06,
+      "loss": 0.7186,
+      "step": 4602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.71440263513905e-06,
+      "loss": 0.6637,
+      "step": 4603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.691956356983731e-06,
+      "loss": 0.7551,
+      "step": 4604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.6695463617699515e-06,
+      "loss": 0.6997,
+      "step": 4605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.647172658211831e-06,
+      "loss": 0.7299,
+      "step": 4606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.624835255009432e-06,
+      "loss": 0.7224,
+      "step": 4607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.602534160848672e-06,
+      "loss": 0.6946,
+      "step": 4608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.58026938440135e-06,
+      "loss": 0.6723,
+      "step": 4609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.558040934325183e-06,
+      "loss": 0.7288,
+      "step": 4610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.535848819263679e-06,
+      "loss": 0.7181,
+      "step": 4611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.5136930478462925e-06,
+      "loss": 0.6279,
+      "step": 4612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.491573628688319e-06,
+      "loss": 0.6675,
+      "step": 4613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4694905703909345e-06,
+      "loss": 0.7635,
+      "step": 4614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.44744388154116e-06,
+      "loss": 0.675,
+      "step": 4615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.425433570711847e-06,
+      "loss": 0.8185,
+      "step": 4616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.403459646461774e-06,
+      "loss": 0.6269,
+      "step": 4617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.381522117335481e-06,
+      "loss": 0.6419,
+      "step": 4618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.359620991863424e-06,
+      "loss": 0.7128,
+      "step": 4619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.337756278561902e-06,
+      "loss": 0.6653,
+      "step": 4620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.315927985933001e-06,
+      "loss": 0.6285,
+      "step": 4621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2941361224647e-06,
+      "loss": 0.6952,
+      "step": 4622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.272380696630764e-06,
+      "loss": 0.7345,
+      "step": 4623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.250661716890838e-06,
+      "loss": 0.6977,
+      "step": 4624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.228979191690387e-06,
+      "loss": 0.7379,
+      "step": 4625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.207333129460646e-06,
+      "loss": 0.6909,
+      "step": 4626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.185723538618738e-06,
+      "loss": 0.7024,
+      "step": 4627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.164150427567572e-06,
+      "loss": 0.6378,
+      "step": 4628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.142613804695885e-06,
+      "loss": 0.734,
+      "step": 4629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.121113678378199e-06,
+      "loss": 0.6995,
+      "step": 4630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.099650056974871e-06,
+      "loss": 0.678,
+      "step": 4631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.078222948832058e-06,
+      "loss": 0.8137,
+      "step": 4632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.056832362281728e-06,
+      "loss": 0.7591,
+      "step": 4633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.035478305641618e-06,
+      "loss": 0.6136,
+      "step": 4634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.014160787215273e-06,
+      "loss": 0.685,
+      "step": 4635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.992879815292052e-06,
+      "loss": 0.7389,
+      "step": 4636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.971635398147091e-06,
+      "loss": 0.7517,
+      "step": 4637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.950427544041293e-06,
+      "loss": 0.6657,
+      "step": 4638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.929256261221361e-06,
+      "loss": 0.5937,
+      "step": 4639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9081215579197545e-06,
+      "loss": 0.69,
+      "step": 4640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.887023442354744e-06,
+      "loss": 0.7702,
+      "step": 4641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.865961922730379e-06,
+      "loss": 0.6262,
+      "step": 4642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.844937007236406e-06,
+      "loss": 0.6121,
+      "step": 4643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.823948704048443e-06,
+      "loss": 0.5728,
+      "step": 4644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8029970213277605e-06,
+      "loss": 0.6999,
+      "step": 4645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.782081967221498e-06,
+      "loss": 0.7118,
+      "step": 4646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.761203549862448e-06,
+      "loss": 0.7901,
+      "step": 4647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.740361777369241e-06,
+      "loss": 0.7152,
+      "step": 4648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.719556657846226e-06,
+      "loss": 0.6735,
+      "step": 4649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.698788199383476e-06,
+      "loss": 0.6799,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.678056410056854e-06,
+      "loss": 0.6299,
+      "step": 4651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.65736129792791e-06,
+      "loss": 0.6131,
+      "step": 4652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.636702871043986e-06,
+      "loss": 0.6811,
+      "step": 4653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.616081137438123e-06,
+      "loss": 0.7404,
+      "step": 4654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5954961051291384e-06,
+      "loss": 0.7152,
+      "step": 4655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.574947782121498e-06,
+      "loss": 0.8683,
+      "step": 4656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.554436176405475e-06,
+      "loss": 0.7343,
+      "step": 4657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.533961295957013e-06,
+      "loss": 0.7754,
+      "step": 4658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.513523148737809e-06,
+      "loss": 0.696,
+      "step": 4659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.493121742695251e-06,
+      "loss": 0.624,
+      "step": 4660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.472757085762459e-06,
+      "loss": 0.6814,
+      "step": 4661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.452429185858221e-06,
+      "loss": 0.6973,
+      "step": 4662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.432138050887114e-06,
+      "loss": 0.6829,
+      "step": 4663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4118836887393275e-06,
+      "loss": 0.739,
+      "step": 4664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.391666107290805e-06,
+      "loss": 0.7577,
+      "step": 4665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.371485314403202e-06,
+      "loss": 0.6734,
+      "step": 4666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.351341317923808e-06,
+      "loss": 0.7258,
+      "step": 4667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.331234125685669e-06,
+      "loss": 0.7037,
+      "step": 4668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3111637455074635e-06,
+      "loss": 0.6365,
+      "step": 4669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.291130185193593e-06,
+      "loss": 0.8323,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.271133452534149e-06,
+      "loss": 0.8164,
+      "step": 4671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.251173555304844e-06,
+      "loss": 0.6623,
+      "step": 4672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.231250501267159e-06,
+      "loss": 0.7444,
+      "step": 4673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2113642981681425e-06,
+      "loss": 0.6256,
+      "step": 4674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.191514953740606e-06,
+      "loss": 0.7148,
+      "step": 4675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.171702475702967e-06,
+      "loss": 0.6595,
+      "step": 4676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.151926871759349e-06,
+      "loss": 0.6697,
+      "step": 4677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.132188149599526e-06,
+      "loss": 0.6388,
+      "step": 4678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1124863168988905e-06,
+      "loss": 0.779,
+      "step": 4679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.092821381318557e-06,
+      "loss": 0.6807,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.073193350505223e-06,
+      "loss": 0.679,
+      "step": 4681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.053602232091303e-06,
+      "loss": 0.6522,
+      "step": 4682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.034048033694816e-06,
+      "loss": 0.7673,
+      "step": 4683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0145307629194225e-06,
+      "loss": 0.6825,
+      "step": 4684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.995050427354464e-06,
+      "loss": 0.6632,
+      "step": 4685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9756070345748586e-06,
+      "loss": 0.7328,
+      "step": 4686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.956200592141214e-06,
+      "loss": 0.6767,
+      "step": 4687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.936831107599749e-06,
+      "loss": 0.7027,
+      "step": 4688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.917498588482294e-06,
+      "loss": 0.7421,
+      "step": 4689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.898203042306338e-06,
+      "loss": 0.6511,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.878944476574965e-06,
+      "loss": 0.6528,
+      "step": 4691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.859722898776897e-06,
+      "loss": 0.6771,
+      "step": 4692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8405383163864555e-06,
+      "loss": 0.6119,
+      "step": 4693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.821390736863607e-06,
+      "loss": 0.681,
+      "step": 4694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8022801676539055e-06,
+      "loss": 0.6947,
+      "step": 4695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.783206616188496e-06,
+      "loss": 0.6915,
+      "step": 4696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.764170089884179e-06,
+      "loss": 0.6285,
+      "step": 4697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.745170596143312e-06,
+      "loss": 0.6665,
+      "step": 4698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7262081423538716e-06,
+      "loss": 0.7068,
+      "step": 4699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.707282735889462e-06,
+      "loss": 0.7227,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.688394384109207e-06,
+      "loss": 0.7044,
+      "step": 4701
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6695430943578864e-06,
+      "loss": 0.6637,
+      "step": 4702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.650728873965826e-06,
+      "loss": 0.7899,
+      "step": 4703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.631951730248995e-06,
+      "loss": 0.6989,
+      "step": 4704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6132116705088655e-06,
+      "loss": 0.7461,
+      "step": 4705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.594508702032552e-06,
+      "loss": 0.7127,
+      "step": 4706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.575842832092736e-06,
+      "loss": 0.7575,
+      "step": 4707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.557214067947646e-06,
+      "loss": 0.854,
+      "step": 4708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.53862241684111e-06,
+      "loss": 0.6622,
+      "step": 4709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5200678860024885e-06,
+      "loss": 0.6423,
+      "step": 4710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5015504826467545e-06,
+      "loss": 0.7939,
+      "step": 4711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.483070213974438e-06,
+      "loss": 0.8101,
+      "step": 4712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.464627087171569e-06,
+      "loss": 0.6743,
+      "step": 4713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.446221109409821e-06,
+      "loss": 0.6843,
+      "step": 4714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.427852287846346e-06,
+      "loss": 0.7459,
+      "step": 4715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4095206296238974e-06,
+      "loss": 0.7342,
+      "step": 4716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.391226141870786e-06,
+      "loss": 0.7803,
+      "step": 4717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3729688317007965e-06,
+      "loss": 0.6732,
+      "step": 4718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.35474870621333e-06,
+      "loss": 0.7719,
+      "step": 4719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.336565772493306e-06,
+      "loss": 0.8147,
+      "step": 4720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3184200376111815e-06,
+      "loss": 0.6512,
+      "step": 4721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.300311508622945e-06,
+      "loss": 0.6267,
+      "step": 4722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.282240192570098e-06,
+      "loss": 0.7315,
+      "step": 4723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.264206096479728e-06,
+      "loss": 0.7843,
+      "step": 4724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.246209227364395e-06,
+      "loss": 0.6356,
+      "step": 4725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2282495922222105e-06,
+      "loss": 0.7451,
+      "step": 4726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.210327198036779e-06,
+      "loss": 0.819,
+      "step": 4727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.192442051777257e-06,
+      "loss": 0.6838,
+      "step": 4728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.174594160398315e-06,
+      "loss": 0.7195,
+      "step": 4729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1567835308401135e-06,
+      "loss": 0.8879,
+      "step": 4730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.139010170028346e-06,
+      "loss": 0.7584,
+      "step": 4731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.121274084874194e-06,
+      "loss": 0.7281,
+      "step": 4732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.103575282274352e-06,
+      "loss": 0.626,
+      "step": 4733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.085913769111049e-06,
+      "loss": 0.7239,
+      "step": 4734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.068289552251947e-06,
+      "loss": 0.6373,
+      "step": 4735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.050702638550275e-06,
+      "loss": 0.7096,
+      "step": 4736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.033153034844695e-06,
+      "loss": 0.745,
+      "step": 4737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0156407479594036e-06,
+      "loss": 0.671,
+      "step": 4738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.998165784704078e-06,
+      "loss": 0.6345,
+      "step": 4739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9807281518738915e-06,
+      "loss": 0.6455,
+      "step": 4740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.963327856249466e-06,
+      "loss": 0.8007,
+      "step": 4741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.945964904596922e-06,
+      "loss": 0.6821,
+      "step": 4742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.928639303667891e-06,
+      "loss": 0.7164,
+      "step": 4743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.911351060199431e-06,
+      "loss": 0.706,
+      "step": 4744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.894100180914095e-06,
+      "loss": 0.71,
+      "step": 4745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.87688667251993e-06,
+      "loss": 0.7553,
+      "step": 4746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.859710541710393e-06,
+      "loss": 0.7315,
+      "step": 4747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.842571795164485e-06,
+      "loss": 0.7178,
+      "step": 4748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.825470439546585e-06,
+      "loss": 0.7324,
+      "step": 4749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.80840648150661e-06,
+      "loss": 0.7126,
+      "step": 4750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7913799276798832e-06,
+      "loss": 0.8538,
+      "step": 4751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7743907846871875e-06,
+      "loss": 0.7328,
+      "step": 4752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.757439059134804e-06,
+      "loss": 0.81,
+      "step": 4753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7405247576144054e-06,
+      "loss": 0.6876,
+      "step": 4754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.723647886703141e-06,
+      "loss": 0.6699,
+      "step": 4755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.706808452963606e-06,
+      "loss": 0.7726,
+      "step": 4756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.690006462943829e-06,
+      "loss": 0.7723,
+      "step": 4757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6732419231773064e-06,
+      "loss": 0.7721,
+      "step": 4758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6565148401829118e-06,
+      "loss": 0.6591,
+      "step": 4759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6398252204650095e-06,
+      "loss": 0.6265,
+      "step": 4760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.623173070513375e-06,
+      "loss": 0.6597,
+      "step": 4761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.606558396803228e-06,
+      "loss": 0.737,
+      "step": 4762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5899812057951897e-06,
+      "loss": 0.7086,
+      "step": 4763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5734415039353262e-06,
+      "loss": 0.6059,
+      "step": 4764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.556939297655115e-06,
+      "loss": 0.5915,
+      "step": 4765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.540474593371468e-06,
+      "loss": 0.7473,
+      "step": 4766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.524047397486696e-06,
+      "loss": 0.6748,
+      "step": 4767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.507657716388535e-06,
+      "loss": 0.657,
+      "step": 4768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.491305556450131e-06,
+      "loss": 0.6739,
+      "step": 4769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4749909240300416e-06,
+      "loss": 0.704,
+      "step": 4770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.458713825472226e-06,
+      "loss": 0.781,
+      "step": 4771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4424742671060773e-06,
+      "loss": 0.6045,
+      "step": 4772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.426272255246321e-06,
+      "loss": 0.7551,
+      "step": 4773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4101077961931626e-06,
+      "loss": 0.6623,
+      "step": 4774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.393980896232174e-06,
+      "loss": 0.6646,
+      "step": 4775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3778915616342943e-06,
+      "loss": 0.6336,
+      "step": 4776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.361839798655897e-06,
+      "loss": 0.7055,
+      "step": 4777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.345825613538722e-06,
+      "loss": 0.683,
+      "step": 4778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3298490125099115e-06,
+      "loss": 0.7576,
+      "step": 4779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3139100017819837e-06,
+      "loss": 0.7119,
+      "step": 4780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2980085875528253e-06,
+      "loss": 0.6553,
+      "step": 4781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.282144776005747e-06,
+      "loss": 0.8142,
+      "step": 4782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.266318573309379e-06,
+      "loss": 0.6649,
+      "step": 4783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2505299856178007e-06,
+      "loss": 0.7465,
+      "step": 4784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2347790190703776e-06,
+      "loss": 0.793,
+      "step": 4785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2190656797919215e-06,
+      "loss": 0.7096,
+      "step": 4786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.203389973892579e-06,
+      "loss": 0.6461,
+      "step": 4787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.187751907467851e-06,
+      "loss": 0.7756,
+      "step": 4788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.172151486598629e-06,
+      "loss": 0.6698,
+      "step": 4789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.15658871735115e-06,
+      "loss": 0.6089,
+      "step": 4790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.141063605777028e-06,
+      "loss": 0.75,
+      "step": 4791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1255761579132125e-06,
+      "loss": 0.7284,
+      "step": 4792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1101263797819967e-06,
+      "loss": 0.7513,
+      "step": 4793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0947142773910863e-06,
+      "loss": 0.7912,
+      "step": 4794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0793398567334544e-06,
+      "loss": 0.7286,
+      "step": 4795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0640031237874957e-06,
+      "loss": 0.7221,
+      "step": 4796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0487040845168846e-06,
+      "loss": 0.7194,
+      "step": 4797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0334427448706847e-06,
+      "loss": 0.7229,
+      "step": 4798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0182191107833046e-06,
+      "loss": 0.6886,
+      "step": 4799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.003033188174431e-06,
+      "loss": 0.6842,
+      "step": 4800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.987884982949163e-06,
+      "loss": 0.7663,
+      "step": 4801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9727745009978992e-06,
+      "loss": 0.6401,
+      "step": 4802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9577017481963398e-06,
+      "loss": 0.5122,
+      "step": 4803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.942666730405552e-06,
+      "loss": 0.6723,
+      "step": 4804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9276694534719375e-06,
+      "loss": 0.7421,
+      "step": 4805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.912709923227197e-06,
+      "loss": 0.7887,
+      "step": 4806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8977881454883447e-06,
+      "loss": 0.7818,
+      "step": 4807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8829041260577396e-06,
+      "loss": 0.6865,
+      "step": 4808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.868057870723073e-06,
+      "loss": 0.6803,
+      "step": 4809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.853249385257295e-06,
+      "loss": 0.8769,
+      "step": 4810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.838478675418732e-06,
+      "loss": 0.6835,
+      "step": 4811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8237457469509676e-06,
+      "loss": 0.6886,
+      "step": 4812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8090506055829304e-06,
+      "loss": 0.7038,
+      "step": 4813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.794393257028838e-06,
+      "loss": 0.6965,
+      "step": 4814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.77977370698822e-06,
+      "loss": 0.7251,
+      "step": 4815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7651919611459075e-06,
+      "loss": 0.7168,
+      "step": 4816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.750648025172031e-06,
+      "loss": 0.7091,
+      "step": 4817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7361419047220116e-06,
+      "loss": 0.6906,
+      "step": 4818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.72167360543657e-06,
+      "loss": 0.6622,
+      "step": 4819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.707243132941717e-06,
+      "loss": 0.6524,
+      "step": 4820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6928504928487754e-06,
+      "loss": 0.7242,
+      "step": 4821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6784956907543236e-06,
+      "loss": 0.6751,
+      "step": 4822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.664178732240252e-06,
+      "loss": 0.7739,
+      "step": 4823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6498996228737195e-06,
+      "loss": 0.7047,
+      "step": 4824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6356583682071946e-06,
+      "loss": 0.6585,
+      "step": 4825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.621454973778381e-06,
+      "loss": 0.6821,
+      "step": 4826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6072894451102834e-06,
+      "loss": 0.7717,
+      "step": 4827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.593161787711218e-06,
+      "loss": 0.6709,
+      "step": 4828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.579072007074701e-06,
+      "loss": 0.7315,
+      "step": 4829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5650201086795945e-06,
+      "loss": 0.6736,
+      "step": 4830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5510060979899607e-06,
+      "loss": 0.6979,
+      "step": 4831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5370299804551967e-06,
+      "loss": 0.642,
+      "step": 4832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.52309176150991e-06,
+      "loss": 0.7896,
+      "step": 4833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5091914465739996e-06,
+      "loss": 0.6813,
+      "step": 4834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.49532904105263e-06,
+      "loss": 0.7522,
+      "step": 4835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4815045503361887e-06,
+      "loss": 0.658,
+      "step": 4836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.467717979800355e-06,
+      "loss": 0.6344,
+      "step": 4837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.453969334806083e-06,
+      "loss": 0.681,
+      "step": 4838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.440258620699498e-06,
+      "loss": 0.6757,
+      "step": 4839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.426585842812079e-06,
+      "loss": 0.7073,
+      "step": 4840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.412951006460451e-06,
+      "loss": 0.6681,
+      "step": 4841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3993541169465837e-06,
+      "loss": 0.7453,
+      "step": 4842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.385795179557604e-06,
+      "loss": 0.6908,
+      "step": 4843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.372274199565938e-06,
+      "loss": 0.7047,
+      "step": 4844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3587911822292365e-06,
+      "loss": 0.5512,
+      "step": 4845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3453461327903824e-06,
+      "loss": 0.7685,
+      "step": 4846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.331939056477517e-06,
+      "loss": 0.8015,
+      "step": 4847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3185699585039577e-06,
+      "loss": 0.6925,
+      "step": 4848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.305238844068325e-06,
+      "loss": 0.8011,
+      "step": 4849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.291945718354438e-06,
+      "loss": 0.7865,
+      "step": 4850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.278690586531329e-06,
+      "loss": 0.6027,
+      "step": 4851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.265473453753286e-06,
+      "loss": 0.7427,
+      "step": 4852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2522943251597873e-06,
+      "loss": 0.7911,
+      "step": 4853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.239153205875577e-06,
+      "loss": 0.7168,
+      "step": 4854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2260501010105793e-06,
+      "loss": 0.7027,
+      "step": 4855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.212985015659941e-06,
+      "loss": 0.7257,
+      "step": 4856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.199957954904075e-06,
+      "loss": 0.6052,
+      "step": 4857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.186968923808519e-06,
+      "loss": 0.8216,
+      "step": 4858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.17401792742411e-06,
+      "loss": 0.6747,
+      "step": 4859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1611049707868423e-06,
+      "loss": 0.7535,
+      "step": 4860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1482300589179326e-06,
+      "loss": 0.7113,
+      "step": 4861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1353931968238207e-06,
+      "loss": 0.7766,
+      "step": 4862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1225943894961132e-06,
+      "loss": 0.7316,
+      "step": 4863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1098336419116625e-06,
+      "loss": 0.7406,
+      "step": 4864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.097110959032489e-06,
+      "loss": 0.7307,
+      "step": 4865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.084426345805812e-06,
+      "loss": 0.7134,
+      "step": 4866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.071779807164076e-06,
+      "loss": 0.5775,
+      "step": 4867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.059171348024891e-06,
+      "loss": 0.7273,
+      "step": 4868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0466009732910906e-06,
+      "loss": 0.8457,
+      "step": 4869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0340686878506656e-06,
+      "loss": 0.6337,
+      "step": 4870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.021574496576806e-06,
+      "loss": 0.581,
+      "step": 4871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.009118404327892e-06,
+      "loss": 0.7865,
+      "step": 4872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9967004159474947e-06,
+      "loss": 0.6533,
+      "step": 4873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9843205362643723e-06,
+      "loss": 0.7435,
+      "step": 4874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.971978770092431e-06,
+      "loss": 0.8292,
+      "step": 4875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9596751222308087e-06,
+      "loss": 0.6134,
+      "step": 4876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9474095974637672e-06,
+      "loss": 0.6641,
+      "step": 4877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.935182200560792e-06,
+      "loss": 0.7437,
+      "step": 4878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9229929362765133e-06,
+      "loss": 0.7593,
+      "step": 4879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9108418093507385e-06,
+      "loss": 0.6878,
+      "step": 4880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8987288245084555e-06,
+      "loss": 0.6644,
+      "step": 4881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8866539864597854e-06,
+      "loss": 0.7571,
+      "step": 4882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8746172999000722e-06,
+      "loss": 0.6874,
+      "step": 4883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8626187695097942e-06,
+      "loss": 0.6332,
+      "step": 4884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8506583999545747e-06,
+      "loss": 0.6528,
+      "step": 4885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.838736195885238e-06,
+      "loss": 0.6732,
+      "step": 4886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.826852161937731e-06,
+      "loss": 0.7959,
+      "step": 4887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8150063027331798e-06,
+      "loss": 0.7184,
+      "step": 4888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8031986228778662e-06,
+      "loss": 0.6901,
+      "step": 4889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7914291269632066e-06,
+      "loss": 0.6411,
+      "step": 4890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7796978195658065e-06,
+      "loss": 0.8413,
+      "step": 4891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7680047052473837e-06,
+      "loss": 0.7381,
+      "step": 4892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7563497885548119e-06,
+      "loss": 0.6429,
+      "step": 4893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7447330740201328e-06,
+      "loss": 0.6895,
+      "step": 4894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7331545661604997e-06,
+      "loss": 0.795,
+      "step": 4895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7216142694782667e-06,
+      "loss": 0.6676,
+      "step": 4896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.710112188460844e-06,
+      "loss": 0.6419,
+      "step": 4897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6986483275808761e-06,
+      "loss": 0.7916,
+      "step": 4898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6872226912960532e-06,
+      "loss": 0.6934,
+      "step": 4899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6758352840492876e-06,
+      "loss": 0.7991,
+      "step": 4900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6644861102685593e-06,
+      "loss": 0.7586,
+      "step": 4901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6531751743670055e-06,
+      "loss": 0.6824,
+      "step": 4902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6419024807429294e-06,
+      "loss": 0.7651,
+      "step": 4903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.630668033779692e-06,
+      "loss": 0.6709,
+      "step": 4904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.619471837845854e-06,
+      "loss": 0.641,
+      "step": 4905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6083138972950328e-06,
+      "loss": 0.7191,
+      "step": 4906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5971942164660357e-06,
+      "loss": 0.6721,
+      "step": 4907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5861127996827597e-06,
+      "loss": 0.7636,
+      "step": 4908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5750696512542241e-06,
+      "loss": 0.6595,
+      "step": 4909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5640647754745608e-06,
+      "loss": 0.7194,
+      "step": 4910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5530981766230357e-06,
+      "loss": 0.7395,
+      "step": 4911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5421698589640266e-06,
+      "loss": 0.72,
+      "step": 4912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.531279826747023e-06,
+      "loss": 0.6903,
+      "step": 4913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5204280842066266e-06,
+      "loss": 0.662,
+      "step": 4914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5096146355625395e-06,
+      "loss": 0.6903,
+      "step": 4915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4988394850195876e-06,
+      "loss": 0.6739,
+      "step": 4916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4881026367677187e-06,
+      "loss": 0.7043,
+      "step": 4917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4774040949819378e-06,
+      "loss": 0.7223,
+      "step": 4918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4667438638224062e-06,
+      "loss": 0.7306,
+      "step": 4919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4561219474343746e-06,
+      "loss": 0.6416,
+      "step": 4920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4455383499481611e-06,
+      "loss": 0.692,
+      "step": 4921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4349930754792407e-06,
+      "loss": 0.7316,
+      "step": 4922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.424486128128122e-06,
+      "loss": 0.7947,
+      "step": 4923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4140175119804699e-06,
+      "loss": 0.7058,
+      "step": 4924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4035872311070175e-06,
+      "loss": 0.7612,
+      "step": 4925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3931952895635647e-06,
+      "loss": 0.6916,
+      "step": 4926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.382841691391068e-06,
+      "loss": 0.8498,
+      "step": 4927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.372526440615507e-06,
+      "loss": 0.7191,
+      "step": 4928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.362249541247984e-06,
+      "loss": 0.6205,
+      "step": 4929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3520109972846917e-06,
+      "loss": 0.7489,
+      "step": 4930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3418108127069007e-06,
+      "loss": 0.6606,
+      "step": 4931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3316489914809604e-06,
+      "loss": 0.6218,
+      "step": 4932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.321525537558299e-06,
+      "loss": 0.6431,
+      "step": 4933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3114404548754567e-06,
+      "loss": 0.6425,
+      "step": 4934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3013937473540073e-06,
+      "loss": 0.7666,
+      "step": 4935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2913854189006259e-06,
+      "loss": 0.8361,
+      "step": 4936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.281415473407077e-06,
+      "loss": 0.7254,
+      "step": 4937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2714839147501712e-06,
+      "loss": 0.8312,
+      "step": 4938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2615907467918187e-06,
+      "loss": 0.7113,
+      "step": 4939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2517359733789758e-06,
+      "loss": 0.7364,
+      "step": 4940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2419195983436881e-06,
+      "loss": 0.678,
+      "step": 4941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2321416255030693e-06,
+      "loss": 0.6938,
+      "step": 4942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2224020586593e-06,
+      "loss": 0.7291,
+      "step": 4943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.212700901599606e-06,
+      "loss": 0.7444,
+      "step": 4944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2030381580963035e-06,
+      "loss": 0.7255,
+      "step": 4945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1934138319067644e-06,
+      "loss": 0.6408,
+      "step": 4946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1838279267734176e-06,
+      "loss": 0.7195,
+      "step": 4947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1742804464237366e-06,
+      "loss": 0.6666,
+      "step": 4948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1647713945702965e-06,
+      "loss": 0.8708,
+      "step": 4949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1553007749106837e-06,
+      "loss": 0.7121,
+      "step": 4950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1458685911275746e-06,
+      "loss": 0.6357,
+      "step": 4951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1364748468886687e-06,
+      "loss": 0.623,
+      "step": 4952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1271195458467554e-06,
+      "loss": 0.6013,
+      "step": 4953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1178026916396355e-06,
+      "loss": 0.8228,
+      "step": 4954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1085242878901892e-06,
+      "loss": 0.6753,
+      "step": 4955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.09928433820633e-06,
+      "loss": 0.7181,
+      "step": 4956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0900828461810175e-06,
+      "loss": 0.7119,
+      "step": 4957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0809198153922672e-06,
+      "loss": 0.7685,
+      "step": 4958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0717952494031513e-06,
+      "loss": 0.7173,
+      "step": 4959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0627091517617427e-06,
+      "loss": 0.6586,
+      "step": 4960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0536615260011928e-06,
+      "loss": 0.7547,
+      "step": 4961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0446523756396764e-06,
+      "loss": 0.6672,
+      "step": 4962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0356817041804246e-06,
+      "loss": 0.771,
+      "step": 4963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.026749515111669e-06,
+      "loss": 0.8035,
+      "step": 4964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0178558119067315e-06,
+      "loss": 0.6581,
+      "step": 4965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0090005980239236e-06,
+      "loss": 0.7257,
+      "step": 4966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0001838769065907e-06,
+      "loss": 0.6182,
+      "step": 4967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.91405651983157e-07,
+      "loss": 0.7257,
+      "step": 4968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.826659266670257e-07,
+      "loss": 0.5725,
+      "step": 4969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.739647043566447e-07,
+      "loss": 0.6788,
+      "step": 4970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.653019884355074e-07,
+      "loss": 0.7343,
+      "step": 4971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.566777822721197e-07,
+      "loss": 0.6975,
+      "step": 4972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.480920892200096e-07,
+      "loss": 0.6478,
+      "step": 4973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.395449126177291e-07,
+      "loss": 0.6539,
+      "step": 4974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.310362557888641e-07,
+      "loss": 0.6931,
+      "step": 4975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.225661220420234e-07,
+      "loss": 0.7782,
+      "step": 4976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.141345146708169e-07,
+      "loss": 0.6752,
+      "step": 4977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.057414369538886e-07,
+      "loss": 0.7676,
+      "step": 4978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.973868921548833e-07,
+      "loss": 0.6726,
+      "step": 4979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.890708835225026e-07,
+      "loss": 0.5958,
+      "step": 4980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.807934142904261e-07,
+      "loss": 0.6008,
+      "step": 4981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.725544876773351e-07,
+      "loss": 0.6764,
+      "step": 4982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.643541068869776e-07,
+      "loss": 0.7443,
+      "step": 4983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.561922751080697e-07,
+      "loss": 0.6972,
+      "step": 4984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.480689955143395e-07,
+      "loss": 0.7731,
+      "step": 4985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.399842712645378e-07,
+      "loss": 0.8191,
+      "step": 4986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.319381055024278e-07,
+      "loss": 0.6816,
+      "step": 4987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.239305013567733e-07,
+      "loss": 0.6376,
+      "step": 4988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.15961461941317e-07,
+      "loss": 0.5943,
+      "step": 4989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.080309903548578e-07,
+      "loss": 0.6835,
+      "step": 4990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.001390896811401e-07,
+      "loss": 0.5595,
+      "step": 4991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.922857629889646e-07,
+      "loss": 0.8278,
+      "step": 4992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.844710133320887e-07,
+      "loss": 0.6763,
+      "step": 4993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.766948437493038e-07,
+      "loss": 0.6728,
+      "step": 4994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.689572572643689e-07,
+      "loss": 0.769,
+      "step": 4995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.612582568860549e-07,
+      "loss": 0.5745,
+      "step": 4996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.535978456081339e-07,
+      "loss": 0.7693,
+      "step": 4997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.459760264093563e-07,
+      "loss": 0.7694,
+      "step": 4998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.383928022534847e-07,
+      "loss": 0.6857,
+      "step": 4999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.308481760892604e-07,
+      "loss": 0.6263,
+      "step": 5000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.233421508504257e-07,
+      "loss": 0.7458,
+      "step": 5001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.1587472945569e-07,
+      "loss": 0.7192,
+      "step": 5002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.084459148087863e-07,
+      "loss": 0.6716,
+      "step": 5003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.01055709798415e-07,
+      "loss": 0.7559,
+      "step": 5004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.937041172982662e-07,
+      "loss": 0.6589,
+      "step": 5005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.863911401669976e-07,
+      "loss": 0.7137,
+      "step": 5006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.791167812483012e-07,
+      "loss": 0.7062,
+      "step": 5007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.718810433707923e-07,
+      "loss": 0.6686,
+      "step": 5008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.646839293481088e-07,
+      "loss": 0.7489,
+      "step": 5009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.575254419788457e-07,
+      "loss": 0.6262,
+      "step": 5010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.504055840465983e-07,
+      "loss": 0.8246,
+      "step": 5011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.433243583199411e-07,
+      "loss": 0.6648,
+      "step": 5012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.362817675523936e-07,
+      "loss": 0.7109,
+      "step": 5013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.292778144824985e-07,
+      "loss": 0.6541,
+      "step": 5014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.223125018337328e-07,
+      "loss": 0.8137,
+      "step": 5015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.153858323145855e-07,
+      "loss": 0.6971,
+      "step": 5016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.0849780861848e-07,
+      "loss": 0.7144,
+      "step": 5017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.016484334238515e-07,
+      "loss": 0.6744,
+      "step": 5018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.948377093940694e-07,
+      "loss": 0.6366,
+      "step": 5019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.880656391775041e-07,
+      "loss": 0.7121,
+      "step": 5020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.813322254074826e-07,
+      "loss": 0.7144,
+      "step": 5021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.746374707022883e-07,
+      "loss": 0.7272,
+      "step": 5022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.679813776651943e-07,
+      "loss": 0.6207,
+      "step": 5023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.613639488844302e-07,
+      "loss": 0.5906,
+      "step": 5024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.547851869331933e-07,
+      "loss": 0.7211,
+      "step": 5025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.482450943696371e-07,
+      "loss": 0.7304,
+      "step": 5026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.41743673736883e-07,
+      "loss": 0.6475,
+      "step": 5027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3528092756302e-07,
+      "loss": 0.7088,
+      "step": 5028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.288568583610931e-07,
+      "loss": 0.7415,
+      "step": 5029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.224714686291154e-07,
+      "loss": 0.7102,
+      "step": 5030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.161247608500453e-07,
+      "loss": 0.7485,
+      "step": 5031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.098167374917973e-07,
+      "loss": 0.6142,
+      "step": 5032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.03547401007276e-07,
+      "loss": 0.7141,
+      "step": 5033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.973167538343204e-07,
+      "loss": 0.7605,
+      "step": 5034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.911247983957035e-07,
+      "loss": 0.7223,
+      "step": 5035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.84971537099177e-07,
+      "loss": 0.6414,
+      "step": 5036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.788569723374603e-07,
+      "loss": 0.7081,
+      "step": 5037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7278110648820704e-07,
+      "loss": 0.7285,
+      "step": 5038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6674394191401627e-07,
+      "loss": 0.6514,
+      "step": 5039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.607454809624434e-07,
+      "loss": 0.7417,
+      "step": 5040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5478572596601155e-07,
+      "loss": 0.7751,
+      "step": 5041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.488646792421558e-07,
+      "loss": 0.7446,
+      "step": 5042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.429823430933122e-07,
+      "loss": 0.5928,
+      "step": 5043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3713871980680663e-07,
+      "loss": 0.738,
+      "step": 5044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.313338116549437e-07,
+      "loss": 0.7509,
+      "step": 5045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2556762089498434e-07,
+      "loss": 0.706,
+      "step": 5046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.198401497691129e-07,
+      "loss": 0.7179,
+      "step": 5047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1415140050445887e-07,
+      "loss": 0.6488,
+      "step": 5048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.085013753130862e-07,
+      "loss": 0.6665,
+      "step": 5049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0289007639203733e-07,
+      "loss": 0.6923,
+      "step": 5050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9731750592325587e-07,
+      "loss": 0.7511,
+      "step": 5051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.917836660736529e-07,
+      "loss": 0.6539,
+      "step": 5052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8628855899506266e-07,
+      "loss": 0.6316,
+      "step": 5053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.808321868242648e-07,
+      "loss": 0.6262,
+      "step": 5054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7541455168297324e-07,
+      "loss": 0.702,
+      "step": 5055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7003565567784723e-07,
+      "loss": 0.657,
+      "step": 5056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.646955009004804e-07,
+      "loss": 0.613,
+      "step": 5057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.593940894273895e-07,
+      "loss": 0.6142,
+      "step": 5058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.541314233200255e-07,
+      "loss": 0.812,
+      "step": 5059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4890750462480694e-07,
+      "loss": 0.8404,
+      "step": 5060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.437223353730423e-07,
+      "loss": 0.753,
+      "step": 5061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385759175809966e-07,
+      "loss": 0.6189,
+      "step": 5062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3346825324986895e-07,
+      "loss": 0.6748,
+      "step": 5063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.283993443657596e-07,
+      "loss": 0.7378,
+      "step": 5064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2336919289973623e-07,
+      "loss": 0.7479,
+      "step": 5065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1837780080777867e-07,
+      "loss": 0.7191,
+      "step": 5066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1342517003079e-07,
+      "loss": 0.7644,
+      "step": 5067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.085113024946074e-07,
+      "loss": 0.7483,
+      "step": 5068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0363620010999127e-07,
+      "loss": 0.6447,
+      "step": 5069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9879986477263644e-07,
+      "loss": 0.7451,
+      "step": 5070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9400229836314964e-07,
+      "loss": 0.68,
+      "step": 5071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.89243502747083e-07,
+      "loss": 0.5986,
+      "step": 5072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.845234797748897e-07,
+      "loss": 0.7624,
+      "step": 5073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.798422312819571e-07,
+      "loss": 0.7112,
+      "step": 5074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7519975908859573e-07,
+      "loss": 0.7222,
+      "step": 5075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.705960650000394e-07,
+      "loss": 0.6296,
+      "step": 5076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.660311508064339e-07,
+      "loss": 0.79,
+      "step": 5077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6150501828284823e-07,
+      "loss": 0.6593,
+      "step": 5078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.570176691892967e-07,
+      "loss": 0.6868,
+      "step": 5079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5256910527066136e-07,
+      "loss": 0.5944,
+      "step": 5080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4815932825679177e-07,
+      "loss": 0.741,
+      "step": 5081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4378833986242745e-07,
+      "loss": 0.7701,
+      "step": 5082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3945614178723097e-07,
+      "loss": 0.7461,
+      "step": 5083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3516273571577708e-07,
+      "loss": 0.6462,
+      "step": 5084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3090812331757472e-07,
+      "loss": 0.7521,
+      "step": 5085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2669230624702276e-07,
+      "loss": 0.6539,
+      "step": 5086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.225152861434654e-07,
+      "loss": 0.8406,
+      "step": 5087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1837706463111453e-07,
+      "loss": 0.7415,
+      "step": 5088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1427764331914956e-07,
+      "loss": 0.7603,
+      "step": 5089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1021702380161767e-07,
+      "loss": 0.7416,
+      "step": 5090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0619520765750023e-07,
+      "loss": 0.7544,
+      "step": 5091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0221219645067956e-07,
+      "loss": 0.8173,
+      "step": 5092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9826799172996125e-07,
+      "loss": 0.6947,
+      "step": 5093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9436259502906286e-07,
+      "loss": 0.7404,
+      "step": 5094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9049600786658073e-07,
+      "loss": 0.5702,
+      "step": 5095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8666823174605662e-07,
+      "loss": 0.7712,
+      "step": 5096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8287926815592217e-07,
+      "loss": 0.7165,
+      "step": 5097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7912911856952097e-07,
+      "loss": 0.7851,
+      "step": 5098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.754177844450977e-07,
+      "loss": 0.7238,
+      "step": 5099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7174526722582017e-07,
+      "loss": 0.6602,
+      "step": 5100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6811156833974606e-07,
+      "loss": 0.8207,
+      "step": 5101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6451668919983397e-07,
+      "loss": 0.8075,
+      "step": 5102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6096063120396576e-07,
+      "loss": 0.6682,
+      "step": 5103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5744339573492416e-07,
+      "loss": 0.7943,
+      "step": 5104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5396498416038186e-07,
+      "loss": 0.6527,
+      "step": 5105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.505253978329235e-07,
+      "loss": 0.6634,
+      "step": 5106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4712463809004595e-07,
+      "loss": 0.829,
+      "step": 5107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.437627062541136e-07,
+      "loss": 0.7806,
+      "step": 5108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4043960363244734e-07,
+      "loss": 0.7406,
+      "step": 5109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.371553315172247e-07,
+      "loss": 0.6266,
+      "step": 5110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3390989118554631e-07,
+      "loss": 0.7593,
+      "step": 5111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3070328389939157e-07,
+      "loss": 0.6609,
+      "step": 5112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2753551090566307e-07,
+      "loss": 0.6246,
+      "step": 5113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2440657343615325e-07,
+      "loss": 0.6596,
+      "step": 5114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2131647270754443e-07,
+      "loss": 0.7528,
+      "step": 5115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1826520992144208e-07,
+      "loss": 0.6048,
+      "step": 5116
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1525278626431934e-07,
+      "loss": 0.7873,
+      "step": 5117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1227920290757255e-07,
+      "loss": 0.6963,
+      "step": 5118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0934446100746565e-07,
+      "loss": 0.7097,
+      "step": 5119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0644856170519691e-07,
+      "loss": 0.6475,
+      "step": 5120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0359150612682111e-07,
+      "loss": 0.7177,
+      "step": 5121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0077329538331626e-07,
+      "loss": 0.656,
+      "step": 5122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.799393057056128e-08,
+      "loss": 0.7042,
+      "step": 5123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.525341276929167e-08,
+      "loss": 0.5768,
+      "step": 5124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.255174304516611e-08,
+      "loss": 0.6305,
+      "step": 5125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.988892244874425e-08,
+      "loss": 0.6819,
+      "step": 5126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.726495201545337e-08,
+      "loss": 0.7135,
+      "step": 5127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.467983276563284e-08,
+      "loss": 0.704,
+      "step": 5128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.213356570452303e-08,
+      "loss": 0.7525,
+      "step": 5129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.962615182222078e-08,
+      "loss": 0.7839,
+      "step": 5130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.715759209373507e-08,
+      "loss": 0.7245,
+      "step": 5131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.472788747899806e-08,
+      "loss": 0.7309,
+      "step": 5132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.233703892277621e-08,
+      "loss": 0.7702,
+      "step": 5133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.99850473547592e-08,
+      "loss": 0.6101,
+      "step": 5134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.767191368952652e-08,
+      "loss": 0.6758,
+      "step": 5135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.53976388265365e-08,
+      "loss": 0.6647,
+      "step": 5136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.316222365014834e-08,
+      "loss": 0.6436,
+      "step": 5137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.096566902961121e-08,
+      "loss": 0.6845,
+      "step": 5138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.880797581904185e-08,
+      "loss": 0.6831,
+      "step": 5139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6689144857480224e-08,
+      "loss": 0.6935,
+      "step": 5140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.460917696882284e-08,
+      "loss": 0.7337,
+      "step": 5141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.256807296187827e-08,
+      "loss": 0.635,
+      "step": 5142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0565833630322745e-08,
+      "loss": 0.6249,
+      "step": 5143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8602459752733475e-08,
+      "loss": 0.7546,
+      "step": 5144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6677952092577525e-08,
+      "loss": 0.743,
+      "step": 5145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.479231139817852e-08,
+      "loss": 0.628,
+      "step": 5146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.294553840279436e-08,
+      "loss": 0.7193,
+      "step": 5147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1137633824539504e-08,
+      "loss": 0.5964,
+      "step": 5148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9368598366418265e-08,
+      "loss": 0.7206,
+      "step": 5149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.763843271631373e-08,
+      "loss": 0.6846,
+      "step": 5150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.594713754702106e-08,
+      "loss": 0.7856,
+      "step": 5151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4294713516180854e-08,
+      "loss": 0.8482,
+      "step": 5152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.26811612663569e-08,
+      "loss": 0.7275,
+      "step": 5153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1106481424969524e-08,
+      "loss": 0.7019,
+      "step": 5154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9570674604340044e-08,
+      "loss": 0.8307,
+      "step": 5155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.807374140166852e-08,
+      "loss": 0.7112,
+      "step": 5156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6615682399044884e-08,
+      "loss": 0.6542,
+      "step": 5157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5196498163426728e-08,
+      "loss": 0.6976,
+      "step": 5158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3816189246672614e-08,
+      "loss": 0.6327,
+      "step": 5159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.247475618550876e-08,
+      "loss": 0.5608,
+      "step": 5160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1172199501573455e-08,
+      "loss": 0.7014,
+      "step": 5161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9908519701339335e-08,
+      "loss": 0.8289,
+      "step": 5162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8683717276202217e-08,
+      "loss": 0.7189,
+      "step": 5163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7497792702436676e-08,
+      "loss": 0.6384,
+      "step": 5164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.635074644117385e-08,
+      "loss": 0.6847,
+      "step": 5165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5242578938468034e-08,
+      "loss": 0.6798,
+      "step": 5166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4173290625207891e-08,
+      "loss": 0.7009,
+      "step": 5167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.314288191719415e-08,
+      "loss": 0.649,
+      "step": 5168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2151353215106297e-08,
+      "loss": 0.7645,
+      "step": 5169
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1198704904502589e-08,
+      "loss": 0.7015,
+      "step": 5170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0284937355820034e-08,
+      "loss": 0.6834,
+      "step": 5171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.410050924374415e-09,
+      "loss": 0.7539,
+      "step": 5172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.574045950360265e-09,
+      "loss": 0.7993,
+      "step": 5173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.776922758884198e-09,
+      "loss": 0.6426,
+      "step": 5174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.018681659876069e-09,
+      "loss": 0.709,
+      "step": 5175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.299322948211117e-09,
+      "loss": 0.6858,
+      "step": 5176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.6188469035767246e-09,
+      "loss": 0.661,
+      "step": 5177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.977253790605651e-09,
+      "loss": 0.6505,
+      "step": 5178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.374543858776115e-09,
+      "loss": 0.6501,
+      "step": 5179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.810717342445092e-09,
+      "loss": 0.6607,
+      "step": 5180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2857744608705277e-09,
+      "loss": 0.7468,
+      "step": 5181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7997154181558238e-09,
+      "loss": 0.8648,
+      "step": 5182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3525404033275523e-09,
+      "loss": 0.7716,
+      "step": 5183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9442495902577406e-09,
+      "loss": 0.68,
+      "step": 5184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5748431377082817e-09,
+      "loss": 0.5859,
+      "step": 5185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2443211893309325e-09,
+      "loss": 0.6092,
+      "step": 5186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.526838736451105e-10,
+      "loss": 0.7156,
+      "step": 5187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.999313040600974e-10,
+      "loss": 0.7838,
+      "step": 5188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.860635788528356e-10,
+      "loss": 0.6527,
+      "step": 5189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1108078119013173e-10,
+      "loss": 0.8096,
+      "step": 5190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.749829791064528e-10,
+      "loss": 0.6148,
+      "step": 5191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.777022553723256e-11,
+      "loss": 0.7854,
+      "step": 5192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.944255827446284e-11,
+      "loss": 0.8104,
+      "step": 5193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.7176,
+      "step": 5194
+    },
+    {
+      "epoch": 1.0,
+      "step": 5194,
+      "total_flos": 5932410235191296.0,
+      "train_loss": 0.7643072148630791,
+      "train_runtime": 171640.7528,
+      "train_samples_per_second": 3.874,
+      "train_steps_per_second": 0.03
+    }
+  ],
+  "max_steps": 5194,
+  "num_train_epochs": 1,
+  "total_flos": 5932410235191296.0,
+  "trial_name": null,
+  "trial_params": null
+}