diff --git "a/llava-v1.5-7b/trainer_state.json" "b/llava-v1.5-7b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/llava-v1.5-7b/trainer_state.json"
@@ -0,0 +1,31165 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 5190,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-07,
+      "loss": 1.3564,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-07,
+      "loss": 1.4004,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 1.4189,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 1.4346,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.41025641025641e-07,
+      "loss": 1.3545,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 1.3447,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.974358974358975e-07,
+      "loss": 1.4473,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 1.3574,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 1.3594,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 1.3018,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4102564102564104e-06,
+      "loss": 1.335,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.3223,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.2129,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.794871794871795e-06,
+      "loss": 1.2373,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 1.2998,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 1.2598,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1794871794871797e-06,
+      "loss": 1.249,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.2207,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.435897435897436e-06,
+      "loss": 1.1279,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 1.1465,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 1.1099,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8205128205128207e-06,
+      "loss": 1.1787,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.948717948717949e-06,
+      "loss": 1.0884,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.0801,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.205128205128206e-06,
+      "loss": 1.1631,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.1348,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 1.1084,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 1.0962,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7179487179487184e-06,
+      "loss": 1.0679,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.1152,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.974358974358974e-06,
+      "loss": 1.0151,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 1.1099,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 1.0693,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.358974358974359e-06,
+      "loss": 1.0796,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.487179487179488e-06,
+      "loss": 1.125,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.0771,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.743589743589744e-06,
+      "loss": 1.0898,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 1.0205,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 1.0947,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 1.0601,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.256410256410257e-06,
+      "loss": 1.0864,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 1.1055,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.512820512820514e-06,
+      "loss": 1.0249,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 1.0342,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 1.0435,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.897435897435898e-06,
+      "loss": 1.0508,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.025641025641026e-06,
+      "loss": 1.0449,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 0.9351,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.282051282051282e-06,
+      "loss": 1.0103,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.410256410256412e-06,
+      "loss": 1.042,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 0.4719,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.0537,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.794871794871796e-06,
+      "loss": 1.0127,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 0.9858,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.051282051282053e-06,
+      "loss": 0.9863,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.17948717948718e-06,
+      "loss": 0.9463,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 1.0078,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.435897435897437e-06,
+      "loss": 0.979,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.564102564102564e-06,
+      "loss": 1.0059,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.9551,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.820512820512822e-06,
+      "loss": 0.4463,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.948717948717949e-06,
+      "loss": 0.9756,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 1.0229,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 1.0293,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.9502,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 0.4026,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.58974358974359e-06,
+      "loss": 1.001,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 1.0405,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 1.0029,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 0.9775,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.102564102564104e-06,
+      "loss": 1.0107,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.9927,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.358974358974359e-06,
+      "loss": 0.9883,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.487179487179487e-06,
+      "loss": 1.0039,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 1.0039,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 0.9722,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.871794871794872e-06,
+      "loss": 1.019,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 1.0029,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.012820512820513e-05,
+      "loss": 0.9766,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 1.019,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 0.9805,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0512820512820514e-05,
+      "loss": 0.9814,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0641025641025643e-05,
+      "loss": 1.0176,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.9761,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0897435897435898e-05,
+      "loss": 0.9937,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1025641025641028e-05,
+      "loss": 0.4253,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 0.9771,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 0.9683,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1410256410256411e-05,
+      "loss": 0.9346,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 0.9575,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.9414,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1794871794871796e-05,
+      "loss": 0.9023,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 1.0107,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2051282051282051e-05,
+      "loss": 0.9321,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.217948717948718e-05,
+      "loss": 0.9917,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.9507,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2435897435897436e-05,
+      "loss": 0.959,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2564102564102565e-05,
+      "loss": 0.9175,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 0.4419,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2820512820512823e-05,
+      "loss": 0.4094,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.294871794871795e-05,
+      "loss": 0.9824,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 0.9688,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3205128205128207e-05,
+      "loss": 0.9917,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.9229,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3461538461538463e-05,
+      "loss": 0.98,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3589743589743592e-05,
+      "loss": 0.9922,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3717948717948718e-05,
+      "loss": 1.0381,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.9834,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3974358974358975e-05,
+      "loss": 0.4187,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4102564102564105e-05,
+      "loss": 1.0137,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4230769230769232e-05,
+      "loss": 0.938,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.435897435897436e-05,
+      "loss": 0.9463,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4487179487179489e-05,
+      "loss": 0.9468,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 0.9072,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4743589743589745e-05,
+      "loss": 0.978,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4871794871794874e-05,
+      "loss": 0.9434,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.9287,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5128205128205129e-05,
+      "loss": 0.9619,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5256410256410257e-05,
+      "loss": 0.9849,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.9531,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5512820512820516e-05,
+      "loss": 0.9404,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5641025641025644e-05,
+      "loss": 0.938,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 0.9673,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 0.4441,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.602564102564103e-05,
+      "loss": 0.9224,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 0.9058,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6282051282051282e-05,
+      "loss": 0.9824,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 0.9971,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 0.9795,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.9341,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6794871794871796e-05,
+      "loss": 0.9023,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 1.0083,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7051282051282053e-05,
+      "loss": 0.9736,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.717948717948718e-05,
+      "loss": 1.001,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 0.4573,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7435897435897438e-05,
+      "loss": 0.9902,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7564102564102566e-05,
+      "loss": 0.9438,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 0.915,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7820512820512823e-05,
+      "loss": 0.9731,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 0.9146,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.807692307692308e-05,
+      "loss": 1.0713,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8205128205128208e-05,
+      "loss": 1.0186,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.9009,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.9634,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8589743589743593e-05,
+      "loss": 1.0161,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8717948717948718e-05,
+      "loss": 0.9453,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 0.9521,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 1.0254,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9102564102564106e-05,
+      "loss": 0.9326,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 0.9399,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.935897435897436e-05,
+      "loss": 0.9849,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 0.5337,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 1.021,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9743589743589745e-05,
+      "loss": 1.0161,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9871794871794873e-05,
+      "loss": 0.9487,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.9268,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999805265315e-05,
+      "loss": 0.8857,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999992210613346e-05,
+      "loss": 0.4536,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998247388287e-05,
+      "loss": 0.9302,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999968842465513e-05,
+      "loss": 1.0166,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999951316366583e-05,
+      "loss": 0.9785,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999929895592906e-05,
+      "loss": 0.8994,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999904580152827e-05,
+      "loss": 0.9595,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999875370056202e-05,
+      "loss": 0.9976,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999984226531441e-05,
+      "loss": 0.9927,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999805265940347e-05,
+      "loss": 0.957,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999764371948416e-05,
+      "loss": 0.9546,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999719583354548e-05,
+      "loss": 0.8926,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999670900176186e-05,
+      "loss": 0.9771,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999961832243229e-05,
+      "loss": 0.9888,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999956185014334e-05,
+      "loss": 0.9053,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999501483331328e-05,
+      "loss": 0.8921,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999437222019768e-05,
+      "loss": 0.9424,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999369066233686e-05,
+      "loss": 0.9355,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999929701599962e-05,
+      "loss": 0.9502,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999221071345643e-05,
+      "loss": 0.9414,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999914123230133e-05,
+      "loss": 0.9731,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999057498897766e-05,
+      "loss": 0.9922,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998969871167574e-05,
+      "loss": 0.8765,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998878349144883e-05,
+      "loss": 0.9238,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998782932865327e-05,
+      "loss": 0.9243,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998683622366073e-05,
+      "loss": 0.9111,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998580417685804e-05,
+      "loss": 0.9404,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998473318864714e-05,
+      "loss": 1.0269,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998362325944506e-05,
+      "loss": 0.9082,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998247438968418e-05,
+      "loss": 0.8823,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999812865798119e-05,
+      "loss": 0.9761,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999800598302909e-05,
+      "loss": 0.9775,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997879414159886e-05,
+      "loss": 0.9849,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997748951422878e-05,
+      "loss": 0.9214,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997614594868882e-05,
+      "loss": 0.916,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999747634455022e-05,
+      "loss": 0.978,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997334200520734e-05,
+      "loss": 0.9614,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999718816283579e-05,
+      "loss": 0.8862,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999703823155226e-05,
+      "loss": 0.9526,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996884406728543e-05,
+      "loss": 0.9819,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999672668842455e-05,
+      "loss": 0.9619,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996565076701698e-05,
+      "loss": 0.9287,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996399571622944e-05,
+      "loss": 0.896,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996230173252734e-05,
+      "loss": 0.957,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996056881657048e-05,
+      "loss": 0.979,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999587969690338e-05,
+      "loss": 0.9624,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995698619060738e-05,
+      "loss": 0.9575,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995513648199645e-05,
+      "loss": 0.9463,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999532478439214e-05,
+      "loss": 0.9937,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995132027711783e-05,
+      "loss": 0.4382,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999493537823364e-05,
+      "loss": 0.8877,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999473483603431e-05,
+      "loss": 0.9746,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999453040119189e-05,
+      "loss": 0.957,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994322073786002e-05,
+      "loss": 0.9326,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999410985389779e-05,
+      "loss": 0.9658,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99938937416099e-05,
+      "loss": 0.8999,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993673737006506e-05,
+      "loss": 0.9453,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993449840173286e-05,
+      "loss": 0.8979,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993222051197447e-05,
+      "loss": 0.917,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992990370167705e-05,
+      "loss": 0.9702,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999275479717429e-05,
+      "loss": 0.9248,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992515332308955e-05,
+      "loss": 0.9604,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992271975664962e-05,
+      "loss": 0.834,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999202472733709e-05,
+      "loss": 0.936,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999177358742163e-05,
+      "loss": 0.9814,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991518556016404e-05,
+      "loss": 0.9717,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999125963322073e-05,
+      "loss": 0.979,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990996819135454e-05,
+      "loss": 0.9268,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999073011386294e-05,
+      "loss": 0.9692,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990459517507046e-05,
+      "loss": 0.8979,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999018503017318e-05,
+      "loss": 0.9351,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989906651968232e-05,
+      "loss": 0.9521,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989624383000626e-05,
+      "loss": 0.9854,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99893382233803e-05,
+      "loss": 0.9502,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989048173218698e-05,
+      "loss": 0.9385,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988754232628795e-05,
+      "loss": 0.9258,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988456401725063e-05,
+      "loss": 0.9121,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988154680623503e-05,
+      "loss": 0.9395,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987849069441627e-05,
+      "loss": 0.981,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987539568298458e-05,
+      "loss": 0.9443,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987226177314535e-05,
+      "loss": 0.9536,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986908896611922e-05,
+      "loss": 0.9512,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986587726314185e-05,
+      "loss": 0.9312,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986262666546406e-05,
+      "loss": 0.9697,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985933717435193e-05,
+      "loss": 0.9546,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985600879108663e-05,
+      "loss": 1.0059,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985264151696436e-05,
+      "loss": 0.9395,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984923535329666e-05,
+      "loss": 0.9224,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998457903014101e-05,
+      "loss": 0.8691,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998423063626464e-05,
+      "loss": 1.0171,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983878353836254e-05,
+      "loss": 0.9209,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998352218299304e-05,
+      "loss": 0.9907,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998316212387373e-05,
+      "loss": 0.9487,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982798176618545e-05,
+      "loss": 0.9507,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982430341369235e-05,
+      "loss": 0.9263,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982058618269065e-05,
+      "loss": 0.9141,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981683007462806e-05,
+      "loss": 0.9683,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998130350909675e-05,
+      "loss": 0.9141,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980920123318692e-05,
+      "loss": 0.8994,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980532850277958e-05,
+      "loss": 0.918,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980141690125373e-05,
+      "loss": 0.9097,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979746643013285e-05,
+      "loss": 0.9365,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979347709095555e-05,
+      "loss": 0.9697,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978944888527546e-05,
+      "loss": 0.96,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978538181466156e-05,
+      "loss": 0.9365,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997812758806978e-05,
+      "loss": 0.9761,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977713108498326e-05,
+      "loss": 0.8804,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977294742913228e-05,
+      "loss": 0.9365,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997687249147743e-05,
+      "loss": 0.9106,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976446354355376e-05,
+      "loss": 0.9399,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997601633171304e-05,
+      "loss": 0.9438,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99755824237179e-05,
+      "loss": 0.9375,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997514463053895e-05,
+      "loss": 0.9678,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974702952346704e-05,
+      "loss": 1.0,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997425738931317e-05,
+      "loss": 0.9312,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973807941611888e-05,
+      "loss": 0.9814,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973354609417903e-05,
+      "loss": 0.9443,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972897392907774e-05,
+      "loss": 1.0054,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972436292259578e-05,
+      "loss": 0.9048,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997197130765289e-05,
+      "loss": 0.9761,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971502439268812e-05,
+      "loss": 1.04,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971029687289955e-05,
+      "loss": 0.9219,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970553051900443e-05,
+      "loss": 0.981,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970072533285903e-05,
+      "loss": 0.9385,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969588131633487e-05,
+      "loss": 0.9238,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996909984713186e-05,
+      "loss": 0.9497,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968607679971185e-05,
+      "loss": 0.9336,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968111630343152e-05,
+      "loss": 0.9253,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996761169844095e-05,
+      "loss": 0.9565,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967107884459298e-05,
+      "loss": 0.9424,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996660018859441e-05,
+      "loss": 0.4939,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966088611044014e-05,
+      "loss": 0.9907,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965573152007364e-05,
+      "loss": 0.9683,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965053811685203e-05,
+      "loss": 0.9619,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964530590279812e-05,
+      "loss": 0.9131,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964003487994955e-05,
+      "loss": 0.8945,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963472505035934e-05,
+      "loss": 0.8774,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962937641609547e-05,
+      "loss": 0.9194,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99623988979241e-05,
+      "loss": 0.9624,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961856274189433e-05,
+      "loss": 0.9443,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961309770616866e-05,
+      "loss": 0.9253,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996075938741925e-05,
+      "loss": 0.9399,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960205124810948e-05,
+      "loss": 0.9688,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959646983007817e-05,
+      "loss": 0.5132,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995908496222725e-05,
+      "loss": 0.8906,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958519062688123e-05,
+      "loss": 0.9253,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957949284610848e-05,
+      "loss": 0.9492,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995737562821733e-05,
+      "loss": 0.8804,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956798093730995e-05,
+      "loss": 0.9551,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956216681376766e-05,
+      "loss": 0.9072,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955631391381095e-05,
+      "loss": 0.9238,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995504222397193e-05,
+      "loss": 0.9453,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954449179378733e-05,
+      "loss": 0.8682,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953852257832483e-05,
+      "loss": 0.9438,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953251459565654e-05,
+      "loss": 0.4507,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952646784812243e-05,
+      "loss": 0.9155,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995203823380775e-05,
+      "loss": 0.9321,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951425806789194e-05,
+      "loss": 0.5,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950809503995087e-05,
+      "loss": 0.9048,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950189325665463e-05,
+      "loss": 0.9561,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949565272041867e-05,
+      "loss": 0.9624,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948937343367346e-05,
+      "loss": 0.9409,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948305539886455e-05,
+      "loss": 0.9531,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947669861845267e-05,
+      "loss": 0.9546,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947030309491357e-05,
+      "loss": 0.9404,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994638688307381e-05,
+      "loss": 0.9238,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994573958284323e-05,
+      "loss": 0.918,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945088409051706e-05,
+      "loss": 0.4551,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994443336195286e-05,
+      "loss": 0.9126,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943774441801814e-05,
+      "loss": 0.9614,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994311164885519e-05,
+      "loss": 0.958,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942444983371128e-05,
+      "loss": 0.4465,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941774445609273e-05,
+      "loss": 0.9121,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941100035830786e-05,
+      "loss": 0.9199,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994042175429832e-05,
+      "loss": 0.9868,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939739601276053e-05,
+      "loss": 0.9995,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939053577029654e-05,
+      "loss": 0.9155,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938363681826317e-05,
+      "loss": 0.9014,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937669915934726e-05,
+      "loss": 0.999,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936972279625087e-05,
+      "loss": 0.9653,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936270773169108e-05,
+      "loss": 0.9761,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935565396840008e-05,
+      "loss": 0.9585,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934856150912497e-05,
+      "loss": 0.9521,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934143035662822e-05,
+      "loss": 0.9072,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933426051368702e-05,
+      "loss": 0.4788,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932705198309393e-05,
+      "loss": 0.5,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993198047676564e-05,
+      "loss": 0.9331,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931251887019702e-05,
+      "loss": 0.8955,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930519429355336e-05,
+      "loss": 0.9463,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929783104057822e-05,
+      "loss": 0.9634,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992904291141393e-05,
+      "loss": 0.8999,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928298851711942e-05,
+      "loss": 0.8589,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992755092524165e-05,
+      "loss": 0.9136,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926799132294343e-05,
+      "loss": 0.9653,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926043473162826e-05,
+      "loss": 0.9805,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925283948141404e-05,
+      "loss": 0.9458,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924520557525892e-05,
+      "loss": 0.53,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923753301613598e-05,
+      "loss": 0.9092,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922982180703352e-05,
+      "loss": 0.9141,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922207195095482e-05,
+      "loss": 0.939,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992142834509182e-05,
+      "loss": 0.9619,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920645630995703e-05,
+      "loss": 0.9478,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919859053111974e-05,
+      "loss": 0.9473,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991906861174698e-05,
+      "loss": 0.9355,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991827430720858e-05,
+      "loss": 0.9512,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917476139806123e-05,
+      "loss": 0.8706,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916674109850477e-05,
+      "loss": 0.9106,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915868217654006e-05,
+      "loss": 0.9097,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915058463530578e-05,
+      "loss": 0.918,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914244847795572e-05,
+      "loss": 0.9756,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913427370765862e-05,
+      "loss": 0.4507,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912606032759832e-05,
+      "loss": 0.9067,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911780834097372e-05,
+      "loss": 0.8921,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910951775099866e-05,
+      "loss": 0.959,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910118856090203e-05,
+      "loss": 0.8799,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990928207739279e-05,
+      "loss": 0.8516,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908441439333523e-05,
+      "loss": 0.9731,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907596942239804e-05,
+      "loss": 0.9819,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906748586440535e-05,
+      "loss": 0.9282,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905896372266133e-05,
+      "loss": 0.9458,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905040300048502e-05,
+      "loss": 0.9551,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904180370121055e-05,
+      "loss": 0.8569,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903316582818716e-05,
+      "loss": 0.9268,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902448938477898e-05,
+      "loss": 0.8887,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901577437436525e-05,
+      "loss": 0.9448,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900702080034015e-05,
+      "loss": 0.9287,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98998228666113e-05,
+      "loss": 0.9224,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98989397975108e-05,
+      "loss": 0.4658,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898052873076446e-05,
+      "loss": 0.9644,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989716209365367e-05,
+      "loss": 0.9492,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98962674595894e-05,
+      "loss": 0.9395,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895368971232072e-05,
+      "loss": 0.9629,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894466628931614e-05,
+      "loss": 0.9282,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893560433039468e-05,
+      "loss": 0.9614,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892650383908566e-05,
+      "loss": 0.8789,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989173648189334e-05,
+      "loss": 0.8965,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890818727349735e-05,
+      "loss": 0.939,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889897120635184e-05,
+      "loss": 0.875,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888971662108622e-05,
+      "loss": 0.4458,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888042352130493e-05,
+      "loss": 0.9038,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887109191062728e-05,
+      "loss": 0.9272,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886172179268773e-05,
+      "loss": 0.9331,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885231317113556e-05,
+      "loss": 0.8984,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884286604963518e-05,
+      "loss": 0.498,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98833380431866e-05,
+      "loss": 0.9199,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988238563215223e-05,
+      "loss": 0.8594,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881429372231348e-05,
+      "loss": 0.938,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880469263796386e-05,
+      "loss": 0.4565,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879505307221278e-05,
+      "loss": 0.8823,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878537502881456e-05,
+      "loss": 0.9287,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987756585115385e-05,
+      "loss": 0.9272,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876590352416882e-05,
+      "loss": 0.9478,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987561100705049e-05,
+      "loss": 0.8755,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874627815436092e-05,
+      "loss": 0.9561,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873640777956613e-05,
+      "loss": 0.9688,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872649894996473e-05,
+      "loss": 0.9614,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871655166941586e-05,
+      "loss": 0.9194,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987065659417938e-05,
+      "loss": 0.9038,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986965417709876e-05,
+      "loss": 0.9067,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868647916090137e-05,
+      "loss": 0.8862,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867637811545425e-05,
+      "loss": 0.918,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866623863858017e-05,
+      "loss": 0.8613,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865606073422826e-05,
+      "loss": 0.9043,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864584440636242e-05,
+      "loss": 0.9014,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863558965896165e-05,
+      "loss": 0.917,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986252964960199e-05,
+      "loss": 0.9219,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986149649215459e-05,
+      "loss": 0.9067,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860459493956362e-05,
+      "loss": 0.9023,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985941865541118e-05,
+      "loss": 0.8799,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858373976924415e-05,
+      "loss": 0.9214,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857325458902944e-05,
+      "loss": 0.915,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985627310175513e-05,
+      "loss": 0.9575,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855216905890827e-05,
+      "loss": 0.8521,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854156871721403e-05,
+      "loss": 0.917,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853092999659705e-05,
+      "loss": 0.9214,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852025290120074e-05,
+      "loss": 0.9365,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850953743518353e-05,
+      "loss": 0.9141,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849878360271875e-05,
+      "loss": 0.8672,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848799140799474e-05,
+      "loss": 0.9849,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847716085521464e-05,
+      "loss": 0.9282,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984662919485967e-05,
+      "loss": 0.9639,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98455384692374e-05,
+      "loss": 0.8989,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844443909079456e-05,
+      "loss": 0.939,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984334551481214e-05,
+      "loss": 0.9497,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984224328686324e-05,
+      "loss": 0.9175,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841137225662042e-05,
+      "loss": 0.9746,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984002733163932e-05,
+      "loss": 0.9072,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838913605227345e-05,
+      "loss": 0.915,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983779604685988e-05,
+      "loss": 0.8843,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836674656972176e-05,
+      "loss": 0.9658,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835549436000988e-05,
+      "loss": 0.9219,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834420384384547e-05,
+      "loss": 0.9287,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833287502562588e-05,
+      "loss": 0.8906,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832150790976333e-05,
+      "loss": 0.8555,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98310102500685e-05,
+      "loss": 0.4119,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982986588028329e-05,
+      "loss": 0.8789,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98287176820664e-05,
+      "loss": 0.9185,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827565655865018e-05,
+      "loss": 0.9277,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982640980212783e-05,
+      "loss": 0.8955,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825250121304992e-05,
+      "loss": 0.9092,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824086613848176e-05,
+      "loss": 0.8682,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822919280210532e-05,
+      "loss": 0.9429,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982174812084669e-05,
+      "loss": 0.9233,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820573136212795e-05,
+      "loss": 0.459,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819394326766455e-05,
+      "loss": 0.9487,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818211692966788e-05,
+      "loss": 0.9634,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981702523527439e-05,
+      "loss": 0.9297,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815834954151353e-05,
+      "loss": 0.9473,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814640850061254e-05,
+      "loss": 0.9385,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813442923469157e-05,
+      "loss": 0.8818,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812241174841622e-05,
+      "loss": 0.9121,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981103560464669e-05,
+      "loss": 0.8916,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809826213353897e-05,
+      "loss": 0.9165,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808613001434258e-05,
+      "loss": 0.9219,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807395969360286e-05,
+      "loss": 0.9141,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980617511760598e-05,
+      "loss": 0.9565,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804950446646817e-05,
+      "loss": 0.957,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803721956959778e-05,
+      "loss": 0.9009,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980248964902332e-05,
+      "loss": 0.9824,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801253523317383e-05,
+      "loss": 0.876,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800013580323403e-05,
+      "loss": 0.489,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798769820524302e-05,
+      "loss": 0.9316,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797522244404484e-05,
+      "loss": 0.957,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796270852449845e-05,
+      "loss": 0.8555,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979501564514776e-05,
+      "loss": 1.0,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793756622987098e-05,
+      "loss": 0.8872,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792493786458204e-05,
+      "loss": 0.9297,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979122713605292e-05,
+      "loss": 0.8491,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789956672264563e-05,
+      "loss": 0.9897,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978868239558794e-05,
+      "loss": 0.9341,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787404306519347e-05,
+      "loss": 0.9429,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786122405556557e-05,
+      "loss": 0.9658,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978483669319883e-05,
+      "loss": 0.9756,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783547169946915e-05,
+      "loss": 0.9624,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978225383630304e-05,
+      "loss": 0.9165,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978095669277092e-05,
+      "loss": 0.9463,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.977965573985575e-05,
+      "loss": 0.8545,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778350978064214e-05,
+      "loss": 0.8994,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777042407904476e-05,
+      "loss": 0.9023,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775730029886185e-05,
+      "loss": 0.957,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977441384452047e-05,
+      "loss": 0.8765,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773093852319944e-05,
+      "loss": 0.916,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771770053798705e-05,
+      "loss": 0.9189,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770442449472336e-05,
+      "loss": 0.9512,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976911103985789e-05,
+      "loss": 0.8789,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767775825473918e-05,
+      "loss": 0.9736,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976643680684044e-05,
+      "loss": 0.959,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765093984478967e-05,
+      "loss": 0.9292,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763747358912485e-05,
+      "loss": 0.9038,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976239693066546e-05,
+      "loss": 0.9248,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976104270026385e-05,
+      "loss": 0.9458,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759684668235076e-05,
+      "loss": 0.9048,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758322835108062e-05,
+      "loss": 0.4756,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975695720141319e-05,
+      "loss": 0.9893,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755587767682337e-05,
+      "loss": 0.9507,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754214534448858e-05,
+      "loss": 0.9165,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752837502247576e-05,
+      "loss": 0.9741,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751456671614814e-05,
+      "loss": 0.8877,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975007204308836e-05,
+      "loss": 0.9375,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974868361720748e-05,
+      "loss": 0.8877,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747291394512926e-05,
+      "loss": 0.9229,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745895375546927e-05,
+      "loss": 0.9419,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974449556085319e-05,
+      "loss": 0.9399,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743091950976896e-05,
+      "loss": 0.9214,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741684546464716e-05,
+      "loss": 0.8955,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974027334786478e-05,
+      "loss": 0.9302,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738858355726716e-05,
+      "loss": 0.9604,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737439570601616e-05,
+      "loss": 0.8867,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736016993042053e-05,
+      "loss": 0.8848,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973459062360208e-05,
+      "loss": 0.9146,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733160462837224e-05,
+      "loss": 0.9253,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731726511304483e-05,
+      "loss": 0.9121,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730288769562342e-05,
+      "loss": 0.9385,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972884723817076e-05,
+      "loss": 0.8555,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727401917691165e-05,
+      "loss": 0.9214,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725952808686467e-05,
+      "loss": 0.8506,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724499911721048e-05,
+      "loss": 0.9341,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972304322736077e-05,
+      "loss": 0.936,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972158275617296e-05,
+      "loss": 0.9189,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720118498726434e-05,
+      "loss": 0.9727,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971865045559147e-05,
+      "loss": 0.9526,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717178627339836e-05,
+      "loss": 0.9526,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971570301454475e-05,
+      "loss": 0.9238,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971422361778093e-05,
+      "loss": 0.9287,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712740437624547e-05,
+      "loss": 0.8882,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971125347465326e-05,
+      "loss": 0.8857,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970976272944619e-05,
+      "loss": 0.9194,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708268202583943e-05,
+      "loss": 0.8979,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706769894648587e-05,
+      "loss": 0.4763,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705267806223668e-05,
+      "loss": 0.8472,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703761937894205e-05,
+      "loss": 0.9336,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702252290246683e-05,
+      "loss": 0.8916,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700738863869072e-05,
+      "loss": 0.9653,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699221659350795e-05,
+      "loss": 0.9272,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697700677282763e-05,
+      "loss": 0.9507,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696175918257357e-05,
+      "loss": 0.9028,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969464738286841e-05,
+      "loss": 0.9341,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969311507171125e-05,
+      "loss": 0.9292,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691578985382663e-05,
+      "loss": 0.9004,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969003912448091e-05,
+      "loss": 0.897,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688495489605714e-05,
+      "loss": 0.9053,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686948081358274e-05,
+      "loss": 0.9058,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685396900341263e-05,
+      "loss": 0.8979,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683841947158816e-05,
+      "loss": 0.9131,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968228322241654e-05,
+      "loss": 0.9175,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968072072672151e-05,
+      "loss": 0.9761,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967915446068227e-05,
+      "loss": 0.8936,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677584424908832e-05,
+      "loss": 0.9287,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967601062001268e-05,
+      "loss": 0.9053,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967443304660676e-05,
+      "loss": 0.8965,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967285170530549e-05,
+      "loss": 0.9346,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671266596724753e-05,
+      "loss": 0.9238,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669677721481898e-05,
+      "loss": 0.875,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966808508019575e-05,
+      "loss": 0.9268,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666488673486585e-05,
+      "loss": 0.9355,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966488850197616e-05,
+      "loss": 0.9478,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663284566287693e-05,
+      "loss": 0.9624,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966167686704587e-05,
+      "loss": 0.8574,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660065404876832e-05,
+      "loss": 0.8677,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658450180408204e-05,
+      "loss": 0.9385,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965683119426906e-05,
+      "loss": 0.9023,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655208447089946e-05,
+      "loss": 0.894,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653581939502876e-05,
+      "loss": 0.9136,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651951672141325e-05,
+      "loss": 0.896,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650317645640226e-05,
+      "loss": 0.9175,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648679860635988e-05,
+      "loss": 0.8794,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647038317766474e-05,
+      "loss": 0.8887,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645393017671022e-05,
+      "loss": 0.9648,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643743960990416e-05,
+      "loss": 0.9258,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642091148366923e-05,
+      "loss": 0.9253,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640434580444254e-05,
+      "loss": 0.9082,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638774257867598e-05,
+      "loss": 0.9194,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637110181283593e-05,
+      "loss": 0.8911,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635442351340352e-05,
+      "loss": 0.9043,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633770768687444e-05,
+      "loss": 0.915,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96320954339759e-05,
+      "loss": 0.9116,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96304163478582e-05,
+      "loss": 0.936,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962873351098831e-05,
+      "loss": 0.8926,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627046924021638e-05,
+      "loss": 0.8652,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962535658761506e-05,
+      "loss": 0.9419,
+      "step": 596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623662502426906e-05,
+      "loss": 0.8608,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621964669116974e-05,
+      "loss": 0.8896,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620263088346517e-05,
+      "loss": 0.9038,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618557760778252e-05,
+      "loss": 0.8555,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616848687076345e-05,
+      "loss": 0.8755,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615135867906435e-05,
+      "loss": 0.9277,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613419303935604e-05,
+      "loss": 0.8848,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961169899583241e-05,
+      "loss": 0.9194,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609974944266854e-05,
+      "loss": 0.8716,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608247149910403e-05,
+      "loss": 0.9502,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606515613435986e-05,
+      "loss": 0.9453,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604780335517974e-05,
+      "loss": 0.9204,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960304131683221e-05,
+      "loss": 0.8599,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601298558055988e-05,
+      "loss": 0.9175,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599552059868058e-05,
+      "loss": 0.8633,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959780182294863e-05,
+      "loss": 0.5029,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959604784797936e-05,
+      "loss": 0.4717,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594290135643377e-05,
+      "loss": 0.9727,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592528686625252e-05,
+      "loss": 0.8765,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590763501611015e-05,
+      "loss": 0.9312,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588994581288153e-05,
+      "loss": 0.877,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587221926345604e-05,
+      "loss": 0.8389,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958544553747377e-05,
+      "loss": 0.8496,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583665415364484e-05,
+      "loss": 0.9111,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958188156071107e-05,
+      "loss": 0.9395,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580093974208266e-05,
+      "loss": 0.8687,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578302656552294e-05,
+      "loss": 0.4246,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957650760844082e-05,
+      "loss": 0.9238,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574708830572945e-05,
+      "loss": 0.3982,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957290632364925e-05,
+      "loss": 0.8799,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571100088371753e-05,
+      "loss": 0.9243,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569290125443925e-05,
+      "loss": 0.9629,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567476435570697e-05,
+      "loss": 0.8594,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956565901945844e-05,
+      "loss": 0.4888,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563837877814987e-05,
+      "loss": 0.9331,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562013011349614e-05,
+      "loss": 0.9106,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560184420773045e-05,
+      "loss": 0.9624,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558352106797473e-05,
+      "loss": 0.8926,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556516070136515e-05,
+      "loss": 0.4053,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955467631150526e-05,
+      "loss": 0.9072,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552832831620232e-05,
+      "loss": 0.8,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550985631199417e-05,
+      "loss": 0.8862,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549134710962232e-05,
+      "loss": 0.9604,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547280071629565e-05,
+      "loss": 0.9116,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545421713923734e-05,
+      "loss": 0.9194,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543559638568515e-05,
+      "loss": 0.8979,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954169384628913e-05,
+      "loss": 0.9097,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539824337812248e-05,
+      "loss": 0.8843,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537951113865983e-05,
+      "loss": 0.9062,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95360741751799e-05,
+      "loss": 0.4624,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534193522485007e-05,
+      "loss": 0.9233,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532309156513766e-05,
+      "loss": 0.9463,
+      "step": 648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530421078000072e-05,
+      "loss": 0.8877,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952852928767928e-05,
+      "loss": 0.9194,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526633786288183e-05,
+      "loss": 0.8955,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952473457456502e-05,
+      "loss": 0.8721,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522831653249475e-05,
+      "loss": 0.9185,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952092502308268e-05,
+      "loss": 0.8682,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519014684807207e-05,
+      "loss": 0.8735,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517100639167074e-05,
+      "loss": 0.9009,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515182886907745e-05,
+      "loss": 0.9263,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513261428776123e-05,
+      "loss": 0.9062,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511336265520562e-05,
+      "loss": 0.5037,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950940739789085e-05,
+      "loss": 0.4917,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507474826638218e-05,
+      "loss": 0.9409,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505538552515352e-05,
+      "loss": 0.9463,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503598576276367e-05,
+      "loss": 0.9165,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501654898676828e-05,
+      "loss": 0.8999,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949970752047373e-05,
+      "loss": 0.8984,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497756442425527e-05,
+      "loss": 0.9087,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495801665292094e-05,
+      "loss": 0.9341,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493843189834765e-05,
+      "loss": 0.8662,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94918810168163e-05,
+      "loss": 0.9209,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948991514700091e-05,
+      "loss": 0.8838,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948794558115424e-05,
+      "loss": 0.8901,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485972320043376e-05,
+      "loss": 0.916,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483995364436842e-05,
+      "loss": 0.9072,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482014715104597e-05,
+      "loss": 0.9326,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480030372818054e-05,
+      "loss": 0.9575,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478042338350042e-05,
+      "loss": 0.9463,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476050612474848e-05,
+      "loss": 0.9404,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947405519596818e-05,
+      "loss": 0.8794,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472056089607205e-05,
+      "loss": 0.8931,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94700532941705e-05,
+      "loss": 0.9492,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468046810438098e-05,
+      "loss": 0.9521,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946603663919146e-05,
+      "loss": 0.8877,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464022781213495e-05,
+      "loss": 0.9033,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946200523728853e-05,
+      "loss": 0.8916,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459984008202338e-05,
+      "loss": 0.9526,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945795909474213e-05,
+      "loss": 0.8975,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455930497696546e-05,
+      "loss": 0.896,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453898217855656e-05,
+      "loss": 0.8657,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945186225601098e-05,
+      "loss": 0.9438,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449822612955458e-05,
+      "loss": 0.9111,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944777928948347e-05,
+      "loss": 0.8936,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445732286390825e-05,
+      "loss": 0.8921,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443681604474776e-05,
+      "loss": 0.9165,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944162724453399e-05,
+      "loss": 0.8423,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439569207368585e-05,
+      "loss": 0.8735,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94375074937801e-05,
+      "loss": 0.8887,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435442104571508e-05,
+      "loss": 0.8877,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433373040547222e-05,
+      "loss": 0.8335,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943130030251307e-05,
+      "loss": 0.9082,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429223891276324e-05,
+      "loss": 0.9023,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9427143807645685e-05,
+      "loss": 0.937,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942506005243128e-05,
+      "loss": 0.8745,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422972626444663e-05,
+      "loss": 0.8599,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420881530498832e-05,
+      "loss": 0.918,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418786765408197e-05,
+      "loss": 0.8838,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416688331988608e-05,
+      "loss": 0.8774,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941458623105734e-05,
+      "loss": 0.9243,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412480463433097e-05,
+      "loss": 0.8833,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941037102993601e-05,
+      "loss": 0.5142,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940825793138764e-05,
+      "loss": 0.9287,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406141168610974e-05,
+      "loss": 0.9141,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404020742430427e-05,
+      "loss": 0.9019,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940189665367184e-05,
+      "loss": 0.8857,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399768903162476e-05,
+      "loss": 0.8584,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397637491731035e-05,
+      "loss": 0.877,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395502420207634e-05,
+      "loss": 0.835,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393363689423818e-05,
+      "loss": 0.9243,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391221300212555e-05,
+      "loss": 0.8218,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938907525340824e-05,
+      "loss": 0.854,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386925549846702e-05,
+      "loss": 0.897,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938477219036517e-05,
+      "loss": 0.8945,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382615175802323e-05,
+      "loss": 0.8882,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380454506998245e-05,
+      "loss": 0.9058,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378290184794454e-05,
+      "loss": 0.8257,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376122210033884e-05,
+      "loss": 0.917,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373950583560897e-05,
+      "loss": 0.8813,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937177530622128e-05,
+      "loss": 0.9165,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936959637886223e-05,
+      "loss": 0.9507,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936741380233237e-05,
+      "loss": 0.9058,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365227577481757e-05,
+      "loss": 0.8989,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936303770516185e-05,
+      "loss": 0.8579,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936084418622554e-05,
+      "loss": 0.9399,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358647021527136e-05,
+      "loss": 0.8564,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356446211922364e-05,
+      "loss": 0.8901,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354241758268374e-05,
+      "loss": 0.8247,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352033661423736e-05,
+      "loss": 0.8945,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934982192224843e-05,
+      "loss": 0.8955,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934760654160386e-05,
+      "loss": 0.896,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934538752035286e-05,
+      "loss": 0.9028,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343164859359658e-05,
+      "loss": 0.8794,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340938559489917e-05,
+      "loss": 0.8667,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338708621610717e-05,
+      "loss": 0.9204,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336475046590544e-05,
+      "loss": 0.9385,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933423783529931e-05,
+      "loss": 0.9609,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331996988608344e-05,
+      "loss": 0.9316,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932975250739038e-05,
+      "loss": 0.8813,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932750439251958e-05,
+      "loss": 0.874,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932525264487151e-05,
+      "loss": 0.9316,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322997265323163e-05,
+      "loss": 0.8794,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320738254752938e-05,
+      "loss": 0.9302,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931847561404065e-05,
+      "loss": 0.8291,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316209344067525e-05,
+      "loss": 0.9092,
+      "step": 752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313939445716216e-05,
+      "loss": 0.8745,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311665919870765e-05,
+      "loss": 0.9243,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309388767416654e-05,
+      "loss": 0.8145,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307107989240756e-05,
+      "loss": 0.9409,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304823586231362e-05,
+      "loss": 0.8354,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302535559278185e-05,
+      "loss": 0.8838,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300243909272338e-05,
+      "loss": 0.9023,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297948637106346e-05,
+      "loss": 0.9175,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929564974367415e-05,
+      "loss": 0.9443,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92933472298711e-05,
+      "loss": 0.8784,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929104109659395e-05,
+      "loss": 0.9351,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288731344740873e-05,
+      "loss": 0.916,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286417975211443e-05,
+      "loss": 0.8984,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284100988906647e-05,
+      "loss": 0.938,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928178038672888e-05,
+      "loss": 0.9292,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279456169581953e-05,
+      "loss": 0.9688,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277128338371064e-05,
+      "loss": 0.9087,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927479689400284e-05,
+      "loss": 0.5342,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927246183738531e-05,
+      "loss": 0.9355,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9270123169427897e-05,
+      "loss": 0.9258,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267780891041453e-05,
+      "loss": 0.9297,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265435003138212e-05,
+      "loss": 0.9224,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263085506631833e-05,
+      "loss": 0.8472,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260732402437372e-05,
+      "loss": 0.9258,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925837569147129e-05,
+      "loss": 0.5464,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256015374651452e-05,
+      "loss": 0.894,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925365145289713e-05,
+      "loss": 0.9302,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251283927129e-05,
+      "loss": 0.915,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248912798269143e-05,
+      "loss": 0.8032,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246538067241036e-05,
+      "loss": 0.9658,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924415973496957e-05,
+      "loss": 0.8984,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241777802381026e-05,
+      "loss": 0.8867,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239392270403097e-05,
+      "loss": 0.9077,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237003139964873e-05,
+      "loss": 0.4668,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234610411996854e-05,
+      "loss": 0.8843,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232214087430924e-05,
+      "loss": 0.8999,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9229814167200382e-05,
+      "loss": 0.9404,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922741065223993e-05,
+      "loss": 0.4763,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922500354348565e-05,
+      "loss": 0.5298,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922259284187505e-05,
+      "loss": 0.8828,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922017854834702e-05,
+      "loss": 0.9204,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217760663841854e-05,
+      "loss": 0.8979,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921533918930124e-05,
+      "loss": 0.8921,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212914125668273e-05,
+      "loss": 0.9136,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210485473887437e-05,
+      "loss": 0.9199,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920805323490462e-05,
+      "loss": 0.9204,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205617409667107e-05,
+      "loss": 0.9121,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203177999123567e-05,
+      "loss": 0.8584,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920073500422409e-05,
+      "loss": 0.8555,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919828842592014e-05,
+      "loss": 0.9048,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919583826516458e-05,
+      "loss": 0.9141,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919338452291168e-05,
+      "loss": 0.9111,
+      "step": 804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190927200117094e-05,
+      "loss": 0.9077,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918846629773788e-05,
+      "loss": 0.9033,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9186001816732476e-05,
+      "loss": 0.896,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183533758060726e-05,
+      "loss": 0.8921,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918106212268386e-05,
+      "loss": 0.8359,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178586911564506e-05,
+      "loss": 0.8687,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176108125666685e-05,
+      "loss": 0.8428,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173625765955805e-05,
+      "loss": 0.9443,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171139833398675e-05,
+      "loss": 0.9102,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916865032896348e-05,
+      "loss": 0.9351,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166157253619813e-05,
+      "loss": 0.9028,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916366060833865e-05,
+      "loss": 0.8721,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161160394092354e-05,
+      "loss": 0.873,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158656611854686e-05,
+      "loss": 0.8652,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915614926260079e-05,
+      "loss": 0.9243,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153638347307205e-05,
+      "loss": 0.8491,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151123866951852e-05,
+      "loss": 0.8633,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148605822514042e-05,
+      "loss": 0.9146,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914608421497448e-05,
+      "loss": 0.9043,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143559045315255e-05,
+      "loss": 0.8652,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914103031451984e-05,
+      "loss": 0.8428,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138498023573104e-05,
+      "loss": 0.8906,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135962173461293e-05,
+      "loss": 0.8589,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133422765172043e-05,
+      "loss": 0.8691,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130879799694376e-05,
+      "loss": 0.8745,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128333278018702e-05,
+      "loss": 0.8838,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125783201136805e-05,
+      "loss": 0.9248,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123229570041874e-05,
+      "loss": 0.9443,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120672385728455e-05,
+      "loss": 0.9268,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118111649192505e-05,
+      "loss": 0.835,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115547361431353e-05,
+      "loss": 0.8589,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91129795234437e-05,
+      "loss": 0.8652,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110408136229652e-05,
+      "loss": 0.8965,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910783320079068e-05,
+      "loss": 0.498,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910525471812964e-05,
+      "loss": 0.8823,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102672689250773e-05,
+      "loss": 0.9092,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91000871151597e-05,
+      "loss": 0.9609,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909749799686343e-05,
+      "loss": 0.8911,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909490533537034e-05,
+      "loss": 0.8379,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909230913169019e-05,
+      "loss": 0.856,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089709386834122e-05,
+      "loss": 0.9282,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087106101814663e-05,
+      "loss": 0.8638,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084499277645708e-05,
+      "loss": 0.8794,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081888915342534e-05,
+      "loss": 0.9111,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90792750159218e-05,
+      "loss": 0.8472,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907665758040154e-05,
+      "loss": 0.9751,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074036609801166e-05,
+      "loss": 0.897,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907141210514146e-05,
+      "loss": 0.9199,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906878406744459e-05,
+      "loss": 0.9507,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066152497734096e-05,
+      "loss": 0.8892,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063517397034898e-05,
+      "loss": 0.9077,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060878766373275e-05,
+      "loss": 0.9175,
+      "step": 856
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905823660677691e-05,
+      "loss": 0.897,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055590919274826e-05,
+      "loss": 0.9243,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052941704897448e-05,
+      "loss": 0.897,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905028896467656e-05,
+      "loss": 0.8921,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047632699645326e-05,
+      "loss": 0.9062,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044972910838274e-05,
+      "loss": 0.8848,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904230959929132e-05,
+      "loss": 0.856,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039642766041732e-05,
+      "loss": 0.8721,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036972412128167e-05,
+      "loss": 0.894,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034298538590642e-05,
+      "loss": 0.9292,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903162114647055e-05,
+      "loss": 0.9272,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028940236810655e-05,
+      "loss": 0.4775,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902625581065509e-05,
+      "loss": 0.8882,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023567869049348e-05,
+      "loss": 0.8418,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902087641304031e-05,
+      "loss": 0.9023,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018181443676212e-05,
+      "loss": 0.873,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901548296200666e-05,
+      "loss": 0.9058,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012780969082633e-05,
+      "loss": 0.9272,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010075465956474e-05,
+      "loss": 0.8931,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007366453681896e-05,
+      "loss": 0.8691,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900465393331397e-05,
+      "loss": 0.8975,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001937905909143e-05,
+      "loss": 0.9233,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8999218372525224e-05,
+      "loss": 0.9248,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996495334221393e-05,
+      "loss": 0.877,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899376879205818e-05,
+      "loss": 0.8711,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991038747097495e-05,
+      "loss": 0.8516,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898830520040261e-05,
+      "loss": 0.9023,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985568153038155e-05,
+      "loss": 0.873,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982827606070123e-05,
+      "loss": 0.8394,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980083560565878e-05,
+      "loss": 0.835,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977336017594142e-05,
+      "loss": 0.9058,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974584978224994e-05,
+      "loss": 0.8857,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897183044352988e-05,
+      "loss": 0.8862,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896907241458161e-05,
+      "loss": 0.8945,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8966310892454353e-05,
+      "loss": 0.9155,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896354587822363e-05,
+      "loss": 0.9434,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896077737296634e-05,
+      "loss": 0.8203,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958005377760724e-05,
+      "loss": 0.9155,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955229893686387e-05,
+      "loss": 0.9014,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89524509218243e-05,
+      "loss": 0.9517,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949668463256788e-05,
+      "loss": 0.8867,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894688251906753e-05,
+      "loss": 0.9087,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944093090341567e-05,
+      "loss": 0.9126,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8941300178165297e-05,
+      "loss": 0.8667,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938503783626476e-05,
+      "loss": 0.9395,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935703907814206e-05,
+      "loss": 0.8613,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932900551818963e-05,
+      "loss": 0.9351,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893009371673256e-05,
+      "loss": 0.833,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892728340364818e-05,
+      "loss": 0.9146,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8924469613660344e-05,
+      "loss": 0.8462,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892165234786495e-05,
+      "loss": 0.8672,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8918831607359226e-05,
+      "loss": 0.897,
+      "step": 908
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891600739324177e-05,
+      "loss": 0.9111,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913179706612526e-05,
+      "loss": 0.896,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8910348548572793e-05,
+      "loss": 0.8618,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890751392022521e-05,
+      "loss": 0.9302,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890467582267379e-05,
+      "loss": 0.9023,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8901834257023883e-05,
+      "loss": 0.9302,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889898922438219e-05,
+      "loss": 0.8755,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889614072585676e-05,
+      "loss": 0.9233,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8893288762557e-05,
+      "loss": 0.8882,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890433335593666e-05,
+      "loss": 0.8311,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888757444607885e-05,
+      "loss": 0.895,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884712095126013e-05,
+      "loss": 0.8623,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888184628384994e-05,
+      "loss": 0.8716,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878977013366787e-05,
+      "loss": 0.9048,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8876104284794047e-05,
+      "loss": 0.8335,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873228099250554e-05,
+      "loss": 0.8472,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870348457856496e-05,
+      "loss": 0.8979,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886746536173341e-05,
+      "loss": 0.8267,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886457881200417e-05,
+      "loss": 0.8999,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886168880979299e-05,
+      "loss": 0.8184,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885879535622545e-05,
+      "loss": 0.813,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885589845242846e-05,
+      "loss": 0.9238,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8852998099530273e-05,
+      "loss": 0.8218,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8850094298660485e-05,
+      "loss": 0.8638,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847187050950037e-05,
+      "loss": 0.9136,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884427635753122e-05,
+      "loss": 0.8696,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841362219537653e-05,
+      "loss": 0.4844,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838444638104308e-05,
+      "loss": 0.9541,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883552361436749e-05,
+      "loss": 0.9062,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883259914946485e-05,
+      "loss": 0.8975,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829671244535377e-05,
+      "loss": 0.8828,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88267399007194e-05,
+      "loss": 0.8491,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882380511915859e-05,
+      "loss": 0.9087,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820866900995954e-05,
+      "loss": 0.9136,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8817925247375837e-05,
+      "loss": 0.8926,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814980159443917e-05,
+      "loss": 0.8872,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812031638347228e-05,
+      "loss": 0.9199,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8809079685234118e-05,
+      "loss": 0.46,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880612430125429e-05,
+      "loss": 0.918,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880316548755877e-05,
+      "loss": 0.9497,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800203245299923e-05,
+      "loss": 0.8818,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797237575631463e-05,
+      "loss": 0.8716,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879426847970842e-05,
+      "loss": 0.9546,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879129595868716e-05,
+      "loss": 0.835,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788320013725395e-05,
+      "loss": 0.895,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878534064598217e-05,
+      "loss": 0.8247,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878235785661785e-05,
+      "loss": 0.509,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877937164679414e-05,
+      "loss": 0.9014,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776382017674084e-05,
+      "loss": 0.8511,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773388970422042e-05,
+      "loss": 0.8726,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770392506203722e-05,
+      "loss": 0.8003,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767392626186146e-05,
+      "loss": 0.9053,
+      "step": 960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764389331537687e-05,
+      "loss": 0.8823,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761382623428027e-05,
+      "loss": 0.9067,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875837250302819e-05,
+      "loss": 0.9136,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875535897151053e-05,
+      "loss": 0.9092,
+      "step": 964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752342030048716e-05,
+      "loss": 0.8857,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749321679817762e-05,
+      "loss": 0.5132,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746297921993997e-05,
+      "loss": 0.8896,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8743270757755084e-05,
+      "loss": 0.8838,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874024018828001e-05,
+      "loss": 0.5312,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737206214749094e-05,
+      "loss": 0.894,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734168838343968e-05,
+      "loss": 0.897,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731128060247605e-05,
+      "loss": 0.4622,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728083881644287e-05,
+      "loss": 0.8862,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872503630371963e-05,
+      "loss": 0.8857,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8721985327660578e-05,
+      "loss": 0.9756,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718930954655388e-05,
+      "loss": 0.8662,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715873185893646e-05,
+      "loss": 0.9624,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8712812022566256e-05,
+      "loss": 0.8945,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8709747465865455e-05,
+      "loss": 0.8779,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706679516984788e-05,
+      "loss": 0.8286,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870360817711913e-05,
+      "loss": 0.9351,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700533447464665e-05,
+      "loss": 0.8726,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869745532921892e-05,
+      "loss": 0.915,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869437382358072e-05,
+      "loss": 0.4917,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8691288931750217e-05,
+      "loss": 0.856,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688200654928882e-05,
+      "loss": 0.8813,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685108994319505e-05,
+      "loss": 0.854,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682013951126196e-05,
+      "loss": 0.8794,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678915526554375e-05,
+      "loss": 0.8076,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867581372181079e-05,
+      "loss": 0.8999,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8672708538103487e-05,
+      "loss": 0.8828,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669599976641852e-05,
+      "loss": 0.9141,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8666488038636566e-05,
+      "loss": 0.8604,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663372725299637e-05,
+      "loss": 0.8643,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.8813,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8657131977485446e-05,
+      "loss": 0.4751,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8654006545438763e-05,
+      "loss": 0.9038,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650877742921595e-05,
+      "loss": 0.9272,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647745571152515e-05,
+      "loss": 0.8345,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644610031351412e-05,
+      "loss": 0.4617,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641471124739478e-05,
+      "loss": 0.8174,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863832885253922e-05,
+      "loss": 0.8813,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8635183215974466e-05,
+      "loss": 0.9355,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8632034216270334e-05,
+      "loss": 0.9277,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862888185465327e-05,
+      "loss": 0.9297,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862572613235102e-05,
+      "loss": 0.8877,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862256705059264e-05,
+      "loss": 0.9058,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86194046106085e-05,
+      "loss": 0.853,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8616238813630264e-05,
+      "loss": 0.8809,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613069660890923e-05,
+      "loss": 0.918,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860989715362476e-05,
+      "loss": 0.9414,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8606721293067372e-05,
+      "loss": 0.4844,
+      "step": 1012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860354208045566e-05,
+      "loss": 0.8354,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8600359517027824e-05,
+      "loss": 0.8628,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859717360402338e-05,
+      "loss": 0.8804,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593984342683145e-05,
+      "loss": 0.8745,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590791734249233e-05,
+      "loss": 0.8687,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858759577996507e-05,
+      "loss": 0.916,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584396481075385e-05,
+      "loss": 0.9585,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8581193838826203e-05,
+      "loss": 0.8906,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577987854464857e-05,
+      "loss": 0.8823,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574778529239975e-05,
+      "loss": 0.9478,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85715658644015e-05,
+      "loss": 0.8931,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856834986120066e-05,
+      "loss": 0.8804,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565130520889994e-05,
+      "loss": 0.9399,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561907844723334e-05,
+      "loss": 0.8848,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855868183395581e-05,
+      "loss": 0.8335,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855545248984386e-05,
+      "loss": 0.8677,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8552219813645213e-05,
+      "loss": 0.9004,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8548983806618893e-05,
+      "loss": 0.8838,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854574447002523e-05,
+      "loss": 0.8745,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854250180512585e-05,
+      "loss": 0.96,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539255813183665e-05,
+      "loss": 0.5383,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536006495462893e-05,
+      "loss": 0.8604,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532753853229042e-05,
+      "loss": 0.8765,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8529497887748918e-05,
+      "loss": 0.9165,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852623860029062e-05,
+      "loss": 0.9644,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852297599212354e-05,
+      "loss": 0.9653,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851971006451836e-05,
+      "loss": 0.8784,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516440818747062e-05,
+      "loss": 0.8579,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513168256082922e-05,
+      "loss": 0.9131,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509892377800495e-05,
+      "loss": 0.8589,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506613185175638e-05,
+      "loss": 0.8604,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8503330679485502e-05,
+      "loss": 0.8408,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500044862008513e-05,
+      "loss": 0.7886,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8496755734024402e-05,
+      "loss": 0.8799,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849346329681418e-05,
+      "loss": 0.8833,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490167551660156e-05,
+      "loss": 0.8682,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486868499845912e-05,
+      "loss": 0.9028,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848356614265634e-05,
+      "loss": 0.8911,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84802604813776e-05,
+      "loss": 0.4985,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476951517297145e-05,
+      "loss": 0.9253,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473639251703716e-05,
+      "loss": 0.9175,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470323685887343e-05,
+      "loss": 0.9082,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467004821139332e-05,
+      "loss": 0.8857,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463682658752283e-05,
+      "loss": 0.9023,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460357200020077e-05,
+      "loss": 0.9326,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845702844623787e-05,
+      "loss": 0.8975,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845369639870212e-05,
+      "loss": 0.8809,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845036105871055e-05,
+      "loss": 0.9009,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447022427562182e-05,
+      "loss": 0.8682,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84436805065573e-05,
+      "loss": 0.8867,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8440335296997486e-05,
+      "loss": 0.8735,
+      "step": 1063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8436986800185593e-05,
+      "loss": 0.8467,
+      "step": 1064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433635017425763e-05,
+      "loss": 0.8525,
+      "step": 1065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843027995002341e-05,
+      "loss": 0.9004,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8426921599285225e-05,
+      "loss": 0.9087,
+      "step": 1067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423559966519195e-05,
+      "loss": 0.8989,
+      "step": 1068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420195053034562e-05,
+      "loss": 0.8647,
+      "step": 1069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841682686014186e-05,
+      "loss": 0.9199,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8413455389152895e-05,
+      "loss": 0.854,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410080641380758e-05,
+      "loss": 0.8232,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406702618139803e-05,
+      "loss": 0.9185,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840332132074567e-05,
+      "loss": 0.8086,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839993675051527e-05,
+      "loss": 0.4707,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396548908766793e-05,
+      "loss": 0.8999,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8393157796819693e-05,
+      "loss": 0.8418,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838976341599471e-05,
+      "loss": 0.9087,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386365767613842e-05,
+      "loss": 0.8726,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382964853000384e-05,
+      "loss": 0.5168,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379560673478874e-05,
+      "loss": 0.9438,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376153230375147e-05,
+      "loss": 0.9062,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837274252501629e-05,
+      "loss": 0.8604,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369328558730668e-05,
+      "loss": 0.8877,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8365911332847922e-05,
+      "loss": 0.8843,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362490848698954e-05,
+      "loss": 0.9043,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359067107615934e-05,
+      "loss": 0.8535,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835564011093231e-05,
+      "loss": 0.4729,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835220985998279e-05,
+      "loss": 0.8594,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834877635610335e-05,
+      "loss": 0.8477,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8345339600631238e-05,
+      "loss": 0.4673,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8341899594904963e-05,
+      "loss": 0.8696,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338456340264303e-05,
+      "loss": 0.8281,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83350098380503e-05,
+      "loss": 0.8428,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331560089605257e-05,
+      "loss": 0.8687,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328107096272752e-05,
+      "loss": 0.833,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324650859397613e-05,
+      "loss": 0.853,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321191380325947e-05,
+      "loss": 0.8506,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317728660405106e-05,
+      "loss": 0.8174,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831426270098372e-05,
+      "loss": 0.9038,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310793503411672e-05,
+      "loss": 0.8555,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307321069040108e-05,
+      "loss": 0.8018,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8303845399221436e-05,
+      "loss": 0.8633,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830036649530932e-05,
+      "loss": 0.8452,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8296884358658687e-05,
+      "loss": 0.8237,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293398990625724e-05,
+      "loss": 0.8569,
+      "step": 1106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828991039256788e-05,
+      "loss": 0.937,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8286418565843848e-05,
+      "loss": 0.8774,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828292351181359e-05,
+      "loss": 0.8618,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279425231838327e-05,
+      "loss": 0.9165,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275923727280525e-05,
+      "loss": 0.8911,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272418999503923e-05,
+      "loss": 0.8511,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8268911049873497e-05,
+      "loss": 0.8242,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265399879755487e-05,
+      "loss": 0.8647,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261885490517387e-05,
+      "loss": 0.8652,
+      "step": 1115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8258367883527942e-05,
+      "loss": 0.9395,
+      "step": 1116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254847060157156e-05,
+      "loss": 0.8594,
+      "step": 1117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825132302177628e-05,
+      "loss": 0.8506,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247795769757818e-05,
+      "loss": 0.874,
+      "step": 1119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824426530547553e-05,
+      "loss": 0.8721,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824073163030442e-05,
+      "loss": 0.8589,
+      "step": 1121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8237194745620748e-05,
+      "loss": 0.4888,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823365465280202e-05,
+      "loss": 0.9302,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230111353226997e-05,
+      "loss": 0.8608,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226564848275684e-05,
+      "loss": 0.8682,
+      "step": 1125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223015139329336e-05,
+      "loss": 0.895,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821946222777046e-05,
+      "loss": 0.9082,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215906114982797e-05,
+      "loss": 0.9038,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8212346802351354e-05,
+      "loss": 0.9092,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8208784291262366e-05,
+      "loss": 0.895,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8205218583103328e-05,
+      "loss": 0.481,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820164967926297e-05,
+      "loss": 0.8916,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198077581131275e-05,
+      "loss": 0.8418,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819450229009946e-05,
+      "loss": 0.9214,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190923807559997e-05,
+      "loss": 0.8657,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818734213490659e-05,
+      "loss": 0.8931,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81837572735342e-05,
+      "loss": 0.9272,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8180169224839012e-05,
+      "loss": 0.392,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817657799021846e-05,
+      "loss": 0.8315,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172983571071226e-05,
+      "loss": 0.8921,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8169385968797225e-05,
+      "loss": 0.8511,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816578518479761e-05,
+      "loss": 0.9302,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8162181220474777e-05,
+      "loss": 0.8408,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815857407723236e-05,
+      "loss": 0.8887,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154963756475236e-05,
+      "loss": 0.8535,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81513502596095e-05,
+      "loss": 0.8604,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8147733588042516e-05,
+      "loss": 0.8833,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8144113743182856e-05,
+      "loss": 0.8667,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8140490726440344e-05,
+      "loss": 0.873,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136864539226023e-05,
+      "loss": 0.8975,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133235182952198e-05,
+      "loss": 0.8955,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8129602659032384e-05,
+      "loss": 0.8784,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8125966968881336e-05,
+      "loss": 0.5247,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8122328113915046e-05,
+      "loss": 0.8682,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8118686095550736e-05,
+      "loss": 0.8584,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8115040915206866e-05,
+      "loss": 0.8877,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111392574303113e-05,
+      "loss": 0.9316,
+      "step": 1157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107741074260397e-05,
+      "loss": 0.896,
+      "step": 1158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810408641650087e-05,
+      "loss": 0.5037,
+      "step": 1159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8100428602447905e-05,
+      "loss": 0.9053,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809676763352611e-05,
+      "loss": 0.9268,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809310351116132e-05,
+      "loss": 0.937,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089436236780598e-05,
+      "loss": 0.8491,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085765811812236e-05,
+      "loss": 0.937,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8082092237685748e-05,
+      "loss": 0.8711,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078415515831885e-05,
+      "loss": 0.855,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074735647682616e-05,
+      "loss": 0.8843,
+      "step": 1167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8071052634671134e-05,
+      "loss": 0.8984,
+      "step": 1168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806736647823186e-05,
+      "loss": 0.8984,
+      "step": 1169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806367717980044e-05,
+      "loss": 0.8604,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8059984740813745e-05,
+      "loss": 0.8989,
+      "step": 1171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056289162709864e-05,
+      "loss": 0.9062,
+      "step": 1172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8052590446928112e-05,
+      "loss": 0.8696,
+      "step": 1173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804888859490903e-05,
+      "loss": 0.8843,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045183608094367e-05,
+      "loss": 0.9292,
+      "step": 1175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804147548792711e-05,
+      "loss": 0.8203,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037764235851452e-05,
+      "loss": 0.8843,
+      "step": 1177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8034049853312817e-05,
+      "loss": 0.9146,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030332341757838e-05,
+      "loss": 0.8535,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802661170263438e-05,
+      "loss": 0.8794,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8022887937391513e-05,
+      "loss": 0.876,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8019161047479526e-05,
+      "loss": 0.8013,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8015431034349932e-05,
+      "loss": 0.8633,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011697899455458e-05,
+      "loss": 0.8403,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800796164425005e-05,
+      "loss": 0.5251,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8004222270188853e-05,
+      "loss": 0.9097,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8000479778728245e-05,
+      "loss": 0.8726,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996734171325817e-05,
+      "loss": 0.8843,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992985449440362e-05,
+      "loss": 0.8706,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798923361453189e-05,
+      "loss": 0.9097,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798547866806163e-05,
+      "loss": 0.915,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981720611492023e-05,
+      "loss": 0.8486,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797795944628671e-05,
+      "loss": 0.9019,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974195173910553e-05,
+      "loss": 0.9033,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7970427795829616e-05,
+      "loss": 0.8662,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796665731351118e-05,
+      "loss": 0.8838,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962883728423737e-05,
+      "loss": 0.8657,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795910704203698e-05,
+      "loss": 0.8838,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7955327255821807e-05,
+      "loss": 0.8818,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795154437125034e-05,
+      "loss": 0.8848,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7947758389795885e-05,
+      "loss": 0.9263,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943969312932976e-05,
+      "loss": 0.8628,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7940177142137335e-05,
+      "loss": 0.8076,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7936381878885897e-05,
+      "loss": 0.9097,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932583524656807e-05,
+      "loss": 0.8779,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7928782080929398e-05,
+      "loss": 0.877,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7924977549184226e-05,
+      "loss": 0.832,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7921169930903035e-05,
+      "loss": 0.8462,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917359227568774e-05,
+      "loss": 0.4983,
+      "step": 1209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7913545440665598e-05,
+      "loss": 0.8979,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7909728571678856e-05,
+      "loss": 0.8594,
+      "step": 1211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905908622095107e-05,
+      "loss": 0.8813,
+      "step": 1212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7902085593402102e-05,
+      "loss": 0.8276,
+      "step": 1213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7898259487088792e-05,
+      "loss": 0.9126,
+      "step": 1214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789443030464533e-05,
+      "loss": 0.4817,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7890598047563067e-05,
+      "loss": 0.9326,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7886762717334546e-05,
+      "loss": 0.7979,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7882924315453515e-05,
+      "loss": 0.894,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879082843414907e-05,
+      "loss": 0.8813,
+      "step": 1219
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7875238302714866e-05,
+      "loss": 0.895,
+      "step": 1220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871390694850718e-05,
+      "loss": 0.8379,
+      "step": 1221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786754002132099e-05,
+      "loss": 0.875,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78636862836254e-05,
+      "loss": 0.499,
+      "step": 1223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785982948326486e-05,
+      "loss": 0.8379,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7855969621741477e-05,
+      "loss": 0.939,
+      "step": 1225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852106700558553e-05,
+      "loss": 0.8799,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7848240721220568e-05,
+      "loss": 0.8652,
+      "step": 1227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844371685233214e-05,
+      "loss": 0.917,
+      "step": 1228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784049959410335e-05,
+      "loss": 0.9229,
+      "step": 1229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7836624449339046e-05,
+      "loss": 0.9224,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783274625244955e-05,
+      "loss": 0.9082,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828865004945296e-05,
+      "loss": 0.8745,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7824980708337917e-05,
+      "loss": 0.8501,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782109336414023e-05,
+      "loss": 0.8086,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7817202973866226e-05,
+      "loss": 0.8281,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7813309539031102e-05,
+      "loss": 0.8433,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809413061151227e-05,
+      "loss": 0.896,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805513541744163e-05,
+      "loss": 0.4663,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7801610982328652e-05,
+      "loss": 0.9121,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779770538442462e-05,
+      "loss": 0.8599,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779379674955318e-05,
+      "loss": 0.877,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778988507923662e-05,
+      "loss": 0.8149,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7785970374998423e-05,
+      "loss": 0.8276,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778205263836325e-05,
+      "loss": 0.5322,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7778131870856924e-05,
+      "loss": 0.877,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7774208074006477e-05,
+      "loss": 0.8535,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77702812493401e-05,
+      "loss": 0.8657,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7766351398387177e-05,
+      "loss": 0.8315,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7762418522678265e-05,
+      "loss": 0.8965,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775848262374509e-05,
+      "loss": 0.8115,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754543703120574e-05,
+      "loss": 0.8359,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77506017623388e-05,
+      "loss": 0.9082,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7746656802935033e-05,
+      "loss": 0.8794,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7742708826445717e-05,
+      "loss": 0.7959,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773875783440847e-05,
+      "loss": 0.8423,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7734803828362074e-05,
+      "loss": 0.8687,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77308468098465e-05,
+      "loss": 0.8506,
+      "step": 1257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7726886780402886e-05,
+      "loss": 0.8477,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772292374157354e-05,
+      "loss": 0.8818,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771895769490194e-05,
+      "loss": 0.8247,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7714988641932748e-05,
+      "loss": 0.9199,
+      "step": 1261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771101658421178e-05,
+      "loss": 0.8726,
+      "step": 1262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770704152328604e-05,
+      "loss": 0.8477,
+      "step": 1263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7703063460703685e-05,
+      "loss": 0.8721,
+      "step": 1264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7699082398014055e-05,
+      "loss": 0.8647,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7695098336767646e-05,
+      "loss": 0.9038,
+      "step": 1266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769111127851613e-05,
+      "loss": 0.8159,
+      "step": 1267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7687121224812342e-05,
+      "loss": 0.8374,
+      "step": 1268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768312817721029e-05,
+      "loss": 0.8965,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767913213726514e-05,
+      "loss": 0.8799,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7675133106533234e-05,
+      "loss": 0.8823,
+      "step": 1271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7671131086572062e-05,
+      "loss": 0.8823,
+      "step": 1272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7667126078940293e-05,
+      "loss": 0.4761,
+      "step": 1273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663118085197758e-05,
+      "loss": 0.9009,
+      "step": 1274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765910710690544e-05,
+      "loss": 0.8345,
+      "step": 1275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76550931456255e-05,
+      "loss": 0.4373,
+      "step": 1276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765107620292125e-05,
+      "loss": 0.8848,
+      "step": 1277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764705628035716e-05,
+      "loss": 0.8721,
+      "step": 1278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764303337949888e-05,
+      "loss": 0.8726,
+      "step": 1279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7639007501913196e-05,
+      "loss": 0.8555,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763497864916807e-05,
+      "loss": 0.8911,
+      "step": 1281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763094682283261e-05,
+      "loss": 0.8887,
+      "step": 1282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7626912024477096e-05,
+      "loss": 0.8716,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7622874255672954e-05,
+      "loss": 0.4851,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7618833517992772e-05,
+      "loss": 0.8735,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7614789813010293e-05,
+      "loss": 0.5491,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761074314230042e-05,
+      "loss": 0.8132,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7606693507439203e-05,
+      "loss": 0.9409,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7602640910003856e-05,
+      "loss": 0.8491,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759858535157273e-05,
+      "loss": 0.7866,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7594526833725355e-05,
+      "loss": 0.8462,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7590465358042393e-05,
+      "loss": 0.9243,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7586400926105663e-05,
+      "loss": 0.9365,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582333539498137e-05,
+      "loss": 0.8677,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757826319980394e-05,
+      "loss": 0.874,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7574189908608345e-05,
+      "loss": 0.9224,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7570113667497767e-05,
+      "loss": 0.9033,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756603447805979e-05,
+      "loss": 0.8184,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7561952341883122e-05,
+      "loss": 0.8711,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7557867260557634e-05,
+      "loss": 0.8711,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755377923567434e-05,
+      "loss": 0.8369,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75496882688254e-05,
+      "loss": 0.9336,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7545594361604122e-05,
+      "loss": 0.8828,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541497515604957e-05,
+      "loss": 0.8076,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75373977324235e-05,
+      "loss": 0.8525,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7533295013656492e-05,
+      "loss": 0.4995,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7529189360901814e-05,
+      "loss": 0.8247,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752508077575849e-05,
+      "loss": 0.8247,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520969259826698e-05,
+      "loss": 0.8921,
+      "step": 1309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7516854814707736e-05,
+      "loss": 0.8491,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7512737442004063e-05,
+      "loss": 0.7651,
+      "step": 1311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7508617143319266e-05,
+      "loss": 0.8691,
+      "step": 1312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7504493920258074e-05,
+      "loss": 0.8848,
+      "step": 1313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750036777442636e-05,
+      "loss": 0.9233,
+      "step": 1314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7496238707431124e-05,
+      "loss": 0.7988,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492106720880517e-05,
+      "loss": 0.9067,
+      "step": 1316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748797181638382e-05,
+      "loss": 0.8638,
+      "step": 1317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7483833995551454e-05,
+      "loss": 0.8604,
+      "step": 1318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7479693259994966e-05,
+      "loss": 0.8599,
+      "step": 1319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747554961132706e-05,
+      "loss": 0.8892,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747140305116154e-05,
+      "loss": 0.8882,
+      "step": 1321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467253581113384e-05,
+      "loss": 0.8682,
+      "step": 1322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7463101202798673e-05,
+      "loss": 0.8433,
+      "step": 1323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7458945917834632e-05,
+      "loss": 0.8457,
+      "step": 1324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454787727839618e-05,
+      "loss": 0.8794,
+      "step": 1325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7450626634433115e-05,
+      "loss": 0.8242,
+      "step": 1326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744646263923575e-05,
+      "loss": 0.8701,
+      "step": 1327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442295743869264e-05,
+      "loss": 0.8521,
+      "step": 1328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743812594995654e-05,
+      "loss": 0.8115,
+      "step": 1329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7433953259121583e-05,
+      "loss": 0.8545,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7429777672989524e-05,
+      "loss": 0.877,
+      "step": 1331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7425599193186635e-05,
+      "loss": 0.4792,
+      "step": 1332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74214178213403e-05,
+      "loss": 0.8677,
+      "step": 1333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7417233559079035e-05,
+      "loss": 0.8984,
+      "step": 1334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7413046408032484e-05,
+      "loss": 0.8525,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7408856369831412e-05,
+      "loss": 0.8496,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740466344610771e-05,
+      "loss": 0.8838,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74004676384944e-05,
+      "loss": 0.7886,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7396268948625615e-05,
+      "loss": 0.8423,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392067378136618e-05,
+      "loss": 0.8838,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7387862928663785e-05,
+      "loss": 0.854,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738365560184463e-05,
+      "loss": 0.8232,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7379445399317774e-05,
+      "loss": 0.9131,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737523232272296e-05,
+      "loss": 0.7915,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7371016373701057e-05,
+      "loss": 0.4661,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7366797553894044e-05,
+      "loss": 0.8955,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362575864945025e-05,
+      "loss": 0.8062,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358351308498214e-05,
+      "loss": 0.8438,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735412388619895e-05,
+      "loss": 0.8447,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7349893599693688e-05,
+      "loss": 0.481,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7345660450629988e-05,
+      "loss": 0.8521,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7341424440656536e-05,
+      "loss": 0.833,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7337185571423124e-05,
+      "loss": 0.8394,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7332943844580667e-05,
+      "loss": 0.8696,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7328699261781183e-05,
+      "loss": 0.8896,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7324451824677812e-05,
+      "loss": 0.9409,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7320201534924796e-05,
+      "loss": 0.8228,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7315948394177495e-05,
+      "loss": 0.8823,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731169240409238e-05,
+      "loss": 0.877,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730743356632702e-05,
+      "loss": 0.8843,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7303171882540112e-05,
+      "loss": 0.8574,
+      "step": 1361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729890735439144e-05,
+      "loss": 0.8198,
+      "step": 1362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294639983541918e-05,
+      "loss": 0.8687,
+      "step": 1363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7290369771653552e-05,
+      "loss": 0.8765,
+      "step": 1364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728609672038946e-05,
+      "loss": 0.8867,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7281820831413862e-05,
+      "loss": 0.8281,
+      "step": 1366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7277542106392085e-05,
+      "loss": 0.9253,
+      "step": 1367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273260546990567e-05,
+      "loss": 0.8511,
+      "step": 1368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7268976154876838e-05,
+      "loss": 0.8438,
+      "step": 1369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726468893171954e-05,
+      "loss": 0.9019,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7260398879188418e-05,
+      "loss": 0.8799,
+      "step": 1371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7256105998954313e-05,
+      "loss": 0.8042,
+      "step": 1372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725181029268917e-05,
+      "loss": 0.8979,
+      "step": 1373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7247511762066033e-05,
+      "loss": 0.8711,
+      "step": 1374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724321040875905e-05,
+      "loss": 0.8872,
+      "step": 1375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238906234443475e-05,
+      "loss": 0.8584,
+      "step": 1376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234599240795634e-05,
+      "loss": 0.8545,
+      "step": 1377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7230289429492986e-05,
+      "loss": 0.8672,
+      "step": 1378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722597680221406e-05,
+      "loss": 0.7881,
+      "step": 1379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72216613606385e-05,
+      "loss": 0.8921,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721734310644703e-05,
+      "loss": 0.9092,
+      "step": 1381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7213022041321485e-05,
+      "loss": 0.8755,
+      "step": 1382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7208698166944784e-05,
+      "loss": 0.9019,
+      "step": 1383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7204371485000947e-05,
+      "loss": 0.8604,
+      "step": 1384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7200041997175074e-05,
+      "loss": 0.8667,
+      "step": 1385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719570970515338e-05,
+      "loss": 0.8999,
+      "step": 1386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7191374610623156e-05,
+      "loss": 0.8643,
+      "step": 1387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7187036715272783e-05,
+      "loss": 0.8564,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7182696020791748e-05,
+      "loss": 0.8398,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717835252887061e-05,
+      "loss": 0.8789,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7174006241201033e-05,
+      "loss": 0.8657,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716965715947576e-05,
+      "loss": 0.8633,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7165305285388622e-05,
+      "loss": 0.8647,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160950620634546e-05,
+      "loss": 0.877,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7156593166909535e-05,
+      "loss": 0.8711,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715223292591069e-05,
+      "loss": 0.8584,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7147869899336186e-05,
+      "loss": 0.8589,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714350408888529e-05,
+      "loss": 0.8721,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7139135496258354e-05,
+      "loss": 0.8462,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7134764123156807e-05,
+      "loss": 0.7964,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7130389971283164e-05,
+      "loss": 0.8882,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7126013042341033e-05,
+      "loss": 0.8555,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712163333803508e-05,
+      "loss": 0.8584,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117250860071074e-05,
+      "loss": 0.8301,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7112865610155855e-05,
+      "loss": 0.8286,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7108477589997345e-05,
+      "loss": 0.9287,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7104086801304538e-05,
+      "loss": 0.8848,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7099693245787517e-05,
+      "loss": 0.8276,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7095296925157433e-05,
+      "loss": 0.9385,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709089784112652e-05,
+      "loss": 0.8887,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708649599540809e-05,
+      "loss": 0.9253,
+      "step": 1411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708209138971652e-05,
+      "loss": 0.8916,
+      "step": 1412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7077684025767278e-05,
+      "loss": 0.8491,
+      "step": 1413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707327390527689e-05,
+      "loss": 0.894,
+      "step": 1414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7068861029962966e-05,
+      "loss": 0.8721,
+      "step": 1415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706444540154418e-05,
+      "loss": 0.8901,
+      "step": 1416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7060027021740296e-05,
+      "loss": 0.855,
+      "step": 1417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7055605892272127e-05,
+      "loss": 0.8945,
+      "step": 1418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705118201486157e-05,
+      "loss": 0.8008,
+      "step": 1419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7046755391231594e-05,
+      "loss": 0.9175,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7042326023106228e-05,
+      "loss": 0.8345,
+      "step": 1421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037893912210577e-05,
+      "loss": 0.8975,
+      "step": 1422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7033459060270812e-05,
+      "loss": 0.8882,
+      "step": 1423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7029021469014174e-05,
+      "loss": 0.8599,
+      "step": 1424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702458114016897e-05,
+      "loss": 0.8657,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7020138075464568e-05,
+      "loss": 0.8467,
+      "step": 1426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7015692276631405e-05,
+      "loss": 0.8911,
+      "step": 1427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7011243745400986e-05,
+      "loss": 0.8154,
+      "step": 1428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7006792483505884e-05,
+      "loss": 0.853,
+      "step": 1429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7002338492679714e-05,
+      "loss": 0.8213,
+      "step": 1430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6997881774657184e-05,
+      "loss": 0.8755,
+      "step": 1431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6993422331174042e-05,
+      "loss": 0.4851,
+      "step": 1432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6988960163967103e-05,
+      "loss": 0.8706,
+      "step": 1433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6984495274774245e-05,
+      "loss": 0.9019,
+      "step": 1434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698002766533441e-05,
+      "loss": 0.856,
+      "step": 1435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975557337387595e-05,
+      "loss": 0.8604,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6971084292674847e-05,
+      "loss": 0.895,
+      "step": 1437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6966608532938297e-05,
+      "loss": 0.8149,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6962130059921096e-05,
+      "loss": 0.8838,
+      "step": 1439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6957648875367483e-05,
+      "loss": 0.8916,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6953164981022745e-05,
+      "loss": 0.8018,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948678378633213e-05,
+      "loss": 0.8589,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944189069946286e-05,
+      "loss": 0.8647,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6939697056710416e-05,
+      "loss": 0.8853,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6935202340675095e-05,
+      "loss": 0.8423,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6930704923590887e-05,
+      "loss": 0.8916,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6926204807209393e-05,
+      "loss": 0.4419,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692170199328327e-05,
+      "loss": 0.5552,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6917196483566224e-05,
+      "loss": 0.832,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691268827981302e-05,
+      "loss": 0.9585,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690817738377946e-05,
+      "loss": 0.8623,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6903663797222403e-05,
+      "loss": 0.8857,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689914752189975e-05,
+      "loss": 0.9072,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894628559570455e-05,
+      "loss": 0.8325,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6890106911994512e-05,
+      "loss": 0.8022,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6885582580932967e-05,
+      "loss": 0.8369,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6881055568147905e-05,
+      "loss": 0.8823,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6876525875402464e-05,
+      "loss": 0.8408,
+      "step": 1458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6871993504460817e-05,
+      "loss": 0.7954,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6867458457088183e-05,
+      "loss": 0.8633,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6862920735050823e-05,
+      "loss": 0.8965,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6858380340116047e-05,
+      "loss": 0.7993,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853837274052193e-05,
+      "loss": 0.8867,
+      "step": 1463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6849291538628648e-05,
+      "loss": 0.9097,
+      "step": 1464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6844743135615842e-05,
+      "loss": 0.8574,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684019206678523e-05,
+      "loss": 0.8691,
+      "step": 1466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6835638333909318e-05,
+      "loss": 0.9082,
+      "step": 1467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683108193876165e-05,
+      "loss": 0.8203,
+      "step": 1468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6826522883116794e-05,
+      "loss": 0.8447,
+      "step": 1469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682196116875037e-05,
+      "loss": 0.9019,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817396797439017e-05,
+      "loss": 0.8726,
+      "step": 1471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681282977096043e-05,
+      "loss": 0.874,
+      "step": 1472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6808260091093315e-05,
+      "loss": 0.8477,
+      "step": 1473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803687759617427e-05,
+      "loss": 0.8857,
+      "step": 1474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6799112778313546e-05,
+      "loss": 0.8901,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6794535148963492e-05,
+      "loss": 0.9033,
+      "step": 1476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6789954873350113e-05,
+      "loss": 0.8091,
+      "step": 1477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6785371953257277e-05,
+      "loss": 0.8594,
+      "step": 1478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.67807863904699e-05,
+      "loss": 0.856,
+      "step": 1479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776198186773914e-05,
+      "loss": 0.8706,
+      "step": 1480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677160734395628e-05,
+      "loss": 0.8701,
+      "step": 1481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6767013863805e-05,
+      "loss": 0.7705,
+      "step": 1482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6762417748109085e-05,
+      "loss": 0.8706,
+      "step": 1483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6757818998658587e-05,
+      "loss": 0.8755,
+      "step": 1484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6753217617244575e-05,
+      "loss": 0.8618,
+      "step": 1485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674861360565915e-05,
+      "loss": 0.8662,
+      "step": 1486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674400696569543e-05,
+      "loss": 0.8408,
+      "step": 1487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6739397699147557e-05,
+      "loss": 0.8623,
+      "step": 1488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6734785807810707e-05,
+      "loss": 0.9199,
+      "step": 1489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6730171293481065e-05,
+      "loss": 0.9009,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6725554157955845e-05,
+      "loss": 0.8672,
+      "step": 1491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672093440303328e-05,
+      "loss": 0.8823,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671631203051262e-05,
+      "loss": 0.8174,
+      "step": 1493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6711687042194142e-05,
+      "loss": 0.8105,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6707059439879137e-05,
+      "loss": 0.895,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6702429225369908e-05,
+      "loss": 0.876,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697796400469788e-05,
+      "loss": 0.8638,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669316096698312e-05,
+      "loss": 0.8872,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668852292671526e-05,
+      "loss": 0.7847,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6683882281472584e-05,
+      "loss": 0.8369,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679239033062486e-05,
+      "loss": 0.8574,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667459318329336e-05,
+      "loss": 0.8901,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666994473397463e-05,
+      "loss": 0.8813,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6665293686916722e-05,
+      "loss": 0.7969,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6660640043931076e-05,
+      "loss": 0.8716,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6655983806830142e-05,
+      "loss": 0.8438,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6651324977427384e-05,
+      "loss": 0.8682,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6646663557537272e-05,
+      "loss": 0.9175,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664199954897529e-05,
+      "loss": 0.8848,
+      "step": 1509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663733295355792e-05,
+      "loss": 0.8296,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663266377310266e-05,
+      "loss": 0.8257,
+      "step": 1511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6627992009428014e-05,
+      "loss": 0.9106,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623317664353494e-05,
+      "loss": 0.7964,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661864073969961e-05,
+      "loss": 0.4944,
+      "step": 1514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661396123728788e-05,
+      "loss": 0.8716,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6609279158940833e-05,
+      "loss": 0.8838,
+      "step": 1516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660459450648199e-05,
+      "loss": 0.853,
+      "step": 1517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659990728173588e-05,
+      "loss": 0.8975,
+      "step": 1518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6595217486528034e-05,
+      "loss": 0.8521,
+      "step": 1519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6590525122684985e-05,
+      "loss": 0.9004,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6585830192034256e-05,
+      "loss": 0.876,
+      "step": 1521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6581132696404398e-05,
+      "loss": 0.8687,
+      "step": 1522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6576432637624922e-05,
+      "loss": 0.8394,
+      "step": 1523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657173001752637e-05,
+      "loss": 0.9199,
+      "step": 1524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656702483794026e-05,
+      "loss": 0.8257,
+      "step": 1525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6562317100699118e-05,
+      "loss": 0.8721,
+      "step": 1526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6557606807636468e-05,
+      "loss": 0.8774,
+      "step": 1527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655289396058682e-05,
+      "loss": 0.8911,
+      "step": 1528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548178561385684e-05,
+      "loss": 0.8594,
+      "step": 1529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6543460611869563e-05,
+      "loss": 0.8457,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.653874011387596e-05,
+      "loss": 0.8994,
+      "step": 1531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6534017069243352e-05,
+      "loss": 0.8428,
+      "step": 1532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6529291479811235e-05,
+      "loss": 0.8892,
+      "step": 1533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652456334742007e-05,
+      "loss": 0.8525,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6519832673911324e-05,
+      "loss": 0.8706,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651509946112745e-05,
+      "loss": 0.8662,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651036371091189e-05,
+      "loss": 0.895,
+      "step": 1537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650562542510907e-05,
+      "loss": 0.875,
+      "step": 1538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6500884605564413e-05,
+      "loss": 0.8623,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6496141254124313e-05,
+      "loss": 0.8584,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6491395372636172e-05,
+      "loss": 0.8691,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648664696294836e-05,
+      "loss": 0.8726,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6481896026910235e-05,
+      "loss": 0.8442,
+      "step": 1543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6477142566372148e-05,
+      "loss": 0.8521,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6472386583185413e-05,
+      "loss": 0.8159,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6467628079202356e-05,
+      "loss": 0.8237,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6462867056276257e-05,
+      "loss": 0.8887,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6458103516261393e-05,
+      "loss": 0.8369,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453337461013018e-05,
+      "loss": 0.8574,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6448568892387363e-05,
+      "loss": 0.8472,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644379781224164e-05,
+      "loss": 0.8857,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439024222434032e-05,
+      "loss": 0.8403,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643424812482372e-05,
+      "loss": 0.8994,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6429469521270836e-05,
+      "loss": 0.8682,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64246884136365e-05,
+      "loss": 0.8096,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6419904803782818e-05,
+      "loss": 0.5093,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415118693572848e-05,
+      "loss": 0.5081,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6410330084870634e-05,
+      "loss": 0.8555,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6405538979541204e-05,
+      "loss": 0.8848,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640074537945053e-05,
+      "loss": 0.8491,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6395949286465588e-05,
+      "loss": 0.873,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6391150702454296e-05,
+      "loss": 0.5393,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386349629285564e-05,
+      "loss": 0.8657,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6381546068829262e-05,
+      "loss": 0.8608,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6376740022956227e-05,
+      "loss": 0.9233,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6371931493538268e-05,
+      "loss": 0.8911,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636712048244816e-05,
+      "loss": 0.8989,
+      "step": 1567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6362306991559645e-05,
+      "loss": 0.8965,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6357491022747428e-05,
+      "loss": 0.7666,
+      "step": 1569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6352672577887185e-05,
+      "loss": 0.8442,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634785165885555e-05,
+      "loss": 0.8657,
+      "step": 1571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6343028267530124e-05,
+      "loss": 0.8848,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633820240578947e-05,
+      "loss": 0.8264,
+      "step": 1573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6333374075513117e-05,
+      "loss": 0.4756,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6328543278581542e-05,
+      "loss": 0.791,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63237100168762e-05,
+      "loss": 0.8022,
+      "step": 1576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6318874292279502e-05,
+      "loss": 0.8735,
+      "step": 1577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6314036106674803e-05,
+      "loss": 0.8965,
+      "step": 1578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630919546194644e-05,
+      "loss": 0.896,
+      "step": 1579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6304352359979687e-05,
+      "loss": 0.8989,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629950680266079e-05,
+      "loss": 0.8125,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629465879187694e-05,
+      "loss": 0.8906,
+      "step": 1582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6289808329516294e-05,
+      "loss": 0.8779,
+      "step": 1583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6284955417467955e-05,
+      "loss": 0.8608,
+      "step": 1584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6280100057621985e-05,
+      "loss": 0.9658,
+      "step": 1585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62752422518694e-05,
+      "loss": 0.9092,
+      "step": 1586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627038200210216e-05,
+      "loss": 0.8271,
+      "step": 1587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626551931021319e-05,
+      "loss": 0.8247,
+      "step": 1588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6260654178096357e-05,
+      "loss": 0.8599,
+      "step": 1589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6255786607646474e-05,
+      "loss": 0.8657,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250916600759325e-05,
+      "loss": 0.8989,
+      "step": 1591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624604415933162e-05,
+      "loss": 0.833,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6241169285261027e-05,
+      "loss": 0.8574,
+      "step": 1593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6236291980446157e-05,
+      "loss": 0.936,
+      "step": 1594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6231412246786574e-05,
+      "loss": 0.8237,
+      "step": 1595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6226530086182787e-05,
+      "loss": 0.8989,
+      "step": 1596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6221645500536248e-05,
+      "loss": 0.8936,
+      "step": 1597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6216758491749347e-05,
+      "loss": 0.7988,
+      "step": 1598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6211869061725427e-05,
+      "loss": 0.8613,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6206977212368774e-05,
+      "loss": 0.8506,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6202082945584613e-05,
+      "loss": 0.8379,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619718626327911e-05,
+      "loss": 0.8877,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619228716735937e-05,
+      "loss": 0.8193,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6187385659733447e-05,
+      "loss": 0.8784,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6182481742310325e-05,
+      "loss": 0.8472,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6177575416999928e-05,
+      "loss": 0.8765,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6172666685713122e-05,
+      "loss": 0.8364,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6167755550361702e-05,
+      "loss": 0.875,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616284201285841e-05,
+      "loss": 0.855,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6157926075116922e-05,
+      "loss": 0.8413,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6153007739051837e-05,
+      "loss": 0.8525,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148087006578698e-05,
+      "loss": 0.8862,
+      "step": 1612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6143163879613982e-05,
+      "loss": 0.7852,
+      "step": 1613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61382383600751e-05,
+      "loss": 0.8447,
+      "step": 1614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6133310449880385e-05,
+      "loss": 0.5354,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6128380150949106e-05,
+      "loss": 0.9121,
+      "step": 1616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6123447465201466e-05,
+      "loss": 0.8398,
+      "step": 1617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61185123945586e-05,
+      "loss": 0.9102,
+      "step": 1618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6113574940942558e-05,
+      "loss": 0.8511,
+      "step": 1619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610863510627633e-05,
+      "loss": 0.9067,
+      "step": 1620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6103692892483832e-05,
+      "loss": 0.8804,
+      "step": 1621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6098748301489906e-05,
+      "loss": 0.8389,
+      "step": 1622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6093801335220317e-05,
+      "loss": 0.4773,
+      "step": 1623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608885199560175e-05,
+      "loss": 0.915,
+      "step": 1624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6083900284561837e-05,
+      "loss": 0.8433,
+      "step": 1625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6078946204029104e-05,
+      "loss": 0.8184,
+      "step": 1626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6073989755933018e-05,
+      "loss": 0.855,
+      "step": 1627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6069030942203963e-05,
+      "loss": 0.9028,
+      "step": 1628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606406976477325e-05,
+      "loss": 0.9077,
+      "step": 1629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60591062255731e-05,
+      "loss": 0.8579,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605414032653666e-05,
+      "loss": 0.8379,
+      "step": 1631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6049172069597996e-05,
+      "loss": 0.9443,
+      "step": 1632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6044201456692095e-05,
+      "loss": 0.8853,
+      "step": 1633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6039228489754857e-05,
+      "loss": 0.5083,
+      "step": 1634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6034253170723097e-05,
+      "loss": 0.8237,
+      "step": 1635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602927550153455e-05,
+      "loss": 0.8472,
+      "step": 1636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6024295484127873e-05,
+      "loss": 0.856,
+      "step": 1637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601931312044262e-05,
+      "loss": 0.8491,
+      "step": 1638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6014328412419275e-05,
+      "loss": 0.8516,
+      "step": 1639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600934136199923e-05,
+      "loss": 0.8013,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6004351971124783e-05,
+      "loss": 0.9087,
+      "step": 1641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5999360241739156e-05,
+      "loss": 0.8306,
+      "step": 1642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5994366175786466e-05,
+      "loss": 0.8701,
+      "step": 1643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5989369775211755e-05,
+      "loss": 0.8262,
+      "step": 1644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5984371041960968e-05,
+      "loss": 0.8491,
+      "step": 1645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597936997798096e-05,
+      "loss": 0.8276,
+      "step": 1646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5974366585219486e-05,
+      "loss": 0.8003,
+      "step": 1647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5969360865625216e-05,
+      "loss": 0.8877,
+      "step": 1648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596435282114773e-05,
+      "loss": 0.9043,
+      "step": 1649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59593424537375e-05,
+      "loss": 0.4951,
+      "step": 1650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5954329765345917e-05,
+      "loss": 0.8457,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949314757925266e-05,
+      "loss": 0.8979,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5944297433428742e-05,
+      "loss": 0.8984,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5939277793810434e-05,
+      "loss": 0.8623,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5934255841025344e-05,
+      "loss": 0.8521,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5929231577029365e-05,
+      "loss": 0.7871,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5924205003779294e-05,
+      "loss": 0.8984,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591917612323283e-05,
+      "loss": 0.8979,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5914144937348565e-05,
+      "loss": 0.4717,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5909111448085992e-05,
+      "loss": 0.8604,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5904075657405503e-05,
+      "loss": 0.8872,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5899037567268383e-05,
+      "loss": 0.8789,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5893997179636812e-05,
+      "loss": 0.8394,
+      "step": 1663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5888954496473868e-05,
+      "loss": 0.9019,
+      "step": 1664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5883909519743526e-05,
+      "loss": 0.8579,
+      "step": 1665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587886225141064e-05,
+      "loss": 0.4897,
+      "step": 1666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587381269344098e-05,
+      "loss": 0.8105,
+      "step": 1667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5868760847801183e-05,
+      "loss": 0.4675,
+      "step": 1668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586370671645879e-05,
+      "loss": 0.8416,
+      "step": 1669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585865030138223e-05,
+      "loss": 0.8237,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585359160454083e-05,
+      "loss": 0.8672,
+      "step": 1671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5848530627904786e-05,
+      "loss": 0.8794,
+      "step": 1672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5843467373445203e-05,
+      "loss": 0.9263,
+      "step": 1673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5838401843134058e-05,
+      "loss": 0.8838,
+      "step": 1674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5833334038944223e-05,
+      "loss": 0.8496,
+      "step": 1675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5828263962849444e-05,
+      "loss": 0.4705,
+      "step": 1676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582319161682437e-05,
+      "loss": 0.8687,
+      "step": 1677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5818117002844523e-05,
+      "loss": 0.8477,
+      "step": 1678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581304012288631e-05,
+      "loss": 0.8945,
+      "step": 1679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5807960978927015e-05,
+      "loss": 0.8853,
+      "step": 1680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5802879572944815e-05,
+      "loss": 0.8408,
+      "step": 1681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5797795906918752e-05,
+      "loss": 0.8218,
+      "step": 1682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5792709982828775e-05,
+      "loss": 0.8999,
+      "step": 1683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578762180265568e-05,
+      "loss": 0.832,
+      "step": 1684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5782531368381166e-05,
+      "loss": 0.8569,
+      "step": 1685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.57774386819878e-05,
+      "loss": 0.8433,
+      "step": 1686
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5772343745459025e-05,
+      "loss": 0.8423,
+      "step": 1687
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5767246560779162e-05,
+      "loss": 0.9014,
+      "step": 1688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762147129933412e-05,
+      "loss": 0.8369,
+      "step": 1689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5757045454907848e-05,
+      "loss": 0.8623,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5751941537689406e-05,
+      "loss": 0.509,
+      "step": 1691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574683538026592e-05,
+      "loss": 0.8223,
+      "step": 1692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574172698462607e-05,
+      "loss": 0.918,
+      "step": 1693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5736616352759425e-05,
+      "loss": 0.7881,
+      "step": 1694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5731503486656422e-05,
+      "loss": 0.8721,
+      "step": 1695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572638838830836e-05,
+      "loss": 0.8608,
+      "step": 1696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5721271059707416e-05,
+      "loss": 0.9033,
+      "step": 1697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5716151502846638e-05,
+      "loss": 0.7646,
+      "step": 1698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5711029719719926e-05,
+      "loss": 0.8721,
+      "step": 1699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5705905712322062e-05,
+      "loss": 0.8125,
+      "step": 1700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5700779482648694e-05,
+      "loss": 0.9043,
+      "step": 1701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5695651032696328e-05,
+      "loss": 0.8618,
+      "step": 1702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5690520364462332e-05,
+      "loss": 0.8237,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568538747994496e-05,
+      "loss": 0.8325,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56802523811433e-05,
+      "loss": 0.833,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567511507005732e-05,
+      "loss": 0.8882,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566997554868784e-05,
+      "loss": 0.8643,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5664833819036557e-05,
+      "loss": 0.876,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565968988310601e-05,
+      "loss": 0.8228,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5654543742899606e-05,
+      "loss": 0.8511,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5649395400421608e-05,
+      "loss": 0.8809,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5644244857677135e-05,
+      "loss": 0.8701,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5639092116672174e-05,
+      "loss": 0.8364,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5633937179413553e-05,
+      "loss": 0.8262,
+      "step": 1714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5628780047908963e-05,
+      "loss": 0.8394,
+      "step": 1715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562362072416695e-05,
+      "loss": 0.8484,
+      "step": 1716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5618459210196915e-05,
+      "loss": 0.8008,
+      "step": 1717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5613295508009106e-05,
+      "loss": 0.8174,
+      "step": 1718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560812961961463e-05,
+      "loss": 0.8721,
+      "step": 1719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5602961547025434e-05,
+      "loss": 0.8008,
+      "step": 1720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5597791292254333e-05,
+      "loss": 0.812,
+      "step": 1721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559261885731498e-05,
+      "loss": 0.8374,
+      "step": 1722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5587444244221882e-05,
+      "loss": 0.9028,
+      "step": 1723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5582267454990386e-05,
+      "loss": 0.8081,
+      "step": 1724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55770884916367e-05,
+      "loss": 0.8682,
+      "step": 1725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5571907356177868e-05,
+      "loss": 0.8354,
+      "step": 1726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5566724050631784e-05,
+      "loss": 0.8242,
+      "step": 1727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5561538577017188e-05,
+      "loss": 0.8623,
+      "step": 1728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5556350937353665e-05,
+      "loss": 0.8848,
+      "step": 1729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555116113366163e-05,
+      "loss": 0.8418,
+      "step": 1730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554596916796237e-05,
+      "loss": 0.8828,
+      "step": 1731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5540775042277986e-05,
+      "loss": 0.8867,
+      "step": 1732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553557875863143e-05,
+      "loss": 0.8623,
+      "step": 1733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55303803190465e-05,
+      "loss": 0.8481,
+      "step": 1734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552517972554783e-05,
+      "loss": 0.8604,
+      "step": 1735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5519976980160883e-05,
+      "loss": 0.9102,
+      "step": 1736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551477208491198e-05,
+      "loss": 0.8423,
+      "step": 1737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5509565041828265e-05,
+      "loss": 0.855,
+      "step": 1738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5504355852937716e-05,
+      "loss": 0.8267,
+      "step": 1739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5499144520269157e-05,
+      "loss": 0.8433,
+      "step": 1740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5493931045852244e-05,
+      "loss": 0.8481,
+      "step": 1741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5488715431717464e-05,
+      "loss": 0.9082,
+      "step": 1742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548349767989614e-05,
+      "loss": 0.4917,
+      "step": 1743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547827779242042e-05,
+      "loss": 0.8496,
+      "step": 1744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54730557713233e-05,
+      "loss": 0.8628,
+      "step": 1745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546783161863859e-05,
+      "loss": 0.8608,
+      "step": 1746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546260533640094e-05,
+      "loss": 0.8188,
+      "step": 1747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545737692664583e-05,
+      "loss": 0.8394,
+      "step": 1748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545214639140956e-05,
+      "loss": 0.8491,
+      "step": 1749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5446913732729268e-05,
+      "loss": 0.8672,
+      "step": 1750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544167895264291e-05,
+      "loss": 0.8276,
+      "step": 1751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5436442053189276e-05,
+      "loss": 0.8311,
+      "step": 1752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543120303640798e-05,
+      "loss": 0.9033,
+      "step": 1753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5425961904339447e-05,
+      "loss": 0.8135,
+      "step": 1754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5420718659024954e-05,
+      "loss": 0.8008,
+      "step": 1755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5415473302506573e-05,
+      "loss": 0.8174,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541022583682721e-05,
+      "loss": 0.8838,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54049762640306e-05,
+      "loss": 0.8413,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539972458616128e-05,
+      "loss": 0.9067,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5394470805264625e-05,
+      "loss": 0.8921,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538921492338682e-05,
+      "loss": 0.8721,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5383956942574872e-05,
+      "loss": 0.8247,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53786968648766e-05,
+      "loss": 0.8184,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5373434692340645e-05,
+      "loss": 0.8315,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5368170427016462e-05,
+      "loss": 0.8467,
+      "step": 1765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536290407095432e-05,
+      "loss": 0.8843,
+      "step": 1766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5357635626205304e-05,
+      "loss": 0.8052,
+      "step": 1767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5352365094821317e-05,
+      "loss": 0.8545,
+      "step": 1768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5347092478855056e-05,
+      "loss": 0.812,
+      "step": 1769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534181778036006e-05,
+      "loss": 0.8379,
+      "step": 1770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533654100139065e-05,
+      "loss": 0.876,
+      "step": 1771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5331262144001974e-05,
+      "loss": 0.8667,
+      "step": 1772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532598121024999e-05,
+      "loss": 0.8057,
+      "step": 1773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5320698202191452e-05,
+      "loss": 0.8579,
+      "step": 1774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315413121883933e-05,
+      "loss": 0.8516,
+      "step": 1775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531012597138581e-05,
+      "loss": 0.8115,
+      "step": 1776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530483675275627e-05,
+      "loss": 0.8486,
+      "step": 1777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299545468055293e-05,
+      "loss": 0.7886,
+      "step": 1778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5294252119343682e-05,
+      "loss": 0.5005,
+      "step": 1779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5288956708683026e-05,
+      "loss": 0.8013,
+      "step": 1780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5283659238135732e-05,
+      "loss": 0.8486,
+      "step": 1781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5278359709764997e-05,
+      "loss": 0.8638,
+      "step": 1782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527305812563483e-05,
+      "loss": 0.8228,
+      "step": 1783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5267754487810032e-05,
+      "loss": 0.833,
+      "step": 1784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5262448798356206e-05,
+      "loss": 0.8599,
+      "step": 1785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257141059339761e-05,
+      "loss": 0.8574,
+      "step": 1786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5251831272827898e-05,
+      "loss": 0.8911,
+      "step": 1787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524651944088861e-05,
+      "loss": 0.875,
+      "step": 1788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52412055655907e-05,
+      "loss": 0.8423,
+      "step": 1789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.523588964900375e-05,
+      "loss": 0.8442,
+      "step": 1790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5230571693198157e-05,
+      "loss": 0.8433,
+      "step": 1791
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5225251700245103e-05,
+      "loss": 0.4678,
+      "step": 1792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521992967221655e-05,
+      "loss": 0.8462,
+      "step": 1793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214605611185275e-05,
+      "loss": 0.7764,
+      "step": 1794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209279519224838e-05,
+      "loss": 0.8647,
+      "step": 1795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5203951398409578e-05,
+      "loss": 0.8789,
+      "step": 1796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198621250814646e-05,
+      "loss": 0.8945,
+      "step": 1797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5193289078515965e-05,
+      "loss": 0.7881,
+      "step": 1798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5187954883590257e-05,
+      "loss": 0.4985,
+      "step": 1799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5182618668115022e-05,
+      "loss": 0.812,
+      "step": 1800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5177280434168558e-05,
+      "loss": 0.8574,
+      "step": 1801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5171940183829938e-05,
+      "loss": 0.8706,
+      "step": 1802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166597919179034e-05,
+      "loss": 0.8564,
+      "step": 1803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5161253642296485e-05,
+      "loss": 0.8135,
+      "step": 1804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155907355263732e-05,
+      "loss": 0.8481,
+      "step": 1805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5150559060162983e-05,
+      "loss": 0.8682,
+      "step": 1806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514520875907724e-05,
+      "loss": 0.8306,
+      "step": 1807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5139856454090275e-05,
+      "loss": 0.8696,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5134502147286654e-05,
+      "loss": 0.8296,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5129145840751714e-05,
+      "loss": 0.918,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5123787536571572e-05,
+      "loss": 0.8735,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118427236833117e-05,
+      "loss": 0.8438,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5113064943624033e-05,
+      "loss": 0.9077,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5107700659032757e-05,
+      "loss": 0.9038,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5102334385148525e-05,
+      "loss": 0.8159,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509696612406133e-05,
+      "loss": 0.8774,
+      "step": 1816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5091595877861942e-05,
+      "loss": 0.8604,
+      "step": 1817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086223648641913e-05,
+      "loss": 0.8833,
+      "step": 1818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5080849438493563e-05,
+      "loss": 0.8438,
+      "step": 1819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5075473249509978e-05,
+      "loss": 0.8125,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5070095083785019e-05,
+      "loss": 0.8799,
+      "step": 1821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506471494341332e-05,
+      "loss": 0.855,
+      "step": 1822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5059332830490276e-05,
+      "loss": 0.8164,
+      "step": 1823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5053948747112061e-05,
+      "loss": 0.9072,
+      "step": 1824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5048562695375611e-05,
+      "loss": 0.8657,
+      "step": 1825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043174677378618e-05,
+      "loss": 0.834,
+      "step": 1826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5037784695219562e-05,
+      "loss": 0.8711,
+      "step": 1827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5032392750997668e-05,
+      "loss": 0.8838,
+      "step": 1828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5026998846812939e-05,
+      "loss": 0.8491,
+      "step": 1829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5021602984766128e-05,
+      "loss": 0.8228,
+      "step": 1830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5016205166958768e-05,
+      "loss": 0.9048,
+      "step": 1831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5010805395493133e-05,
+      "loss": 0.8433,
+      "step": 1832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005403672472274e-05,
+      "loss": 0.8408,
+      "step": 1833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.7554,
+      "step": 1834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4994594380180873e-05,
+      "loss": 0.8306,
+      "step": 1835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498918681512021e-05,
+      "loss": 0.7998,
+      "step": 1836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4983777306924105e-05,
+      "loss": 0.8223,
+      "step": 1837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4978365857699384e-05,
+      "loss": 0.9053,
+      "step": 1838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4972952469553645e-05,
+      "loss": 0.8691,
+      "step": 1839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967537144595241e-05,
+      "loss": 0.853,
+      "step": 1840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4962119884933266e-05,
+      "loss": 0.8843,
+      "step": 1841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4956700692677583e-05,
+      "loss": 0.8745,
+      "step": 1842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4951279569938803e-05,
+      "loss": 0.8711,
+      "step": 1843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4945856518828283e-05,
+      "loss": 0.9277,
+      "step": 1844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4940431541458137e-05,
+      "loss": 0.8828,
+      "step": 1845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4935004639941228e-05,
+      "loss": 0.8843,
+      "step": 1846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4929575816391166e-05,
+      "loss": 0.8857,
+      "step": 1847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4924145072922316e-05,
+      "loss": 0.8374,
+      "step": 1848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491871241164978e-05,
+      "loss": 0.8555,
+      "step": 1849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913277834689418e-05,
+      "loss": 0.8423,
+      "step": 1850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4907841344157827e-05,
+      "loss": 0.8613,
+      "step": 1851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4902402942172357e-05,
+      "loss": 0.8325,
+      "step": 1852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4896962630851099e-05,
+      "loss": 0.8286,
+      "step": 1853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4891520412312887e-05,
+      "loss": 0.8354,
+      "step": 1854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4886076288677298e-05,
+      "loss": 0.5686,
+      "step": 1855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488063026206465e-05,
+      "loss": 0.8833,
+      "step": 1856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4875182334596005e-05,
+      "loss": 0.8721,
+      "step": 1857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869732508393168e-05,
+      "loss": 0.8218,
+      "step": 1858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4864280785578674e-05,
+      "loss": 0.834,
+      "step": 1859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4858827168275798e-05,
+      "loss": 0.8154,
+      "step": 1860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4853371658608569e-05,
+      "loss": 0.8901,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847914258701732e-05,
+      "loss": 0.8403,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842454970680775e-05,
+      "loss": 0.876,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836993796671934e-05,
+      "loss": 0.853,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483153073880216e-05,
+      "loss": 0.8672,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482606579919915e-05,
+      "loss": 0.8164,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4820598979991332e-05,
+      "loss": 0.9302,
+      "step": 1867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4815130283307859e-05,
+      "loss": 0.9146,
+      "step": 1868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809659711278625e-05,
+      "loss": 0.8867,
+      "step": 1869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4804187266034254e-05,
+      "loss": 0.8574,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4798712949706087e-05,
+      "loss": 0.8389,
+      "step": 1871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793236764426211e-05,
+      "loss": 0.8911,
+      "step": 1872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4787758712327427e-05,
+      "loss": 0.8599,
+      "step": 1873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4782278795543268e-05,
+      "loss": 0.8364,
+      "step": 1874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4776797016207998e-05,
+      "loss": 0.8945,
+      "step": 1875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.47713133764566e-05,
+      "loss": 0.8535,
+      "step": 1876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4765827878424783e-05,
+      "loss": 0.8853,
+      "step": 1877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760340524248982e-05,
+      "loss": 0.8105,
+      "step": 1878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4754851316066354e-05,
+      "loss": 0.8789,
+      "step": 1879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4749360256014774e-05,
+      "loss": 0.8184,
+      "step": 1880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4743867346232841e-05,
+      "loss": 0.936,
+      "step": 1881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738372588859884e-05,
+      "loss": 0.877,
+      "step": 1882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4732875986035932e-05,
+      "loss": 0.8809,
+      "step": 1883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4727377539901748e-05,
+      "loss": 0.9121,
+      "step": 1884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4721877252598812e-05,
+      "loss": 0.8467,
+      "step": 1885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471637512626931e-05,
+      "loss": 0.5051,
+      "step": 1886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4710871163056157e-05,
+      "loss": 0.8657,
+      "step": 1887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470536536510298e-05,
+      "loss": 0.8286,
+      "step": 1888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4699857734554108e-05,
+      "loss": 0.8696,
+      "step": 1889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4694348273554607e-05,
+      "loss": 0.8613,
+      "step": 1890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4688836984250239e-05,
+      "loss": 0.8135,
+      "step": 1891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4683323868787476e-05,
+      "loss": 0.9141,
+      "step": 1892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4677808929313515e-05,
+      "loss": 0.855,
+      "step": 1893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4672292167976254e-05,
+      "loss": 0.8687,
+      "step": 1894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666773586924304e-05,
+      "loss": 0.8682,
+      "step": 1895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466125318830698e-05,
+      "loss": 0.8892,
+      "step": 1896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465573097427431e-05,
+      "loss": 0.8467,
+      "step": 1897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465020694697703e-05,
+      "loss": 0.8843,
+      "step": 1898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4644681108566575e-05,
+      "loss": 0.8374,
+      "step": 1899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4639153461195088e-05,
+      "loss": 0.8193,
+      "step": 1900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4633624007015427e-05,
+      "loss": 0.7817,
+      "step": 1901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4628092748181137e-05,
+      "loss": 0.7847,
+      "step": 1902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4622559686846477e-05,
+      "loss": 0.7964,
+      "step": 1903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4617024825166409e-05,
+      "loss": 0.8662,
+      "step": 1904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611488165296578e-05,
+      "loss": 0.832,
+      "step": 1905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460594970939336e-05,
+      "loss": 0.812,
+      "step": 1906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4600409459613809e-05,
+      "loss": 0.8643,
+      "step": 1907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4594867418115674e-05,
+      "loss": 0.8916,
+      "step": 1908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4589323587057422e-05,
+      "loss": 0.8457,
+      "step": 1909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45837779685982e-05,
+      "loss": 0.8745,
+      "step": 1910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578230564897857e-05,
+      "loss": 0.4741,
+      "step": 1911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4572681378116936e-05,
+      "loss": 0.8359,
+      "step": 1912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4567130410416677e-05,
+      "loss": 0.7769,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4561577663959009e-05,
+      "loss": 0.8369,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455602314090656e-05,
+      "loss": 0.8159,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4550466843422647e-05,
+      "loss": 0.8354,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4544908773671273e-05,
+      "loss": 0.8647,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4539348933817139e-05,
+      "loss": 0.8628,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4533787326025635e-05,
+      "loss": 0.8369,
+      "step": 1919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4528223952462832e-05,
+      "loss": 0.8799,
+      "step": 1920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4522658815295496e-05,
+      "loss": 0.8784,
+      "step": 1921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4517091916691079e-05,
+      "loss": 0.8052,
+      "step": 1922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4511523258817715e-05,
+      "loss": 0.8975,
+      "step": 1923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505952843844222e-05,
+      "loss": 0.8735,
+      "step": 1924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4500380673940116e-05,
+      "loss": 0.8369,
+      "step": 1925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4494806751275578e-05,
+      "loss": 0.897,
+      "step": 1926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4489231078021484e-05,
+      "loss": 0.8848,
+      "step": 1927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4483653656349388e-05,
+      "loss": 0.8564,
+      "step": 1928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478074488431522e-05,
+      "loss": 0.9033,
+      "step": 1929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4472493576440803e-05,
+      "loss": 0.8545,
+      "step": 1930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4466910922550828e-05,
+      "loss": 0.8833,
+      "step": 1931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4461326528935864e-05,
+      "loss": 0.8521,
+      "step": 1932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4455740397770864e-05,
+      "loss": 0.873,
+      "step": 1933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4450152531231458e-05,
+      "loss": 0.8394,
+      "step": 1934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4444562931493945e-05,
+      "loss": 0.9072,
+      "step": 1935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438971600735301e-05,
+      "loss": 0.7935,
+      "step": 1936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4433378541133183e-05,
+      "loss": 0.8521,
+      "step": 1937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4427783754865914e-05,
+      "loss": 0.8838,
+      "step": 1938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4422187244112493e-05,
+      "loss": 0.8716,
+      "step": 1939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4416589011052589e-05,
+      "loss": 0.8257,
+      "step": 1940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4410989057866543e-05,
+      "loss": 0.8599,
+      "step": 1941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4405387386735363e-05,
+      "loss": 0.9204,
+      "step": 1942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4399783999840731e-05,
+      "loss": 0.8647,
+      "step": 1943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439417889936499e-05,
+      "loss": 0.873,
+      "step": 1944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4388572087491161e-05,
+      "loss": 0.8682,
+      "step": 1945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4382963566402924e-05,
+      "loss": 0.8032,
+      "step": 1946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437735333828462e-05,
+      "loss": 0.7954,
+      "step": 1947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437174140532127e-05,
+      "loss": 0.8867,
+      "step": 1948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4366127769698546e-05,
+      "loss": 0.854,
+      "step": 1949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4360512433602785e-05,
+      "loss": 0.8271,
+      "step": 1950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4354895399220989e-05,
+      "loss": 0.7939,
+      "step": 1951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4349276668740827e-05,
+      "loss": 0.8081,
+      "step": 1952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4343656244350612e-05,
+      "loss": 0.8564,
+      "step": 1953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4338034128239338e-05,
+      "loss": 0.8525,
+      "step": 1954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4332410322596641e-05,
+      "loss": 0.8447,
+      "step": 1955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4326784829612818e-05,
+      "loss": 0.8125,
+      "step": 1956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4321157651478834e-05,
+      "loss": 0.8442,
+      "step": 1957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43155287903863e-05,
+      "loss": 0.8306,
+      "step": 1958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430989824852748e-05,
+      "loss": 0.4822,
+      "step": 1959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304266028095302e-05,
+      "loss": 0.8521,
+      "step": 1960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4298632131283344e-05,
+      "loss": 0.8364,
+      "step": 1961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4292996560285834e-05,
+      "loss": 0.8164,
+      "step": 1962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4287359317297655e-05,
+      "loss": 0.8428,
+      "step": 1963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428172040451434e-05,
+      "loss": 0.8701,
+      "step": 1964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276079824132075e-05,
+      "loss": 0.897,
+      "step": 1965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427043757834769e-05,
+      "loss": 0.8662,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4264793669358673e-05,
+      "loss": 0.8477,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4259148099363143e-05,
+      "loss": 0.8247,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4253500870559887e-05,
+      "loss": 0.8271,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4247851985148322e-05,
+      "loss": 0.874,
+      "step": 1970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4242201445328519e-05,
+      "loss": 0.8066,
+      "step": 1971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4236549253301184e-05,
+      "loss": 0.8574,
+      "step": 1972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4230895411267685e-05,
+      "loss": 0.855,
+      "step": 1973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4225239921430008e-05,
+      "loss": 0.4478,
+      "step": 1974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4219582785990796e-05,
+      "loss": 0.9165,
+      "step": 1975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4213924007153334e-05,
+      "loss": 0.873,
+      "step": 1976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4208263587121538e-05,
+      "loss": 0.9297,
+      "step": 1977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4202601528099975e-05,
+      "loss": 0.7988,
+      "step": 1978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196937832293838e-05,
+      "loss": 0.8467,
+      "step": 1979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4191272501908962e-05,
+      "loss": 0.7598,
+      "step": 1980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4185605539151825e-05,
+      "loss": 0.7686,
+      "step": 1981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4179936946229533e-05,
+      "loss": 0.874,
+      "step": 1982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4174266725349825e-05,
+      "loss": 0.835,
+      "step": 1983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4168594878721087e-05,
+      "loss": 0.8013,
+      "step": 1984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416292140855232e-05,
+      "loss": 0.8071,
+      "step": 1985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4157246317053177e-05,
+      "loss": 0.8179,
+      "step": 1986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4151569606433923e-05,
+      "loss": 0.8203,
+      "step": 1987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4145891278905467e-05,
+      "loss": 0.874,
+      "step": 1988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4140211336679346e-05,
+      "loss": 0.7979,
+      "step": 1989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4134529781967721e-05,
+      "loss": 0.8315,
+      "step": 1990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412884661698338e-05,
+      "loss": 0.873,
+      "step": 1991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412316184393975e-05,
+      "loss": 0.8804,
+      "step": 1992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4117475465050868e-05,
+      "loss": 0.8311,
+      "step": 1993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111787482531408e-05,
+      "loss": 0.8286,
+      "step": 1994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4106097898596665e-05,
+      "loss": 0.8345,
+      "step": 1995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4100406715462555e-05,
+      "loss": 0.8589,
+      "step": 1996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4094713935345627e-05,
+      "loss": 0.8267,
+      "step": 1997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4089019560463035e-05,
+      "loss": 0.8276,
+      "step": 1998
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4083323593032567e-05,
+      "loss": 0.8672,
+      "step": 1999
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4077626035272634e-05,
+      "loss": 0.8271,
+      "step": 2000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407192688940225e-05,
+      "loss": 0.875,
+      "step": 2001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4066226157641065e-05,
+      "loss": 0.8477,
+      "step": 2002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406052384220934e-05,
+      "loss": 0.8423,
+      "step": 2003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4054819945327944e-05,
+      "loss": 0.9019,
+      "step": 2004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4049114469218378e-05,
+      "loss": 0.897,
+      "step": 2005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404340741610275e-05,
+      "loss": 0.8755,
+      "step": 2006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4037698788203776e-05,
+      "loss": 0.8633,
+      "step": 2007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4031988587744796e-05,
+      "loss": 0.8354,
+      "step": 2008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402627681694976e-05,
+      "loss": 0.8257,
+      "step": 2009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4020563478043223e-05,
+      "loss": 0.8594,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4014848573250362e-05,
+      "loss": 0.811,
+      "step": 2011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4009132104796953e-05,
+      "loss": 0.8257,
+      "step": 2012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4003414074909385e-05,
+      "loss": 0.7969,
+      "step": 2013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3997694485814654e-05,
+      "loss": 0.8325,
+      "step": 2014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399197333974037e-05,
+      "loss": 0.7964,
+      "step": 2015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3986250638914741e-05,
+      "loss": 0.4841,
+      "step": 2016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3980526385566586e-05,
+      "loss": 0.8501,
+      "step": 2017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3974800581925322e-05,
+      "loss": 0.8843,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3969073230220978e-05,
+      "loss": 0.8389,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3963344332684177e-05,
+      "loss": 0.8643,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3957613891546155e-05,
+      "loss": 0.9375,
+      "step": 2021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3951881909038745e-05,
+      "loss": 0.4497,
+      "step": 2022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3946148387394369e-05,
+      "loss": 0.8413,
+      "step": 2023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940413328846068e-05,
+      "loss": 0.8696,
+      "step": 2024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3934676735627464e-05,
+      "loss": 0.8257,
+      "step": 2025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3928938609972788e-05,
+      "loss": 0.8223,
+      "step": 2026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3923198954116867e-05,
+      "loss": 0.873,
+      "step": 2027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3917457770295117e-05,
+      "loss": 0.833,
+      "step": 2028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3911715060743554e-05,
+      "loss": 0.8076,
+      "step": 2029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3905970827698787e-05,
+      "loss": 0.8276,
+      "step": 2030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390022507339802e-05,
+      "loss": 0.8794,
+      "step": 2031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894477800079049e-05,
+      "loss": 0.8311,
+      "step": 2032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388872900998026e-05,
+      "loss": 0.8994,
+      "step": 2033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388297870534063e-05,
+      "loss": 0.8457,
+      "step": 2034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3877226888399727e-05,
+      "loss": 0.8008,
+      "step": 2035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3871473561397703e-05,
+      "loss": 0.8311,
+      "step": 2036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3865718726575315e-05,
+      "loss": 0.8306,
+      "step": 2037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3859962386173885e-05,
+      "loss": 0.8618,
+      "step": 2038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385420454243533e-05,
+      "loss": 0.8794,
+      "step": 2039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384844519760216e-05,
+      "loss": 0.8706,
+      "step": 2040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3842684353917462e-05,
+      "loss": 0.8774,
+      "step": 2041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836922013624903e-05,
+      "loss": 0.7715,
+      "step": 2042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3831158178968745e-05,
+      "loss": 0.8662,
+      "step": 2043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382539285219382e-05,
+      "loss": 0.5261,
+      "step": 2044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3819626035545548e-05,
+      "loss": 0.853,
+      "step": 2045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3813857731269931e-05,
+      "loss": 0.8384,
+      "step": 2046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3808087941613537e-05,
+      "loss": 0.8633,
+      "step": 2047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3802316668823534e-05,
+      "loss": 0.7803,
+      "step": 2048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3796543915147649e-05,
+      "loss": 0.8354,
+      "step": 2049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3790769682834195e-05,
+      "loss": 0.8384,
+      "step": 2050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3784993974132057e-05,
+      "loss": 0.792,
+      "step": 2051
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3779216791290699e-05,
+      "loss": 0.834,
+      "step": 2052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3773438136560153e-05,
+      "loss": 0.7671,
+      "step": 2053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3767658012191034e-05,
+      "loss": 0.8521,
+      "step": 2054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3761876420434516e-05,
+      "loss": 0.8608,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375609336354236e-05,
+      "loss": 0.8906,
+      "step": 2056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3750308843766883e-05,
+      "loss": 0.8521,
+      "step": 2057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374452286336098e-05,
+      "loss": 0.4968,
+      "step": 2058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3738735424578116e-05,
+      "loss": 0.8027,
+      "step": 2059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732946529672315e-05,
+      "loss": 0.8384,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3727156180898183e-05,
+      "loss": 0.8945,
+      "step": 2061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3721364380510875e-05,
+      "loss": 0.8521,
+      "step": 2062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3715571130766124e-05,
+      "loss": 0.7754,
+      "step": 2063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3709776433920225e-05,
+      "loss": 0.8413,
+      "step": 2064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3703980292230033e-05,
+      "loss": 0.8589,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3698182707952963e-05,
+      "loss": 0.8135,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3692383683347006e-05,
+      "loss": 0.8862,
+      "step": 2067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3686583220670694e-05,
+      "loss": 0.8848,
+      "step": 2068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3680781322183138e-05,
+      "loss": 0.8882,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3674977990143998e-05,
+      "loss": 0.8311,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3669173226813491e-05,
+      "loss": 0.9131,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3663367034452394e-05,
+      "loss": 0.8315,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3657559415322049e-05,
+      "loss": 0.8232,
+      "step": 2073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651750371684336e-05,
+      "loss": 0.8418,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3645939905801703e-05,
+      "loss": 0.7544,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3640128019937153e-05,
+      "loss": 0.8413,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3634314716354231e-05,
+      "loss": 0.8452,
+      "step": 2077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3628499997317044e-05,
+      "loss": 0.7842,
+      "step": 2078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3622683865090247e-05,
+      "loss": 0.8726,
+      "step": 2079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3616866321939045e-05,
+      "loss": 0.853,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361104737012919e-05,
+      "loss": 0.8496,
+      "step": 2081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3605227011926992e-05,
+      "loss": 0.9292,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3599405249599296e-05,
+      "loss": 0.8882,
+      "step": 2083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593582085413502e-05,
+      "loss": 0.8115,
+      "step": 2084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3587757521637554e-05,
+      "loss": 0.8154,
+      "step": 2085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3581931560539942e-05,
+      "loss": 0.8198,
+      "step": 2086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35761042043897e-05,
+      "loss": 0.8208,
+      "step": 2087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3570275455456405e-05,
+      "loss": 0.9385,
+      "step": 2088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3564445316010176e-05,
+      "loss": 0.8257,
+      "step": 2089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3558613788321668e-05,
+      "loss": 0.8394,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3552780874662093e-05,
+      "loss": 0.8008,
+      "step": 2091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546946577303182e-05,
+      "loss": 0.8193,
+      "step": 2092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3541110898517218e-05,
+      "loss": 0.8379,
+      "step": 2093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3535273840577028e-05,
+      "loss": 0.7993,
+      "step": 2094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3529435405755953e-05,
+      "loss": 0.9287,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3523595596327895e-05,
+      "loss": 0.8306,
+      "step": 2096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517754414567276e-05,
+      "loss": 0.8291,
+      "step": 2097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.351191186274906e-05,
+      "loss": 0.4541,
+      "step": 2098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350606794314874e-05,
+      "loss": 0.8628,
+      "step": 2099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3500222658042343e-05,
+      "loss": 0.8022,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3494376009706432e-05,
+      "loss": 0.8179,
+      "step": 2101
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3488528000418095e-05,
+      "loss": 0.8555,
+      "step": 2102
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3482678632454953e-05,
+      "loss": 0.8218,
+      "step": 2103
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3476827908095156e-05,
+      "loss": 0.8281,
+      "step": 2104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3470975829617381e-05,
+      "loss": 0.8179,
+      "step": 2105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3465122399300834e-05,
+      "loss": 0.8638,
+      "step": 2106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3459267619425247e-05,
+      "loss": 0.814,
+      "step": 2107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3453411492270876e-05,
+      "loss": 0.8628,
+      "step": 2108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3447554020118505e-05,
+      "loss": 0.8433,
+      "step": 2109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344169520524944e-05,
+      "loss": 0.8237,
+      "step": 2110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3435835049945504e-05,
+      "loss": 0.7954,
+      "step": 2111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342997355648906e-05,
+      "loss": 0.9102,
+      "step": 2112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342411072716297e-05,
+      "loss": 0.7471,
+      "step": 2113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3418246564250629e-05,
+      "loss": 0.9048,
+      "step": 2114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3412381070035954e-05,
+      "loss": 0.8325,
+      "step": 2115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3406514246803366e-05,
+      "loss": 0.875,
+      "step": 2116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3400646096837822e-05,
+      "loss": 0.8462,
+      "step": 2117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339477662242478e-05,
+      "loss": 0.895,
+      "step": 2118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388905825850222e-05,
+      "loss": 0.7944,
+      "step": 2119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3383033709400648e-05,
+      "loss": 0.8564,
+      "step": 2120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.337716027536306e-05,
+      "loss": 0.9111,
+      "step": 2121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3371285526024988e-05,
+      "loss": 0.8374,
+      "step": 2122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3365409463674464e-05,
+      "loss": 0.8369,
+      "step": 2123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3359532090600034e-05,
+      "loss": 0.4534,
+      "step": 2124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3353653409090752e-05,
+      "loss": 0.8374,
+      "step": 2125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334777342143619e-05,
+      "loss": 0.8442,
+      "step": 2126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3341892129926418e-05,
+      "loss": 0.9048,
+      "step": 2127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3336009536852022e-05,
+      "loss": 0.8809,
+      "step": 2128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3330125644504091e-05,
+      "loss": 0.8647,
+      "step": 2129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332424045517422e-05,
+      "loss": 0.8159,
+      "step": 2130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3318353971154512e-05,
+      "loss": 0.8618,
+      "step": 2131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331246619473757e-05,
+      "loss": 0.8799,
+      "step": 2132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3306577128216505e-05,
+      "loss": 0.875,
+      "step": 2133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3300686773884925e-05,
+      "loss": 0.8086,
+      "step": 2134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3294795134036945e-05,
+      "loss": 0.8677,
+      "step": 2135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3288902210967175e-05,
+      "loss": 0.8447,
+      "step": 2136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3283008006970737e-05,
+      "loss": 0.8086,
+      "step": 2137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327711252434323e-05,
+      "loss": 0.8794,
+      "step": 2138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3271215765380774e-05,
+      "loss": 0.7988,
+      "step": 2139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326531773237997e-05,
+      "loss": 0.8677,
+      "step": 2140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259418427637924e-05,
+      "loss": 0.8013,
+      "step": 2141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3253517853452236e-05,
+      "loss": 0.8311,
+      "step": 2142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3247616012120994e-05,
+      "loss": 0.8438,
+      "step": 2143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324171290594279e-05,
+      "loss": 0.8569,
+      "step": 2144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3235808537216701e-05,
+      "loss": 0.8252,
+      "step": 2145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3229902908242296e-05,
+      "loss": 0.8628,
+      "step": 2146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3223996021319636e-05,
+      "loss": 0.8691,
+      "step": 2147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3218087878749275e-05,
+      "loss": 0.8159,
+      "step": 2148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3212178482832255e-05,
+      "loss": 0.8306,
+      "step": 2149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32062678358701e-05,
+      "loss": 0.833,
+      "step": 2150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3200355940164831e-05,
+      "loss": 0.8613,
+      "step": 2151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3194442798018949e-05,
+      "loss": 0.8662,
+      "step": 2152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3188528411735437e-05,
+      "loss": 0.5024,
+      "step": 2153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3182612783617772e-05,
+      "loss": 0.7681,
+      "step": 2154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317669591596991e-05,
+      "loss": 0.8525,
+      "step": 2155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170777811096288e-05,
+      "loss": 0.8047,
+      "step": 2156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3164858471301827e-05,
+      "loss": 0.8872,
+      "step": 2157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.315893789889193e-05,
+      "loss": 0.8242,
+      "step": 2158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3153016096172475e-05,
+      "loss": 0.7852,
+      "step": 2159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3147093065449825e-05,
+      "loss": 0.8296,
+      "step": 2160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3141168809030824e-05,
+      "loss": 0.8525,
+      "step": 2161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135243329222777e-05,
+      "loss": 0.918,
+      "step": 2162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3129316628333489e-05,
+      "loss": 0.8662,
+      "step": 2163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312338870867122e-05,
+      "loss": 0.8882,
+      "step": 2164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3117459572544718e-05,
+      "loss": 0.8154,
+      "step": 2165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3111529222263198e-05,
+      "loss": 0.814,
+      "step": 2166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105597660136346e-05,
+      "loss": 0.8506,
+      "step": 2167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309966488847433e-05,
+      "loss": 0.8911,
+      "step": 2168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3093730909587782e-05,
+      "loss": 0.4927,
+      "step": 2169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087795725787802e-05,
+      "loss": 0.8511,
+      "step": 2170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3081859339385965e-05,
+      "loss": 0.8574,
+      "step": 2171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307592175269431e-05,
+      "loss": 0.8345,
+      "step": 2172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3069982968025343e-05,
+      "loss": 0.8828,
+      "step": 2173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3064042987692042e-05,
+      "loss": 0.873,
+      "step": 2174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3058101814007848e-05,
+      "loss": 0.8408,
+      "step": 2175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3052159449286668e-05,
+      "loss": 0.4861,
+      "step": 2176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3046215895842865e-05,
+      "loss": 0.8359,
+      "step": 2177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040271155991272e-05,
+      "loss": 0.7886,
+      "step": 2178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034325232047186e-05,
+      "loss": 0.8325,
+      "step": 2179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3028378126326364e-05,
+      "loss": 0.8076,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3022429841145015e-05,
+      "loss": 0.7969,
+      "step": 2181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016480378819822e-05,
+      "loss": 0.8281,
+      "step": 2182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3010529741667908e-05,
+      "loss": 0.8677,
+      "step": 2183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3004577932006874e-05,
+      "loss": 0.5173,
+      "step": 2184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2998624952154766e-05,
+      "loss": 0.8281,
+      "step": 2185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2992670804430078e-05,
+      "loss": 0.8159,
+      "step": 2186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298671549115178e-05,
+      "loss": 0.9019,
+      "step": 2187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298075901463928e-05,
+      "loss": 0.9116,
+      "step": 2188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2974801377212441e-05,
+      "loss": 0.8062,
+      "step": 2189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296884258119158e-05,
+      "loss": 0.8628,
+      "step": 2190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2962882628897469e-05,
+      "loss": 0.8418,
+      "step": 2191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2956921522651323e-05,
+      "loss": 0.8354,
+      "step": 2192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2950959264774815e-05,
+      "loss": 0.8262,
+      "step": 2193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2944995857590054e-05,
+      "loss": 0.8491,
+      "step": 2194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2939031303419617e-05,
+      "loss": 0.769,
+      "step": 2195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2933065604586501e-05,
+      "loss": 0.8574,
+      "step": 2196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2927098763414175e-05,
+      "loss": 0.854,
+      "step": 2197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2921130782226532e-05,
+      "loss": 0.8311,
+      "step": 2198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291516166334792e-05,
+      "loss": 0.8442,
+      "step": 2199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2909191409103134e-05,
+      "loss": 0.8193,
+      "step": 2200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29032200218174e-05,
+      "loss": 0.8892,
+      "step": 2201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2897247503816386e-05,
+      "loss": 0.8721,
+      "step": 2202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2891273857426217e-05,
+      "loss": 0.9028,
+      "step": 2203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2885299084973433e-05,
+      "loss": 0.5488,
+      "step": 2204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.287932318878503e-05,
+      "loss": 0.8423,
+      "step": 2205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2873346171188442e-05,
+      "loss": 0.7979,
+      "step": 2206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2867368034511521e-05,
+      "loss": 0.8374,
+      "step": 2207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286138878108258e-05,
+      "loss": 0.8262,
+      "step": 2208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2855408413230352e-05,
+      "loss": 0.8872,
+      "step": 2209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2849426933284004e-05,
+      "loss": 0.7969,
+      "step": 2210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2843444343573142e-05,
+      "loss": 0.8535,
+      "step": 2211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2837460646427803e-05,
+      "loss": 0.8203,
+      "step": 2212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2831475844178449e-05,
+      "loss": 0.9185,
+      "step": 2213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825489939155979e-05,
+      "loss": 0.4834,
+      "step": 2214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2819502933691723e-05,
+      "loss": 0.8315,
+      "step": 2215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2813514830117434e-05,
+      "loss": 0.8999,
+      "step": 2216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2807525630765292e-05,
+      "loss": 0.854,
+      "step": 2217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280153533796791e-05,
+      "loss": 0.4678,
+      "step": 2218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2795543954058322e-05,
+      "loss": 0.8359,
+      "step": 2219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2789551481369992e-05,
+      "loss": 0.8354,
+      "step": 2220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27835579222368e-05,
+      "loss": 0.8359,
+      "step": 2221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2777563278993057e-05,
+      "loss": 0.8525,
+      "step": 2222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277156755397349e-05,
+      "loss": 0.8906,
+      "step": 2223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2765570749513254e-05,
+      "loss": 0.8813,
+      "step": 2224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2759572867947918e-05,
+      "loss": 0.7983,
+      "step": 2225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275357391161347e-05,
+      "loss": 0.8716,
+      "step": 2226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274757388284633e-05,
+      "loss": 0.8467,
+      "step": 2227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2741572783983313e-05,
+      "loss": 0.8477,
+      "step": 2228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273557061736167e-05,
+      "loss": 0.8491,
+      "step": 2229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2729567385319064e-05,
+      "loss": 0.8989,
+      "step": 2230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272356309019356e-05,
+      "loss": 0.8018,
+      "step": 2231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2717557734323657e-05,
+      "loss": 0.8291,
+      "step": 2232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711551320048256e-05,
+      "loss": 0.8662,
+      "step": 2233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2705543849706663e-05,
+      "loss": 0.8706,
+      "step": 2234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2699535325638611e-05,
+      "loss": 0.7793,
+      "step": 2235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2693525750184237e-05,
+      "loss": 0.8608,
+      "step": 2236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2687515125684083e-05,
+      "loss": 0.854,
+      "step": 2237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2681503454479103e-05,
+      "loss": 0.8779,
+      "step": 2238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267549073891066e-05,
+      "loss": 0.854,
+      "step": 2239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2669476981320523e-05,
+      "loss": 0.4602,
+      "step": 2240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2663462184050865e-05,
+      "loss": 0.8711,
+      "step": 2241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2657446349444267e-05,
+      "loss": 0.7837,
+      "step": 2242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2651429479843714e-05,
+      "loss": 0.8711,
+      "step": 2243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2645411577592586e-05,
+      "loss": 0.8467,
+      "step": 2244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2639392645034677e-05,
+      "loss": 0.873,
+      "step": 2245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2633372684514176e-05,
+      "loss": 0.8301,
+      "step": 2246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2627351698375671e-05,
+      "loss": 0.8618,
+      "step": 2247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2621329688964156e-05,
+      "loss": 0.874,
+      "step": 2248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2615306658625018e-05,
+      "loss": 0.8311,
+      "step": 2249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2609282609704035e-05,
+      "loss": 0.8696,
+      "step": 2250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2603257544547403e-05,
+      "loss": 0.8589,
+      "step": 2251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259723146550169e-05,
+      "loss": 0.854,
+      "step": 2252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259120437491387e-05,
+      "loss": 0.8813,
+      "step": 2253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2585176275131314e-05,
+      "loss": 0.8804,
+      "step": 2254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2579147168501784e-05,
+      "loss": 0.8394,
+      "step": 2255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2573117057373426e-05,
+      "loss": 0.8892,
+      "step": 2256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2567085944094787e-05,
+      "loss": 0.8101,
+      "step": 2257
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2561053831014803e-05,
+      "loss": 0.8267,
+      "step": 2258
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2555020720482793e-05,
+      "loss": 0.4839,
+      "step": 2259
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2548986614848473e-05,
+      "loss": 0.813,
+      "step": 2260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2542951516461939e-05,
+      "loss": 0.8501,
+      "step": 2261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2536915427673679e-05,
+      "loss": 0.854,
+      "step": 2262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2530878350834564e-05,
+      "loss": 0.8735,
+      "step": 2263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2524840288295848e-05,
+      "loss": 0.8438,
+      "step": 2264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518801242409175e-05,
+      "loss": 0.8008,
+      "step": 2265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2512761215526566e-05,
+      "loss": 0.8208,
+      "step": 2266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250672021000043e-05,
+      "loss": 0.8594,
+      "step": 2267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250067822818355e-05,
+      "loss": 0.8535,
+      "step": 2268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2494635272429092e-05,
+      "loss": 0.877,
+      "step": 2269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2488591345090608e-05,
+      "loss": 0.8477,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2482546448522015e-05,
+      "loss": 0.8862,
+      "step": 2271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247650058507762e-05,
+      "loss": 0.8228,
+      "step": 2272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2470453757112104e-05,
+      "loss": 0.8076,
+      "step": 2273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2464405966980513e-05,
+      "loss": 0.8901,
+      "step": 2274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2458357217038284e-05,
+      "loss": 0.8081,
+      "step": 2275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2452307509641217e-05,
+      "loss": 0.8345,
+      "step": 2276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2446256847145485e-05,
+      "loss": 0.8364,
+      "step": 2277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244020523190764e-05,
+      "loss": 0.8418,
+      "step": 2278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2434152666284597e-05,
+      "loss": 0.8174,
+      "step": 2279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2428099152633648e-05,
+      "loss": 0.813,
+      "step": 2280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2422044693312452e-05,
+      "loss": 0.8813,
+      "step": 2281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2415989290679028e-05,
+      "loss": 0.8828,
+      "step": 2282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240993294709178e-05,
+      "loss": 0.8115,
+      "step": 2283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2403875664909464e-05,
+      "loss": 0.7983,
+      "step": 2284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2397817446491202e-05,
+      "loss": 0.8481,
+      "step": 2285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2391758294196492e-05,
+      "loss": 0.8735,
+      "step": 2286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2385698210385181e-05,
+      "loss": 0.8774,
+      "step": 2287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2379637197417492e-05,
+      "loss": 0.8696,
+      "step": 2288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2373575257653998e-05,
+      "loss": 0.8105,
+      "step": 2289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2367512393455641e-05,
+      "loss": 0.832,
+      "step": 2290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2361448607183726e-05,
+      "loss": 0.8442,
+      "step": 2291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2355383901199907e-05,
+      "loss": 0.7915,
+      "step": 2292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23493182778662e-05,
+      "loss": 0.7974,
+      "step": 2293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2343251739544987e-05,
+      "loss": 0.7754,
+      "step": 2294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2337184288598986e-05,
+      "loss": 0.8374,
+      "step": 2295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2331115927391296e-05,
+      "loss": 0.8555,
+      "step": 2296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232504665828535e-05,
+      "loss": 0.8394,
+      "step": 2297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2318976483644947e-05,
+      "loss": 0.8115,
+      "step": 2298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231290540583423e-05,
+      "loss": 0.7529,
+      "step": 2299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2306833427217705e-05,
+      "loss": 0.8247,
+      "step": 2300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2300760550160211e-05,
+      "loss": 0.519,
+      "step": 2301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2294686777026956e-05,
+      "loss": 0.8345,
+      "step": 2302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2288612110183486e-05,
+      "loss": 0.8315,
+      "step": 2303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2282536551995696e-05,
+      "loss": 0.8452,
+      "step": 2304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2276460104829835e-05,
+      "loss": 0.8486,
+      "step": 2305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2270382771052487e-05,
+      "loss": 0.7935,
+      "step": 2306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2264304553030591e-05,
+      "loss": 0.897,
+      "step": 2307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2258225453131423e-05,
+      "loss": 0.8672,
+      "step": 2308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2252145473722609e-05,
+      "loss": 0.8369,
+      "step": 2309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2246064617172118e-05,
+      "loss": 0.8213,
+      "step": 2310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2239982885848251e-05,
+      "loss": 0.8276,
+      "step": 2311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223390028211966e-05,
+      "loss": 0.8398,
+      "step": 2312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2227816808355333e-05,
+      "loss": 0.7842,
+      "step": 2313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2221732466924592e-05,
+      "loss": 0.8921,
+      "step": 2314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2215647260197109e-05,
+      "loss": 0.8276,
+      "step": 2315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220956119054288e-05,
+      "loss": 0.8589,
+      "step": 2316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2203474260332244e-05,
+      "loss": 0.8613,
+      "step": 2317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2197386471935877e-05,
+      "loss": 0.7769,
+      "step": 2318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2191297827724782e-05,
+      "loss": 0.9121,
+      "step": 2319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21852083300703e-05,
+      "loss": 0.8843,
+      "step": 2320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2179117981344107e-05,
+      "loss": 0.793,
+      "step": 2321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173026783918203e-05,
+      "loss": 0.8022,
+      "step": 2322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2166934740164925e-05,
+      "loss": 0.8516,
+      "step": 2323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2160841852456939e-05,
+      "loss": 0.8765,
+      "step": 2324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2154748123167236e-05,
+      "loss": 0.8447,
+      "step": 2325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2148653554669135e-05,
+      "loss": 0.7598,
+      "step": 2326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2142558149336287e-05,
+      "loss": 0.8174,
+      "step": 2327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2136461909542667e-05,
+      "loss": 0.8574,
+      "step": 2328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2130364837662572e-05,
+      "loss": 0.8442,
+      "step": 2329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2124266936070622e-05,
+      "loss": 0.895,
+      "step": 2330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2118168207141764e-05,
+      "loss": 0.7891,
+      "step": 2331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211206865325127e-05,
+      "loss": 0.7876,
+      "step": 2332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2105968276774727e-05,
+      "loss": 0.8413,
+      "step": 2333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2099867080088043e-05,
+      "loss": 0.8545,
+      "step": 2334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2093765065567448e-05,
+      "loss": 0.7983,
+      "step": 2335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2087662235589492e-05,
+      "loss": 0.8555,
+      "step": 2336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2081558592531037e-05,
+      "loss": 0.4939,
+      "step": 2337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2075454138769266e-05,
+      "loss": 0.8413,
+      "step": 2338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2069348876681683e-05,
+      "loss": 0.8423,
+      "step": 2339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2063242808646092e-05,
+      "loss": 0.8555,
+      "step": 2340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205713593704062e-05,
+      "loss": 0.896,
+      "step": 2341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051028264243714e-05,
+      "loss": 0.8447,
+      "step": 2342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2044919792634115e-05,
+      "loss": 0.856,
+      "step": 2343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2038810524590896e-05,
+      "loss": 0.833,
+      "step": 2344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2032700462493426e-05,
+      "loss": 0.8003,
+      "step": 2345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2026589608721379e-05,
+      "loss": 0.9111,
+      "step": 2346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2020477965654758e-05,
+      "loss": 0.7905,
+      "step": 2347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2014365535673855e-05,
+      "loss": 0.5386,
+      "step": 2348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2008252321159273e-05,
+      "loss": 0.8164,
+      "step": 2349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2002138324491922e-05,
+      "loss": 0.8403,
+      "step": 2350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199602354805302e-05,
+      "loss": 0.8384,
+      "step": 2351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1989907994224081e-05,
+      "loss": 0.8809,
+      "step": 2352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198379166538693e-05,
+      "loss": 0.8706,
+      "step": 2353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977674563923687e-05,
+      "loss": 0.8374,
+      "step": 2354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1971556692216775e-05,
+      "loss": 0.8799,
+      "step": 2355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1965438052648918e-05,
+      "loss": 0.8862,
+      "step": 2356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.195931864760314e-05,
+      "loss": 0.8047,
+      "step": 2357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1953198479462761e-05,
+      "loss": 0.8525,
+      "step": 2358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.19470775506114e-05,
+      "loss": 0.8276,
+      "step": 2359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1940955863432972e-05,
+      "loss": 0.8638,
+      "step": 2360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1934833420311684e-05,
+      "loss": 0.8823,
+      "step": 2361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1928710223632038e-05,
+      "loss": 0.8306,
+      "step": 2362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1922586275778842e-05,
+      "loss": 0.8081,
+      "step": 2363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1916461579137175e-05,
+      "loss": 0.5137,
+      "step": 2364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191033613609242e-05,
+      "loss": 0.4753,
+      "step": 2365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1904209949030256e-05,
+      "loss": 0.8691,
+      "step": 2366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189808302033664e-05,
+      "loss": 0.8442,
+      "step": 2367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1891955352397823e-05,
+      "loss": 0.8018,
+      "step": 2368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1885826947600353e-05,
+      "loss": 0.856,
+      "step": 2369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1879697808331039e-05,
+      "loss": 0.8179,
+      "step": 2370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1873567936977007e-05,
+      "loss": 0.8135,
+      "step": 2371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1867437335925649e-05,
+      "loss": 0.8608,
+      "step": 2372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1861306007564645e-05,
+      "loss": 0.8721,
+      "step": 2373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1855173954281964e-05,
+      "loss": 0.8267,
+      "step": 2374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1849041178465849e-05,
+      "loss": 0.895,
+      "step": 2375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1842907682504831e-05,
+      "loss": 0.7646,
+      "step": 2376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1836773468787716e-05,
+      "loss": 0.8257,
+      "step": 2377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1830638539703596e-05,
+      "loss": 0.8662,
+      "step": 2378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1824502897641835e-05,
+      "loss": 0.8208,
+      "step": 2379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181836654499208e-05,
+      "loss": 0.9155,
+      "step": 2380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1812229484144247e-05,
+      "loss": 0.4397,
+      "step": 2381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1806091717488542e-05,
+      "loss": 0.8345,
+      "step": 2382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1799953247415427e-05,
+      "loss": 0.8008,
+      "step": 2383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1793814076315655e-05,
+      "loss": 0.7935,
+      "step": 2384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1787674206580248e-05,
+      "loss": 0.8408,
+      "step": 2385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1781533640600487e-05,
+      "loss": 0.5432,
+      "step": 2386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1775392380767943e-05,
+      "loss": 0.876,
+      "step": 2387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1769250429474448e-05,
+      "loss": 0.8545,
+      "step": 2388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1763107789112096e-05,
+      "loss": 0.8228,
+      "step": 2389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1756964462073265e-05,
+      "loss": 0.835,
+      "step": 2390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1750820450750593e-05,
+      "loss": 0.7432,
+      "step": 2391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1744675757536979e-05,
+      "loss": 0.8569,
+      "step": 2392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1738530384825598e-05,
+      "loss": 0.8345,
+      "step": 2393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1732384335009882e-05,
+      "loss": 0.8477,
+      "step": 2394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1726237610483526e-05,
+      "loss": 0.8379,
+      "step": 2395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1720090213640496e-05,
+      "loss": 0.8154,
+      "step": 2396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1713942146875013e-05,
+      "loss": 0.4849,
+      "step": 2397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.170779341258156e-05,
+      "loss": 0.7642,
+      "step": 2398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1701644013154883e-05,
+      "loss": 0.7847,
+      "step": 2399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169549395098998e-05,
+      "loss": 0.8677,
+      "step": 2400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1689343228482117e-05,
+      "loss": 0.8535,
+      "step": 2401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168319184802681e-05,
+      "loss": 0.4871,
+      "step": 2402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1677039812019833e-05,
+      "loss": 0.8423,
+      "step": 2403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1670887122857216e-05,
+      "loss": 0.8057,
+      "step": 2404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166473378293524e-05,
+      "loss": 0.8242,
+      "step": 2405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1658579794650449e-05,
+      "loss": 0.8149,
+      "step": 2406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1652425160399628e-05,
+      "loss": 0.8101,
+      "step": 2407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1646269882579816e-05,
+      "loss": 0.8413,
+      "step": 2408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1640113963588312e-05,
+      "loss": 0.8359,
+      "step": 2409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1633957405822653e-05,
+      "loss": 0.4746,
+      "step": 2410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1627800211680634e-05,
+      "loss": 0.813,
+      "step": 2411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1621642383560292e-05,
+      "loss": 0.8208,
+      "step": 2412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1615483923859907e-05,
+      "loss": 0.8325,
+      "step": 2413
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609324834978018e-05,
+      "loss": 0.8276,
+      "step": 2414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1603165119313395e-05,
+      "loss": 0.8799,
+      "step": 2415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1597004779265063e-05,
+      "loss": 0.7993,
+      "step": 2416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1590843817232287e-05,
+      "loss": 0.8364,
+      "step": 2417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1584682235614568e-05,
+      "loss": 0.8677,
+      "step": 2418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1578520036811658e-05,
+      "loss": 0.8296,
+      "step": 2419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1572357223223541e-05,
+      "loss": 0.8252,
+      "step": 2420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1566193797250446e-05,
+      "loss": 0.8604,
+      "step": 2421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1560029761292839e-05,
+      "loss": 0.8491,
+      "step": 2422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1553865117751422e-05,
+      "loss": 0.8428,
+      "step": 2423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1547699869027135e-05,
+      "loss": 0.8091,
+      "step": 2424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1541534017521154e-05,
+      "loss": 0.8608,
+      "step": 2425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.153536756563489e-05,
+      "loss": 0.8389,
+      "step": 2426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1529200515769988e-05,
+      "loss": 0.7876,
+      "step": 2427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1523032870328319e-05,
+      "loss": 0.8481,
+      "step": 2428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1516864631711996e-05,
+      "loss": 0.7375,
+      "step": 2429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510695802323366e-05,
+      "loss": 0.8862,
+      "step": 2430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1504526384564988e-05,
+      "loss": 0.4604,
+      "step": 2431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1498356380839665e-05,
+      "loss": 0.8779,
+      "step": 2432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149218579355043e-05,
+      "loss": 0.8408,
+      "step": 2433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1486014625100525e-05,
+      "loss": 0.8018,
+      "step": 2434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1479842877893445e-05,
+      "loss": 0.9043,
+      "step": 2435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1473670554332892e-05,
+      "loss": 0.8228,
+      "step": 2436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146749765682279e-05,
+      "loss": 0.8179,
+      "step": 2437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1461324187767301e-05,
+      "loss": 0.7944,
+      "step": 2438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1455150149570804e-05,
+      "loss": 0.8931,
+      "step": 2439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.144897554463789e-05,
+      "loss": 0.7783,
+      "step": 2440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1442800375373384e-05,
+      "loss": 0.8481,
+      "step": 2441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1436624644182323e-05,
+      "loss": 0.9014,
+      "step": 2442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1430448353469965e-05,
+      "loss": 0.8154,
+      "step": 2443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1424271505641789e-05,
+      "loss": 0.8413,
+      "step": 2444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1418094103103485e-05,
+      "loss": 0.8325,
+      "step": 2445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1411916148260961e-05,
+      "loss": 0.7881,
+      "step": 2446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1405737643520347e-05,
+      "loss": 0.8179,
+      "step": 2447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399558591287976e-05,
+      "loss": 0.4846,
+      "step": 2448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13933789939704e-05,
+      "loss": 0.8125,
+      "step": 2449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1387198853974387e-05,
+      "loss": 0.8589,
+      "step": 2450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1381018173706905e-05,
+      "loss": 0.8662,
+      "step": 2451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1374836955575147e-05,
+      "loss": 0.8477,
+      "step": 2452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1368655201986503e-05,
+      "loss": 0.4768,
+      "step": 2453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136247291534858e-05,
+      "loss": 0.8218,
+      "step": 2454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135629009806919e-05,
+      "loss": 0.8228,
+      "step": 2455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1350106752556344e-05,
+      "loss": 0.8296,
+      "step": 2456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1343922881218277e-05,
+      "loss": 0.5088,
+      "step": 2457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1337738486463406e-05,
+      "loss": 0.4468,
+      "step": 2458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133155357070037e-05,
+      "loss": 0.7715,
+      "step": 2459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1325368136338004e-05,
+      "loss": 0.4734,
+      "step": 2460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1319182185785343e-05,
+      "loss": 0.8438,
+      "step": 2461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1312995721451625e-05,
+      "loss": 0.8599,
+      "step": 2462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1306808745746294e-05,
+      "loss": 0.8647,
+      "step": 2463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1300621261078977e-05,
+      "loss": 0.4797,
+      "step": 2464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.129443326985952e-05,
+      "loss": 0.8662,
+      "step": 2465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128824477449795e-05,
+      "loss": 0.8579,
+      "step": 2466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12820557774045e-05,
+      "loss": 0.8574,
+      "step": 2467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127586628098959e-05,
+      "loss": 0.4617,
+      "step": 2468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269676287663847e-05,
+      "loss": 0.8276,
+      "step": 2469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1263485799838076e-05,
+      "loss": 0.8403,
+      "step": 2470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1257294819923285e-05,
+      "loss": 0.8286,
+      "step": 2471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1251103350330675e-05,
+      "loss": 0.8154,
+      "step": 2472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1244911393471629e-05,
+      "loss": 0.7832,
+      "step": 2473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1238718951757726e-05,
+      "loss": 0.7642,
+      "step": 2474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123252602760073e-05,
+      "loss": 0.8384,
+      "step": 2475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12263326234126e-05,
+      "loss": 0.8726,
+      "step": 2476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1220138741605474e-05,
+      "loss": 0.9058,
+      "step": 2477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1213944384591683e-05,
+      "loss": 0.5081,
+      "step": 2478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1207749554783733e-05,
+      "loss": 0.8726,
+      "step": 2479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1201554254594323e-05,
+      "loss": 0.8438,
+      "step": 2480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1195358486436335e-05,
+      "loss": 0.8574,
+      "step": 2481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118916225272283e-05,
+      "loss": 0.8237,
+      "step": 2482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1182965555867051e-05,
+      "loss": 0.8237,
+      "step": 2483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1176768398282422e-05,
+      "loss": 0.8408,
+      "step": 2484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1170570782382543e-05,
+      "loss": 0.7993,
+      "step": 2485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1164372710581199e-05,
+      "loss": 0.853,
+      "step": 2486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1158174185292352e-05,
+      "loss": 0.7988,
+      "step": 2487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1151975208930129e-05,
+      "loss": 0.7959,
+      "step": 2488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1145775783908848e-05,
+      "loss": 0.814,
+      "step": 2489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1139575912642992e-05,
+      "loss": 0.5049,
+      "step": 2490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133375597547225e-05,
+      "loss": 0.8027,
+      "step": 2491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127174841036374e-05,
+      "loss": 0.8169,
+      "step": 2492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1120973645525451e-05,
+      "loss": 0.8257,
+      "step": 2493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1114772013429625e-05,
+      "loss": 0.8652,
+      "step": 2494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108569947164245e-05,
+      "loss": 0.8291,
+      "step": 2495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102367449144821e-05,
+      "loss": 0.4592,
+      "step": 2496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1096164521787048e-05,
+      "loss": 0.8281,
+      "step": 2497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1089961167506765e-05,
+      "loss": 0.7847,
+      "step": 2498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1083757388719989e-05,
+      "loss": 0.7944,
+      "step": 2499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107755318784291e-05,
+      "loss": 0.8491,
+      "step": 2500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1071348567291866e-05,
+      "loss": 0.8359,
+      "step": 2501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1065143529483371e-05,
+      "loss": 0.8271,
+      "step": 2502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1058938076834095e-05,
+      "loss": 0.7886,
+      "step": 2503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1052732211760872e-05,
+      "loss": 0.8423,
+      "step": 2504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1046525936680699e-05,
+      "loss": 0.8564,
+      "step": 2505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1040319254010726e-05,
+      "loss": 0.7993,
+      "step": 2506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1034112166168267e-05,
+      "loss": 0.8384,
+      "step": 2507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1027904675570796e-05,
+      "loss": 0.8599,
+      "step": 2508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1021696784635937e-05,
+      "loss": 0.8345,
+      "step": 2509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1015488495781474e-05,
+      "loss": 0.8257,
+      "step": 2510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1009279811425346e-05,
+      "loss": 0.7842,
+      "step": 2511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1003070733985644e-05,
+      "loss": 0.853,
+      "step": 2512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0996861265880616e-05,
+      "loss": 0.7622,
+      "step": 2513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0990651409528657e-05,
+      "loss": 0.8765,
+      "step": 2514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0984441167348316e-05,
+      "loss": 0.8267,
+      "step": 2515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0978230541758295e-05,
+      "loss": 0.8706,
+      "step": 2516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0972019535177437e-05,
+      "loss": 0.8335,
+      "step": 2517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0965808150024745e-05,
+      "loss": 0.8076,
+      "step": 2518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0959596388719362e-05,
+      "loss": 0.875,
+      "step": 2519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953384253680574e-05,
+      "loss": 0.7827,
+      "step": 2520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0947171747327825e-05,
+      "loss": 0.7852,
+      "step": 2521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094095887208069e-05,
+      "loss": 0.8398,
+      "step": 2522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934745630358889e-05,
+      "loss": 0.8257,
+      "step": 2523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0928532024582304e-05,
+      "loss": 0.8379,
+      "step": 2524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092231805717093e-05,
+      "loss": 0.8696,
+      "step": 2525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0916103730544924e-05,
+      "loss": 0.8345,
+      "step": 2526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090988904712458e-05,
+      "loss": 0.8003,
+      "step": 2527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0903674009330314e-05,
+      "loss": 0.8042,
+      "step": 2528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0897458619582704e-05,
+      "loss": 0.8213,
+      "step": 2529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0891242880302448e-05,
+      "loss": 0.8589,
+      "step": 2530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0885026793910391e-05,
+      "loss": 0.8521,
+      "step": 2531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0878810362827504e-05,
+      "loss": 0.8452,
+      "step": 2532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0872593589474898e-05,
+      "loss": 0.8853,
+      "step": 2533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0866376476273817e-05,
+      "loss": 0.8228,
+      "step": 2534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0860159025645633e-05,
+      "loss": 0.7803,
+      "step": 2535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0853941240011856e-05,
+      "loss": 0.8535,
+      "step": 2536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0847723121794121e-05,
+      "loss": 0.8091,
+      "step": 2537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0841504673414196e-05,
+      "loss": 0.8208,
+      "step": 2538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0835285897293975e-05,
+      "loss": 0.8174,
+      "step": 2539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082906679585548e-05,
+      "loss": 0.8564,
+      "step": 2540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082284737152086e-05,
+      "loss": 0.8066,
+      "step": 2541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0816627626712398e-05,
+      "loss": 0.8291,
+      "step": 2542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0810407563852486e-05,
+      "loss": 0.8486,
+      "step": 2543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0804187185363647e-05,
+      "loss": 0.8057,
+      "step": 2544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0797966493668533e-05,
+      "loss": 0.7852,
+      "step": 2545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0791745491189912e-05,
+      "loss": 0.8042,
+      "step": 2546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0785524180350668e-05,
+      "loss": 0.4568,
+      "step": 2547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0779302563573821e-05,
+      "loss": 0.8359,
+      "step": 2548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0773080643282493e-05,
+      "loss": 0.7998,
+      "step": 2549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0766858421899934e-05,
+      "loss": 0.7979,
+      "step": 2550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760635901849508e-05,
+      "loss": 0.8853,
+      "step": 2551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0754413085554694e-05,
+      "loss": 0.8462,
+      "step": 2552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0748189975439091e-05,
+      "loss": 0.8364,
+      "step": 2553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074196657392641e-05,
+      "loss": 0.79,
+      "step": 2554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0735742883440474e-05,
+      "loss": 0.7637,
+      "step": 2555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.072951890640522e-05,
+      "loss": 0.8755,
+      "step": 2556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723294645244698e-05,
+      "loss": 0.7915,
+      "step": 2557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0717070102383065e-05,
+      "loss": 0.8403,
+      "step": 2558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071084528024459e-05,
+      "loss": 0.8418,
+      "step": 2559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070462018125365e-05,
+      "loss": 0.8296,
+      "step": 2560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0698394807834733e-05,
+      "loss": 0.8774,
+      "step": 2561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.069216916241243e-05,
+      "loss": 0.8193,
+      "step": 2562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0685943247411439e-05,
+      "loss": 0.8496,
+      "step": 2563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.067971706525656e-05,
+      "loss": 0.8403,
+      "step": 2564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673490618372705e-05,
+      "loss": 0.8213,
+      "step": 2565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0667263909184887e-05,
+      "loss": 0.8188,
+      "step": 2566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0661036940118208e-05,
+      "loss": 0.7939,
+      "step": 2567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0654809713597888e-05,
+      "loss": 0.7856,
+      "step": 2568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0648582232049247e-05,
+      "loss": 0.7671,
+      "step": 2569
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0642354497897687e-05,
+      "loss": 0.8687,
+      "step": 2570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0636126513568725e-05,
+      "loss": 0.8457,
+      "step": 2571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0629898281487974e-05,
+      "loss": 0.4971,
+      "step": 2572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062366980408113e-05,
+      "loss": 0.7842,
+      "step": 2573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0617441083774003e-05,
+      "loss": 0.875,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.061121212299249e-05,
+      "loss": 0.896,
+      "step": 2575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0604982924162568e-05,
+      "loss": 0.8384,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0598753489710329e-05,
+      "loss": 0.8232,
+      "step": 2577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0592523822061949e-05,
+      "loss": 0.8281,
+      "step": 2578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0586293923643687e-05,
+      "loss": 0.8306,
+      "step": 2579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05800637968819e-05,
+      "loss": 0.8286,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057383344420303e-05,
+      "loss": 0.8525,
+      "step": 2581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0567602868033609e-05,
+      "loss": 0.8794,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0561372070800257e-05,
+      "loss": 0.854,
+      "step": 2583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0555141054929679e-05,
+      "loss": 0.8467,
+      "step": 2584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0548909822848664e-05,
+      "loss": 0.8569,
+      "step": 2585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0542678376984083e-05,
+      "loss": 0.8452,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0536446719762898e-05,
+      "loss": 0.8496,
+      "step": 2587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0530214853612146e-05,
+      "loss": 0.7688,
+      "step": 2588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0523982780958946e-05,
+      "loss": 0.8726,
+      "step": 2589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0517750504230507e-05,
+      "loss": 0.8027,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0511518025854102e-05,
+      "loss": 0.8345,
+      "step": 2591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0505285348257093e-05,
+      "loss": 0.7812,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049905247386692e-05,
+      "loss": 0.8237,
+      "step": 2593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0492819405111092e-05,
+      "loss": 0.8555,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0486586144417199e-05,
+      "loss": 0.4539,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048035269421291e-05,
+      "loss": 0.8828,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0474119056925956e-05,
+      "loss": 0.8901,
+      "step": 2597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0467885234984154e-05,
+      "loss": 0.8042,
+      "step": 2598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0461651230815383e-05,
+      "loss": 0.8203,
+      "step": 2599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0455417046847598e-05,
+      "loss": 0.8389,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044918268550882e-05,
+      "loss": 0.8794,
+      "step": 2601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0442948149227148e-05,
+      "loss": 0.4956,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0436713440430737e-05,
+      "loss": 0.8398,
+      "step": 2603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0430478561547817e-05,
+      "loss": 0.772,
+      "step": 2604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0424243515006683e-05,
+      "loss": 0.8105,
+      "step": 2605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0418008303235694e-05,
+      "loss": 0.7832,
+      "step": 2606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0411772928663275e-05,
+      "loss": 0.8379,
+      "step": 2607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040553739371791e-05,
+      "loss": 0.8853,
+      "step": 2608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0399301700828155e-05,
+      "loss": 0.4927,
+      "step": 2609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0393065852422614e-05,
+      "loss": 0.8408,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0386829850929968e-05,
+      "loss": 0.7881,
+      "step": 2611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0380593698778938e-05,
+      "loss": 0.7485,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037435739839832e-05,
+      "loss": 0.8589,
+      "step": 2613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0368120952216965e-05,
+      "loss": 0.8213,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0361884362663774e-05,
+      "loss": 0.8501,
+      "step": 2615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355647632167705e-05,
+      "loss": 0.8604,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0349410763157778e-05,
+      "loss": 0.8799,
+      "step": 2617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0343173758063065e-05,
+      "loss": 0.8491,
+      "step": 2618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0336936619312679e-05,
+      "loss": 0.8149,
+      "step": 2619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0330699349335803e-05,
+      "loss": 0.4736,
+      "step": 2620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0324461950561657e-05,
+      "loss": 0.8218,
+      "step": 2621
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031822442541952e-05,
+      "loss": 0.7915,
+      "step": 2622
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0311986776338716e-05,
+      "loss": 0.8413,
+      "step": 2623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030574900574862e-05,
+      "loss": 0.814,
+      "step": 2624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029951111607865e-05,
+      "loss": 0.8838,
+      "step": 2625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0293273109758275e-05,
+      "loss": 0.8228,
+      "step": 2626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0287034989217007e-05,
+      "loss": 0.7671,
+      "step": 2627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0280796756884402e-05,
+      "loss": 0.8389,
+      "step": 2628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0274558415190058e-05,
+      "loss": 0.4331,
+      "step": 2629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0268319966563624e-05,
+      "loss": 0.8525,
+      "step": 2630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0262081413434782e-05,
+      "loss": 0.8403,
+      "step": 2631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0255842758233251e-05,
+      "loss": 0.8335,
+      "step": 2632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024960400338881e-05,
+      "loss": 0.8755,
+      "step": 2633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024336515133125e-05,
+      "loss": 0.8213,
+      "step": 2634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0237126204490418e-05,
+      "loss": 0.792,
+      "step": 2635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0230887165296194e-05,
+      "loss": 0.8174,
+      "step": 2636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224648036178488e-05,
+      "loss": 0.8423,
+      "step": 2637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0218408819567256e-05,
+      "loss": 0.8232,
+      "step": 2638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0212169517892478e-05,
+      "loss": 0.8306,
+      "step": 2639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.020593013358417e-05,
+      "loss": 0.7983,
+      "step": 2640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0199690669072385e-05,
+      "loss": 0.8403,
+      "step": 2641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.01934511267872e-05,
+      "loss": 0.8389,
+      "step": 2642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0187211509158723e-05,
+      "loss": 0.8433,
+      "step": 2643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.01809718186171e-05,
+      "loss": 0.8076,
+      "step": 2644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0174732057592496e-05,
+      "loss": 0.8892,
+      "step": 2645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0168492228515107e-05,
+      "loss": 0.8442,
+      "step": 2646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0162252333815157e-05,
+      "loss": 0.8423,
+      "step": 2647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0156012375922892e-05,
+      "loss": 0.8071,
+      "step": 2648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0149772357268586e-05,
+      "loss": 0.811,
+      "step": 2649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0143532280282533e-05,
+      "loss": 0.4497,
+      "step": 2650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0137292147395054e-05,
+      "loss": 0.4934,
+      "step": 2651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0131051961036488e-05,
+      "loss": 0.854,
+      "step": 2652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.01248117236372e-05,
+      "loss": 0.835,
+      "step": 2653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0118571437627567e-05,
+      "loss": 0.8633,
+      "step": 2654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.011233110543799e-05,
+      "loss": 0.832,
+      "step": 2655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0106090729498885e-05,
+      "loss": 0.7905,
+      "step": 2656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099850312240696e-05,
+      "loss": 0.8174,
+      "step": 2657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0093609856093866e-05,
+      "loss": 0.832,
+      "step": 2658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087369363488861e-05,
+      "loss": 0.8101,
+      "step": 2659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0081128836856174e-05,
+      "loss": 0.8477,
+      "step": 2660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.007488827862628e-05,
+      "loss": 0.8354,
+      "step": 2661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0068647691229701e-05,
+      "loss": 0.8315,
+      "step": 2662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006240707709695e-05,
+      "loss": 0.7886,
+      "step": 2663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0056166438658549e-05,
+      "loss": 0.8672,
+      "step": 2664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0049925778345042e-05,
+      "loss": 0.8613,
+      "step": 2665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043685098586977e-05,
+      "loss": 0.833,
+      "step": 2666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.00374444018149e-05,
+      "loss": 0.8403,
+      "step": 2667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0031203690459378e-05,
+      "loss": 0.8091,
+      "step": 2668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0024962966950973e-05,
+      "loss": 0.8135,
+      "step": 2669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0018722233720257e-05,
+      "loss": 0.8774,
+      "step": 2670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0012481493197806e-05,
+      "loss": 0.812,
+      "step": 2671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0006240747814193e-05,
+      "loss": 0.7812,
+      "step": 2672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.8384,
+      "step": 2673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99375925218581e-06,
+      "loss": 0.811,
+      "step": 2674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987518506802196e-06,
+      "loss": 0.7925,
+      "step": 2675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.981277766279744e-06,
+      "loss": 0.812,
+      "step": 2676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.975037033049032e-06,
+      "loss": 0.8125,
+      "step": 2677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.968796309540626e-06,
+      "loss": 0.791,
+      "step": 2678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.962555598185103e-06,
+      "loss": 0.8267,
+      "step": 2679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956314901413026e-06,
+      "loss": 0.8604,
+      "step": 2680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.950074221654958e-06,
+      "loss": 0.8081,
+      "step": 2681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943833561341453e-06,
+      "loss": 0.8911,
+      "step": 2682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937592922903056e-06,
+      "loss": 0.832,
+      "step": 2683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931352308770302e-06,
+      "loss": 0.7407,
+      "step": 2684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925111721373722e-06,
+      "loss": 0.793,
+      "step": 2685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.918871163143831e-06,
+      "loss": 0.834,
+      "step": 2686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912630636511137e-06,
+      "loss": 0.7866,
+      "step": 2687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.906390143906137e-06,
+      "loss": 0.7793,
+      "step": 2688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900149687759309e-06,
+      "loss": 0.7949,
+      "step": 2689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.893909270501117e-06,
+      "loss": 0.8169,
+      "step": 2690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887668894562013e-06,
+      "loss": 0.8174,
+      "step": 2691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.881428562372438e-06,
+      "loss": 0.8096,
+      "step": 2692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.875188276362804e-06,
+      "loss": 0.8174,
+      "step": 2693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86894803896351e-06,
+      "loss": 0.7954,
+      "step": 2694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86270785260495e-06,
+      "loss": 0.8525,
+      "step": 2695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856467719717469e-06,
+      "loss": 0.8252,
+      "step": 2696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850227642731417e-06,
+      "loss": 0.8545,
+      "step": 2697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84398762407711e-06,
+      "loss": 0.8506,
+      "step": 2698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837747666184843e-06,
+      "loss": 0.8027,
+      "step": 2699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.831507771484893e-06,
+      "loss": 0.4941,
+      "step": 2700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.825267942407507e-06,
+      "loss": 0.4761,
+      "step": 2701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.819028181382904e-06,
+      "loss": 0.8179,
+      "step": 2702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81278849084128e-06,
+      "loss": 0.854,
+      "step": 2703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806548873212804e-06,
+      "loss": 0.7925,
+      "step": 2704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800309330927617e-06,
+      "loss": 0.7891,
+      "step": 2705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794069866415831e-06,
+      "loss": 0.792,
+      "step": 2706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.787830482107525e-06,
+      "loss": 0.8252,
+      "step": 2707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.781591180432746e-06,
+      "loss": 0.8394,
+      "step": 2708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775351963821515e-06,
+      "loss": 0.772,
+      "step": 2709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.769112834703807e-06,
+      "loss": 0.8735,
+      "step": 2710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.762873795509584e-06,
+      "loss": 0.7866,
+      "step": 2711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.756634848668753e-06,
+      "loss": 0.8169,
+      "step": 2712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750395996611195e-06,
+      "loss": 0.9141,
+      "step": 2713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74415724176675e-06,
+      "loss": 0.8149,
+      "step": 2714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.737918586565223e-06,
+      "loss": 0.7886,
+      "step": 2715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731680033436379e-06,
+      "loss": 0.8452,
+      "step": 2716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725441584809943e-06,
+      "loss": 0.877,
+      "step": 2717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.7192032431156e-06,
+      "loss": 0.8701,
+      "step": 2718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.712965010782998e-06,
+      "loss": 0.8374,
+      "step": 2719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.706726890241728e-06,
+      "loss": 0.811,
+      "step": 2720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700488883921353e-06,
+      "loss": 0.873,
+      "step": 2721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.694250994251383e-06,
+      "loss": 0.8501,
+      "step": 2722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.688013223661284e-06,
+      "loss": 0.7969,
+      "step": 2723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.681775574580481e-06,
+      "loss": 0.8335,
+      "step": 2724
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.675538049438347e-06,
+      "loss": 0.7852,
+      "step": 2725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669300650664202e-06,
+      "loss": 0.8491,
+      "step": 2726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663063380687324e-06,
+      "loss": 0.8545,
+      "step": 2727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.656826241936938e-06,
+      "loss": 0.8398,
+      "step": 2728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.650589236842222e-06,
+      "loss": 0.8125,
+      "step": 2729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644352367832295e-06,
+      "loss": 0.7856,
+      "step": 2730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63811563733623e-06,
+      "loss": 0.8252,
+      "step": 2731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.631879047783039e-06,
+      "loss": 0.8848,
+      "step": 2732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.625642601601683e-06,
+      "loss": 0.8535,
+      "step": 2733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619406301221063e-06,
+      "loss": 0.8203,
+      "step": 2734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.613170149070036e-06,
+      "loss": 0.8379,
+      "step": 2735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.606934147577387e-06,
+      "loss": 0.8281,
+      "step": 2736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.600698299171851e-06,
+      "loss": 0.814,
+      "step": 2737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594462606282093e-06,
+      "loss": 0.8276,
+      "step": 2738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588227071336728e-06,
+      "loss": 0.8467,
+      "step": 2739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58199169676431e-06,
+      "loss": 0.8516,
+      "step": 2740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57575648499332e-06,
+      "loss": 0.4868,
+      "step": 2741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.569521438452183e-06,
+      "loss": 0.8618,
+      "step": 2742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563286559569268e-06,
+      "loss": 0.7715,
+      "step": 2743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557051850772857e-06,
+      "loss": 0.8423,
+      "step": 2744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.550817314491182e-06,
+      "loss": 0.8208,
+      "step": 2745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.544582953152405e-06,
+      "loss": 0.7773,
+      "step": 2746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.538348769184618e-06,
+      "loss": 0.7949,
+      "step": 2747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.532114765015848e-06,
+      "loss": 0.8369,
+      "step": 2748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.525880943074046e-06,
+      "loss": 0.8618,
+      "step": 2749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.519647305787095e-06,
+      "loss": 0.8433,
+      "step": 2750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.513413855582803e-06,
+      "loss": 0.8335,
+      "step": 2751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507180594888911e-06,
+      "loss": 0.8735,
+      "step": 2752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.500947526133083e-06,
+      "loss": 0.8687,
+      "step": 2753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.494714651742909e-06,
+      "loss": 0.4624,
+      "step": 2754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.488481974145903e-06,
+      "loss": 0.8599,
+      "step": 2755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.482249495769498e-06,
+      "loss": 0.7832,
+      "step": 2756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.476017219041057e-06,
+      "loss": 0.8511,
+      "step": 2757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.469785146387857e-06,
+      "loss": 0.8438,
+      "step": 2758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.463553280237105e-06,
+      "loss": 0.8774,
+      "step": 2759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45732162301592e-06,
+      "loss": 0.7915,
+      "step": 2760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451090177151343e-06,
+      "loss": 0.8242,
+      "step": 2761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.444858945070326e-06,
+      "loss": 0.8276,
+      "step": 2762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.438627929199744e-06,
+      "loss": 0.7842,
+      "step": 2763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.432397131966393e-06,
+      "loss": 0.8291,
+      "step": 2764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.426166555796974e-06,
+      "loss": 0.8145,
+      "step": 2765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.419936203118102e-06,
+      "loss": 0.8599,
+      "step": 2766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413706076356318e-06,
+      "loss": 0.793,
+      "step": 2767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407476177938055e-06,
+      "loss": 0.8662,
+      "step": 2768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.401246510289673e-06,
+      "loss": 0.8091,
+      "step": 2769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395017075837435e-06,
+      "loss": 0.8408,
+      "step": 2770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388787877007515e-06,
+      "loss": 0.8921,
+      "step": 2771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.382558916225997e-06,
+      "loss": 0.8545,
+      "step": 2772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.376330195918872e-06,
+      "loss": 0.8555,
+      "step": 2773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.370101718512031e-06,
+      "loss": 0.8477,
+      "step": 2774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.363873486431279e-06,
+      "loss": 0.8291,
+      "step": 2775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.357645502102315e-06,
+      "loss": 0.7979,
+      "step": 2776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.351417767950756e-06,
+      "loss": 0.9009,
+      "step": 2777
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.345190286402112e-06,
+      "loss": 0.8188,
+      "step": 2778
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.338963059881796e-06,
+      "loss": 0.8506,
+      "step": 2779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33273609081512e-06,
+      "loss": 0.7769,
+      "step": 2780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326509381627297e-06,
+      "loss": 0.8228,
+      "step": 2781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.320282934743441e-06,
+      "loss": 0.4558,
+      "step": 2782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.314056752588564e-06,
+      "loss": 0.8047,
+      "step": 2783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.307830837587572e-06,
+      "loss": 0.8018,
+      "step": 2784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30160519216527e-06,
+      "loss": 0.8262,
+      "step": 2785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.295379818746353e-06,
+      "loss": 0.4871,
+      "step": 2786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.289154719755413e-06,
+      "loss": 0.8413,
+      "step": 2787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.282929897616938e-06,
+      "loss": 0.8101,
+      "step": 2788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276705354755304e-06,
+      "loss": 0.8149,
+      "step": 2789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27048109359478e-06,
+      "loss": 0.4753,
+      "step": 2790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26425711655953e-06,
+      "loss": 0.834,
+      "step": 2791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.258033426073592e-06,
+      "loss": 0.8105,
+      "step": 2792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25181002456091e-06,
+      "loss": 0.8794,
+      "step": 2793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24558691444531e-06,
+      "loss": 0.7646,
+      "step": 2794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.239364098150493e-06,
+      "loss": 0.8042,
+      "step": 2795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.233141578100068e-06,
+      "loss": 0.7739,
+      "step": 2796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.226919356717509e-06,
+      "loss": 0.7803,
+      "step": 2797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.220697436426182e-06,
+      "loss": 0.7876,
+      "step": 2798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214475819649334e-06,
+      "loss": 0.8286,
+      "step": 2799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.208254508810091e-06,
+      "loss": 0.8374,
+      "step": 2800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.202033506331469e-06,
+      "loss": 0.8574,
+      "step": 2801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195812814636355e-06,
+      "loss": 0.7974,
+      "step": 2802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18959243614752e-06,
+      "loss": 0.8291,
+      "step": 2803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.183372373287607e-06,
+      "loss": 0.8438,
+      "step": 2804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.177152628479142e-06,
+      "loss": 0.8193,
+      "step": 2805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.170933204144523e-06,
+      "loss": 0.8252,
+      "step": 2806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.164714102706028e-06,
+      "loss": 0.8022,
+      "step": 2807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.158495326585806e-06,
+      "loss": 0.8682,
+      "step": 2808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.152276878205879e-06,
+      "loss": 0.8384,
+      "step": 2809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.146058759988149e-06,
+      "loss": 0.8008,
+      "step": 2810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.139840974354369e-06,
+      "loss": 0.8213,
+      "step": 2811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.133623523726187e-06,
+      "loss": 0.8228,
+      "step": 2812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127406410525106e-06,
+      "loss": 0.8467,
+      "step": 2813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121189637172496e-06,
+      "loss": 0.7803,
+      "step": 2814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11497320608961e-06,
+      "loss": 0.8452,
+      "step": 2815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.108757119697554e-06,
+      "loss": 0.9087,
+      "step": 2816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1025413804173e-06,
+      "loss": 0.771,
+      "step": 2817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.096325990669689e-06,
+      "loss": 0.811,
+      "step": 2818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090110952875423e-06,
+      "loss": 0.8438,
+      "step": 2819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.083896269455076e-06,
+      "loss": 0.7744,
+      "step": 2820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07768194282907e-06,
+      "loss": 0.7988,
+      "step": 2821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0714679754177e-06,
+      "loss": 0.8462,
+      "step": 2822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.065254369641113e-06,
+      "loss": 0.793,
+      "step": 2823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.059041127919315e-06,
+      "loss": 0.8555,
+      "step": 2824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.052828252672179e-06,
+      "loss": 0.8833,
+      "step": 2825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046615746319427e-06,
+      "loss": 0.8066,
+      "step": 2826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.040403611280641e-06,
+      "loss": 0.8374,
+      "step": 2827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.034191849975258e-06,
+      "loss": 0.8047,
+      "step": 2828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.027980464822565e-06,
+      "loss": 0.7891,
+      "step": 2829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.021769458241708e-06,
+      "loss": 0.8394,
+      "step": 2830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015558832651687e-06,
+      "loss": 0.7681,
+      "step": 2831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.009348590471346e-06,
+      "loss": 0.8511,
+      "step": 2832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.003138734119385e-06,
+      "loss": 0.8667,
+      "step": 2833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.996929266014361e-06,
+      "loss": 0.7896,
+      "step": 2834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.990720188574657e-06,
+      "loss": 0.8252,
+      "step": 2835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984511504218529e-06,
+      "loss": 0.4634,
+      "step": 2836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.978303215364066e-06,
+      "loss": 0.7871,
+      "step": 2837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.972095324429204e-06,
+      "loss": 0.8555,
+      "step": 2838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.965887833831732e-06,
+      "loss": 0.8086,
+      "step": 2839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.959680745989278e-06,
+      "loss": 0.835,
+      "step": 2840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.953474063319305e-06,
+      "loss": 0.8921,
+      "step": 2841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.947267788239131e-06,
+      "loss": 0.8623,
+      "step": 2842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941061923165907e-06,
+      "loss": 0.7939,
+      "step": 2843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93485647051663e-06,
+      "loss": 0.8667,
+      "step": 2844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.928651432708135e-06,
+      "loss": 0.8501,
+      "step": 2845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922446812157094e-06,
+      "loss": 0.853,
+      "step": 2846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916242611280013e-06,
+      "loss": 0.7876,
+      "step": 2847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.910038832493237e-06,
+      "loss": 0.7817,
+      "step": 2848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.903835478212954e-06,
+      "loss": 0.8696,
+      "step": 2849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897632550855179e-06,
+      "loss": 0.8472,
+      "step": 2850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891430052835758e-06,
+      "loss": 0.8262,
+      "step": 2851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88522798657038e-06,
+      "loss": 0.8569,
+      "step": 2852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.879026354474556e-06,
+      "loss": 0.8682,
+      "step": 2853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872825158963628e-06,
+      "loss": 0.77,
+      "step": 2854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86662440245278e-06,
+      "loss": 0.8135,
+      "step": 2855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.860424087357011e-06,
+      "loss": 0.8457,
+      "step": 2856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.854224216091154e-06,
+      "loss": 0.8418,
+      "step": 2857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848024791069876e-06,
+      "loss": 0.8164,
+      "step": 2858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.841825814707653e-06,
+      "loss": 0.8179,
+      "step": 2859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.835627289418803e-06,
+      "loss": 0.8164,
+      "step": 2860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82942921761746e-06,
+      "loss": 0.7983,
+      "step": 2861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82323160171758e-06,
+      "loss": 0.8223,
+      "step": 2862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81703444413295e-06,
+      "loss": 0.8633,
+      "step": 2863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.810837747277174e-06,
+      "loss": 0.8462,
+      "step": 2864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.804641513563668e-06,
+      "loss": 0.8423,
+      "step": 2865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79844574540568e-06,
+      "loss": 0.8315,
+      "step": 2866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79225044521627e-06,
+      "loss": 0.8569,
+      "step": 2867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78605561540832e-06,
+      "loss": 0.769,
+      "step": 2868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.779861258394528e-06,
+      "loss": 0.8564,
+      "step": 2869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.773667376587404e-06,
+      "loss": 0.8516,
+      "step": 2870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.767473972399275e-06,
+      "loss": 0.8511,
+      "step": 2871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761281048242279e-06,
+      "loss": 0.8232,
+      "step": 2872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755088606528375e-06,
+      "loss": 0.7388,
+      "step": 2873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.748896649669328e-06,
+      "loss": 0.8574,
+      "step": 2874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.742705180076717e-06,
+      "loss": 0.7988,
+      "step": 2875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.736514200161929e-06,
+      "loss": 0.854,
+      "step": 2876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73032371233616e-06,
+      "loss": 0.8105,
+      "step": 2877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724133719010413e-06,
+      "loss": 0.8687,
+      "step": 2878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.717944222595505e-06,
+      "loss": 0.8564,
+      "step": 2879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.711755225502053e-06,
+      "loss": 0.8262,
+      "step": 2880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.705566730140482e-06,
+      "loss": 0.7827,
+      "step": 2881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.699378738921028e-06,
+      "loss": 0.7837,
+      "step": 2882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.693191254253712e-06,
+      "loss": 0.9023,
+      "step": 2883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.687004278548378e-06,
+      "loss": 0.7373,
+      "step": 2884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.680817814214662e-06,
+      "loss": 0.8213,
+      "step": 2885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.674631863661998e-06,
+      "loss": 0.8369,
+      "step": 2886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66844642929963e-06,
+      "loss": 0.8032,
+      "step": 2887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.662261513536598e-06,
+      "loss": 0.8345,
+      "step": 2888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.656077118781728e-06,
+      "loss": 0.4932,
+      "step": 2889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.649893247443658e-06,
+      "loss": 0.7935,
+      "step": 2890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.643709901930814e-06,
+      "loss": 0.5105,
+      "step": 2891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.637527084651421e-06,
+      "loss": 0.8359,
+      "step": 2892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.631344798013497e-06,
+      "loss": 0.8433,
+      "step": 2893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.625163044424858e-06,
+      "loss": 0.8286,
+      "step": 2894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6189818262931e-06,
+      "loss": 0.8237,
+      "step": 2895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.612801146025616e-06,
+      "loss": 0.8218,
+      "step": 2896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.606621006029601e-06,
+      "loss": 0.8389,
+      "step": 2897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600441408712027e-06,
+      "loss": 0.8501,
+      "step": 2898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.594262356479656e-06,
+      "loss": 0.7949,
+      "step": 2899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.588083851739042e-06,
+      "loss": 0.8018,
+      "step": 2900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58190589689652e-06,
+      "loss": 0.8237,
+      "step": 2901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.575728494358215e-06,
+      "loss": 0.8706,
+      "step": 2902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.569551646530036e-06,
+      "loss": 0.8418,
+      "step": 2903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56337535581768e-06,
+      "loss": 0.7969,
+      "step": 2904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.557199624626618e-06,
+      "loss": 0.7939,
+      "step": 2905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.551024455362116e-06,
+      "loss": 0.7983,
+      "step": 2906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5448498504292e-06,
+      "loss": 0.7695,
+      "step": 2907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5386758122327e-06,
+      "loss": 0.8413,
+      "step": 2908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.532502343177212e-06,
+      "loss": 0.8345,
+      "step": 2909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.526329445667111e-06,
+      "loss": 0.8784,
+      "step": 2910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.520157122106556e-06,
+      "loss": 0.7798,
+      "step": 2911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.513985374899477e-06,
+      "loss": 0.4343,
+      "step": 2912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.507814206449576e-06,
+      "loss": 0.8359,
+      "step": 2913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.501643619160338e-06,
+      "loss": 0.7905,
+      "step": 2914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.495473615435015e-06,
+      "loss": 0.7935,
+      "step": 2915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489304197676637e-06,
+      "loss": 0.834,
+      "step": 2916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.483135368288004e-06,
+      "loss": 0.7979,
+      "step": 2917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.476967129671683e-06,
+      "loss": 0.8457,
+      "step": 2918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.470799484230019e-06,
+      "loss": 0.8276,
+      "step": 2919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.464632434365112e-06,
+      "loss": 0.4541,
+      "step": 2920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.458465982478848e-06,
+      "loss": 0.814,
+      "step": 2921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.452300130972868e-06,
+      "loss": 0.8696,
+      "step": 2922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.446134882248581e-06,
+      "loss": 0.8594,
+      "step": 2923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.439970238707163e-06,
+      "loss": 0.8076,
+      "step": 2924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.433806202749559e-06,
+      "loss": 0.8032,
+      "step": 2925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.427642776776462e-06,
+      "loss": 0.833,
+      "step": 2926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.421479963188345e-06,
+      "loss": 0.8027,
+      "step": 2927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.415317764385434e-06,
+      "loss": 0.8433,
+      "step": 2928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.409156182767713e-06,
+      "loss": 0.813,
+      "step": 2929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.402995220734936e-06,
+      "loss": 0.8237,
+      "step": 2930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.396834880686607e-06,
+      "loss": 0.79,
+      "step": 2931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390675165021987e-06,
+      "loss": 0.5056,
+      "step": 2932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.384516076140096e-06,
+      "loss": 0.8262,
+      "step": 2933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.378357616439711e-06,
+      "loss": 0.8574,
+      "step": 2934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.372199788319366e-06,
+      "loss": 0.8462,
+      "step": 2935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.366042594177345e-06,
+      "loss": 0.8418,
+      "step": 2936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.359886036411691e-06,
+      "loss": 0.7588,
+      "step": 2937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.353730117420186e-06,
+      "loss": 0.8345,
+      "step": 2938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.347574839600375e-06,
+      "loss": 0.8354,
+      "step": 2939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.341420205349554e-06,
+      "loss": 0.8618,
+      "step": 2940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.335266217064761e-06,
+      "loss": 0.8545,
+      "step": 2941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329112877142787e-06,
+      "loss": 0.7637,
+      "step": 2942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32296018798017e-06,
+      "loss": 0.834,
+      "step": 2943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.316808151973193e-06,
+      "loss": 0.7039,
+      "step": 2944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.310656771517886e-06,
+      "loss": 0.7905,
+      "step": 2945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.304506049010022e-06,
+      "loss": 0.8218,
+      "step": 2946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29835598684512e-06,
+      "loss": 0.7842,
+      "step": 2947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29220658741844e-06,
+      "loss": 0.8569,
+      "step": 2948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.286057853124992e-06,
+      "loss": 0.7622,
+      "step": 2949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.279909786359506e-06,
+      "loss": 0.7764,
+      "step": 2950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.273762389516478e-06,
+      "loss": 0.877,
+      "step": 2951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.267615664990123e-06,
+      "loss": 0.4536,
+      "step": 2952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.261469615174404e-06,
+      "loss": 0.8057,
+      "step": 2953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.255324242463021e-06,
+      "loss": 0.4346,
+      "step": 2954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.24917954924941e-06,
+      "loss": 0.7368,
+      "step": 2955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.243035537926736e-06,
+      "loss": 0.832,
+      "step": 2956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.236892210887908e-06,
+      "loss": 0.8975,
+      "step": 2957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.230749570525555e-06,
+      "loss": 0.7927,
+      "step": 2958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.224607619232057e-06,
+      "loss": 0.8042,
+      "step": 2959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.218466359399513e-06,
+      "loss": 0.8604,
+      "step": 2960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.212325793419757e-06,
+      "loss": 0.8096,
+      "step": 2961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.206185923684347e-06,
+      "loss": 0.832,
+      "step": 2962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.200046752584574e-06,
+      "loss": 0.4749,
+      "step": 2963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193908282511463e-06,
+      "loss": 0.7798,
+      "step": 2964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.187770515855754e-06,
+      "loss": 0.8379,
+      "step": 2965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.181633455007925e-06,
+      "loss": 0.8164,
+      "step": 2966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17549710235817e-06,
+      "loss": 0.7847,
+      "step": 2967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.169361460296406e-06,
+      "loss": 0.8408,
+      "step": 2968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.163226531212286e-06,
+      "loss": 0.8477,
+      "step": 2969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.157092317495172e-06,
+      "loss": 0.8115,
+      "step": 2970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150958821534154e-06,
+      "loss": 0.7881,
+      "step": 2971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.144826045718037e-06,
+      "loss": 0.7954,
+      "step": 2972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13869399243536e-06,
+      "loss": 0.8662,
+      "step": 2973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.132562664074354e-06,
+      "loss": 0.8569,
+      "step": 2974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126432063022997e-06,
+      "loss": 0.7515,
+      "step": 2975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.120302191668965e-06,
+      "loss": 0.7944,
+      "step": 2976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.114173052399652e-06,
+      "loss": 0.8062,
+      "step": 2977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.108044647602175e-06,
+      "loss": 0.8447,
+      "step": 2978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.101916979663363e-06,
+      "loss": 0.811,
+      "step": 2979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.095790050969747e-06,
+      "loss": 0.7939,
+      "step": 2980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.089663863907584e-06,
+      "loss": 0.7754,
+      "step": 2981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.083538420862828e-06,
+      "loss": 0.8389,
+      "step": 2982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.077413724221161e-06,
+      "loss": 0.7773,
+      "step": 2983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.071289776367962e-06,
+      "loss": 0.8394,
+      "step": 2984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.065166579688321e-06,
+      "loss": 0.7888,
+      "step": 2985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.059044136567033e-06,
+      "loss": 0.8076,
+      "step": 2986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.052922449388601e-06,
+      "loss": 0.8242,
+      "step": 2987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04680152053724e-06,
+      "loss": 0.8135,
+      "step": 2988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.040681352396862e-06,
+      "loss": 0.8608,
+      "step": 2989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034561947351084e-06,
+      "loss": 0.8682,
+      "step": 2990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02844330778323e-06,
+      "loss": 0.8647,
+      "step": 2991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.022325436076317e-06,
+      "loss": 0.8442,
+      "step": 2992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.016208334613073e-06,
+      "loss": 0.8213,
+      "step": 2993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01009200577592e-06,
+      "loss": 0.7778,
+      "step": 2994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.003976451946981e-06,
+      "loss": 0.8193,
+      "step": 2995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.997861675508078e-06,
+      "loss": 0.8149,
+      "step": 2996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.991747678840732e-06,
+      "loss": 0.8013,
+      "step": 2997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.985634464326148e-06,
+      "loss": 0.8506,
+      "step": 2998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.979522034345244e-06,
+      "loss": 0.8423,
+      "step": 2999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.973410391278623e-06,
+      "loss": 0.8423,
+      "step": 3000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.967299537506578e-06,
+      "loss": 0.7363,
+      "step": 3001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.961189475409104e-06,
+      "loss": 0.8184,
+      "step": 3002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.955080207365886e-06,
+      "loss": 0.8721,
+      "step": 3003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.94897173575629e-06,
+      "loss": 0.8101,
+      "step": 3004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.942864062959383e-06,
+      "loss": 0.8286,
+      "step": 3005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.936757191353912e-06,
+      "loss": 0.8135,
+      "step": 3006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93065112331832e-06,
+      "loss": 0.8032,
+      "step": 3007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.924545861230734e-06,
+      "loss": 0.7832,
+      "step": 3008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.918441407468966e-06,
+      "loss": 0.4714,
+      "step": 3009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.912337764410514e-06,
+      "loss": 0.8467,
+      "step": 3010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.906234934432555e-06,
+      "loss": 0.8379,
+      "step": 3011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90013291991196e-06,
+      "loss": 0.8892,
+      "step": 3012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.894031723225278e-06,
+      "loss": 0.7961,
+      "step": 3013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.887931346748733e-06,
+      "loss": 0.8057,
+      "step": 3014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88183179285824e-06,
+      "loss": 0.8135,
+      "step": 3015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.875733063929382e-06,
+      "loss": 0.8237,
+      "step": 3016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.869635162337433e-06,
+      "loss": 0.7803,
+      "step": 3017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.863538090457336e-06,
+      "loss": 0.8428,
+      "step": 3018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.857441850663714e-06,
+      "loss": 0.8115,
+      "step": 3019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.851346445330865e-06,
+      "loss": 0.8022,
+      "step": 3020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.845251876832771e-06,
+      "loss": 0.7983,
+      "step": 3021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.839158147543064e-06,
+      "loss": 0.8154,
+      "step": 3022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.833065259835077e-06,
+      "loss": 0.7974,
+      "step": 3023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8269732160818e-06,
+      "loss": 0.8188,
+      "step": 3024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.820882018655895e-06,
+      "loss": 0.8218,
+      "step": 3025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8147916699297e-06,
+      "loss": 0.7725,
+      "step": 3026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.808702172275221e-06,
+      "loss": 0.8149,
+      "step": 3027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.802613528064127e-06,
+      "loss": 0.8022,
+      "step": 3028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796525739667758e-06,
+      "loss": 0.832,
+      "step": 3029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.790438809457122e-06,
+      "loss": 0.8154,
+      "step": 3030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.784352739802893e-06,
+      "loss": 0.8726,
+      "step": 3031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77826753307541e-06,
+      "loss": 0.7905,
+      "step": 3032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.772183191644667e-06,
+      "loss": 0.8252,
+      "step": 3033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.766099717880344e-06,
+      "loss": 0.8184,
+      "step": 3034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76001711415175e-06,
+      "loss": 0.8008,
+      "step": 3035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.753935382827885e-06,
+      "loss": 0.8408,
+      "step": 3036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.747854526277393e-06,
+      "loss": 0.8491,
+      "step": 3037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74177454686858e-06,
+      "loss": 0.7686,
+      "step": 3038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73569544696941e-06,
+      "loss": 0.8486,
+      "step": 3039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72961722894752e-06,
+      "loss": 0.8252,
+      "step": 3040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.723539895170168e-06,
+      "loss": 0.7944,
+      "step": 3041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.717463448004305e-06,
+      "loss": 0.7622,
+      "step": 3042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.711387889816517e-06,
+      "loss": 0.7576,
+      "step": 3043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.705313222973044e-06,
+      "loss": 0.8242,
+      "step": 3044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69923944983979e-06,
+      "loss": 0.8369,
+      "step": 3045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6931665727823e-06,
+      "loss": 0.8208,
+      "step": 3046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.687094594165772e-06,
+      "loss": 0.8066,
+      "step": 3047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681023516355057e-06,
+      "loss": 0.8179,
+      "step": 3048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.674953341714651e-06,
+      "loss": 0.8037,
+      "step": 3049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.668884072608706e-06,
+      "loss": 0.7871,
+      "step": 3050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.662815711401016e-06,
+      "loss": 0.853,
+      "step": 3051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.65674826045502e-06,
+      "loss": 0.833,
+      "step": 3052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.650681722133803e-06,
+      "loss": 0.8252,
+      "step": 3053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.644616098800095e-06,
+      "loss": 0.4849,
+      "step": 3054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.638551392816275e-06,
+      "loss": 0.4788,
+      "step": 3055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632487606544359e-06,
+      "loss": 0.811,
+      "step": 3056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6264247423460025e-06,
+      "loss": 0.7808,
+      "step": 3057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.620362802582515e-06,
+      "loss": 0.7798,
+      "step": 3058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.614301789614822e-06,
+      "loss": 0.8672,
+      "step": 3059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.608241705803512e-06,
+      "loss": 0.8149,
+      "step": 3060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.602182553508799e-06,
+      "loss": 0.8389,
+      "step": 3061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.59612433509054e-06,
+      "loss": 0.4756,
+      "step": 3062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.59006705290822e-06,
+      "loss": 0.8892,
+      "step": 3063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.584010709320975e-06,
+      "loss": 0.7935,
+      "step": 3064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.577955306687552e-06,
+      "loss": 0.8726,
+      "step": 3065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.571900847366353e-06,
+      "loss": 0.7896,
+      "step": 3066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.565847333715405e-06,
+      "loss": 0.8423,
+      "step": 3067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.559794768092361e-06,
+      "loss": 0.8193,
+      "step": 3068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.553743152854516e-06,
+      "loss": 0.8569,
+      "step": 3069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5476924903587865e-06,
+      "loss": 0.8228,
+      "step": 3070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.541642782961719e-06,
+      "loss": 0.8403,
+      "step": 3071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.535594033019489e-06,
+      "loss": 0.8169,
+      "step": 3072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.529546242887898e-06,
+      "loss": 0.7783,
+      "step": 3073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.52349941492238e-06,
+      "loss": 0.833,
+      "step": 3074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.517453551477986e-06,
+      "loss": 0.7209,
+      "step": 3075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.511408654909397e-06,
+      "loss": 0.856,
+      "step": 3076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.505364727570912e-06,
+      "loss": 0.8564,
+      "step": 3077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.499321771816452e-06,
+      "loss": 0.8613,
+      "step": 3078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.493279789999573e-06,
+      "loss": 0.4663,
+      "step": 3079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4872387844734365e-06,
+      "loss": 0.8833,
+      "step": 3080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.481198757590826e-06,
+      "loss": 0.8521,
+      "step": 3081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.475159711704156e-06,
+      "loss": 0.8433,
+      "step": 3082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.46912164916544e-06,
+      "loss": 0.8408,
+      "step": 3083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.463084572326325e-06,
+      "loss": 0.7861,
+      "step": 3084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.457048483538063e-06,
+      "loss": 0.8379,
+      "step": 3085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.451013385151528e-06,
+      "loss": 0.7266,
+      "step": 3086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444979279517206e-06,
+      "loss": 0.8184,
+      "step": 3087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.438946168985202e-06,
+      "loss": 0.4714,
+      "step": 3088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.432914055905213e-06,
+      "loss": 0.812,
+      "step": 3089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.426882942626577e-06,
+      "loss": 0.8193,
+      "step": 3090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.42085283149822e-06,
+      "loss": 0.8105,
+      "step": 3091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.414823724868686e-06,
+      "loss": 0.8569,
+      "step": 3092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408795625086131e-06,
+      "loss": 0.8481,
+      "step": 3093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4027685344983155e-06,
+      "loss": 0.7993,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.396742455452604e-06,
+      "loss": 0.8252,
+      "step": 3095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.390717390295967e-06,
+      "loss": 0.8389,
+      "step": 3096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.384693341374987e-06,
+      "loss": 0.811,
+      "step": 3097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.378670311035846e-06,
+      "loss": 0.7871,
+      "step": 3098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.37264830162433e-06,
+      "loss": 0.8062,
+      "step": 3099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3666273154858284e-06,
+      "loss": 0.7295,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.360607354965327e-06,
+      "loss": 0.8691,
+      "step": 3101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354588422407416e-06,
+      "loss": 0.7993,
+      "step": 3102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34857052015629e-06,
+      "loss": 0.8354,
+      "step": 3103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.342553650555735e-06,
+      "loss": 0.8335,
+      "step": 3104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.336537815949134e-06,
+      "loss": 0.8555,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.330523018679483e-06,
+      "loss": 0.8706,
+      "step": 3106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.324509261089343e-06,
+      "loss": 0.8804,
+      "step": 3107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3184965455209015e-06,
+      "loss": 0.8115,
+      "step": 3108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.312484874315922e-06,
+      "loss": 0.8428,
+      "step": 3109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.306474249815767e-06,
+      "loss": 0.8433,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.300464674361389e-06,
+      "loss": 0.7925,
+      "step": 3111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.294456150293342e-06,
+      "loss": 0.8389,
+      "step": 3112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2884486799517496e-06,
+      "loss": 0.8203,
+      "step": 3113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.282442265676346e-06,
+      "loss": 0.8364,
+      "step": 3114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.276436909806442e-06,
+      "loss": 0.8354,
+      "step": 3115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.270432614680939e-06,
+      "loss": 0.4888,
+      "step": 3116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.26442938263833e-06,
+      "loss": 0.769,
+      "step": 3117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25842721601669e-06,
+      "loss": 0.8071,
+      "step": 3118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2524261171536755e-06,
+      "loss": 0.812,
+      "step": 3119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.246426088386532e-06,
+      "loss": 0.8374,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.240427132052085e-06,
+      "loss": 0.7915,
+      "step": 3121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.234429250486747e-06,
+      "loss": 0.8174,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.228432446026511e-06,
+      "loss": 0.8555,
+      "step": 3123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.222436721006947e-06,
+      "loss": 0.8584,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.216442077763204e-06,
+      "loss": 0.8398,
+      "step": 3125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.21044851863001e-06,
+      "loss": 0.4983,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204456045941679e-06,
+      "loss": 0.7832,
+      "step": 3127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.198464662032092e-06,
+      "loss": 0.7886,
+      "step": 3128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1924743692347084e-06,
+      "loss": 0.8848,
+      "step": 3129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186485169882572e-06,
+      "loss": 0.8018,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.18049706630828e-06,
+      "loss": 0.7734,
+      "step": 3131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.174510060844022e-06,
+      "loss": 0.8452,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.168524155821555e-06,
+      "loss": 0.7808,
+      "step": 3133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1625393535722025e-06,
+      "loss": 0.8271,
+      "step": 3134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.156555656426858e-06,
+      "loss": 0.8184,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.150573066716001e-06,
+      "loss": 0.8149,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.144591586769652e-06,
+      "loss": 0.791,
+      "step": 3137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.138611218917424e-06,
+      "loss": 0.7886,
+      "step": 3138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.132631965488482e-06,
+      "loss": 0.8022,
+      "step": 3139
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.126653828811563e-06,
+      "loss": 0.8379,
+      "step": 3140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.12067681121497e-06,
+      "loss": 0.8584,
+      "step": 3141
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.114700915026571e-06,
+      "loss": 0.8623,
+      "step": 3142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.10872614257379e-06,
+      "loss": 0.8315,
+      "step": 3143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102752496183617e-06,
+      "loss": 0.8394,
+      "step": 3144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.096779978182604e-06,
+      "loss": 0.8413,
+      "step": 3145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.090808590896869e-06,
+      "loss": 0.8545,
+      "step": 3146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.08483833665208e-06,
+      "loss": 0.749,
+      "step": 3147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078869217773474e-06,
+      "loss": 0.8208,
+      "step": 3148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.072901236585832e-06,
+      "loss": 0.8003,
+      "step": 3149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.066934395413501e-06,
+      "loss": 0.4753,
+      "step": 3150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0609686965803884e-06,
+      "loss": 0.8208,
+      "step": 3151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.055004142409947e-06,
+      "loss": 0.8003,
+      "step": 3152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.04904073522519e-06,
+      "loss": 0.8784,
+      "step": 3153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.043078477348677e-06,
+      "loss": 0.7681,
+      "step": 3154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.037117371102535e-06,
+      "loss": 0.8306,
+      "step": 3155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.031157418808423e-06,
+      "loss": 0.7388,
+      "step": 3156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.025198622787563e-06,
+      "loss": 0.8643,
+      "step": 3157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.019240985360725e-06,
+      "loss": 0.749,
+      "step": 3158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013284508848219e-06,
+      "loss": 0.8296,
+      "step": 3159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.007329195569921e-06,
+      "loss": 0.7949,
+      "step": 3160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.001375047845239e-06,
+      "loss": 0.8008,
+      "step": 3161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.995422067993128e-06,
+      "loss": 0.8271,
+      "step": 3162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.989470258332094e-06,
+      "loss": 0.8423,
+      "step": 3163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983519621180182e-06,
+      "loss": 0.8213,
+      "step": 3164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.977570158854987e-06,
+      "loss": 0.8408,
+      "step": 3165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97162187367364e-06,
+      "loss": 0.4565,
+      "step": 3166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.965674767952817e-06,
+      "loss": 0.8872,
+      "step": 3167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9597288440087326e-06,
+      "loss": 0.877,
+      "step": 3168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.95378410415714e-06,
+      "loss": 0.4792,
+      "step": 3169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.947840550713336e-06,
+      "loss": 0.8003,
+      "step": 3170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.941898185992153e-06,
+      "loss": 0.8931,
+      "step": 3171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.935957012307957e-06,
+      "loss": 0.4968,
+      "step": 3172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.930017031974661e-06,
+      "loss": 0.8252,
+      "step": 3173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.924078247305694e-06,
+      "loss": 0.8027,
+      "step": 3174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.918140660614039e-06,
+      "loss": 0.8413,
+      "step": 3175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9122042742122e-06,
+      "loss": 0.79,
+      "step": 3176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906269090412221e-06,
+      "loss": 0.8247,
+      "step": 3177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.900335111525669e-06,
+      "loss": 0.7783,
+      "step": 3178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894402339863655e-06,
+      "loss": 0.8101,
+      "step": 3179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.888470777736806e-06,
+      "loss": 0.8247,
+      "step": 3180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.882540427455285e-06,
+      "loss": 0.7881,
+      "step": 3181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.876611291328783e-06,
+      "loss": 0.7466,
+      "step": 3182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8706833716665125e-06,
+      "loss": 0.8247,
+      "step": 3183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.864756670777222e-06,
+      "loss": 0.7935,
+      "step": 3184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.85883119096918e-06,
+      "loss": 0.8252,
+      "step": 3185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.852906934550177e-06,
+      "loss": 0.8892,
+      "step": 3186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.846983903827529e-06,
+      "loss": 0.771,
+      "step": 3187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.841062101108072e-06,
+      "loss": 0.8247,
+      "step": 3188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.835141528698174e-06,
+      "loss": 0.8472,
+      "step": 3189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829222188903713e-06,
+      "loss": 0.7393,
+      "step": 3190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.823304084030094e-06,
+      "loss": 0.8057,
+      "step": 3191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.817387216382232e-06,
+      "loss": 0.8623,
+      "step": 3192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811471588264565e-06,
+      "loss": 0.8633,
+      "step": 3193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8055572019810545e-06,
+      "loss": 0.7827,
+      "step": 3194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799644059835171e-06,
+      "loss": 0.8062,
+      "step": 3195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7937321641298985e-06,
+      "loss": 0.8848,
+      "step": 3196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7878215171677494e-06,
+      "loss": 0.8301,
+      "step": 3197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7819121212507265e-06,
+      "loss": 0.8428,
+      "step": 3198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.776003978680366e-06,
+      "loss": 0.8228,
+      "step": 3199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.770097091757708e-06,
+      "loss": 0.8662,
+      "step": 3200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.764191462783303e-06,
+      "loss": 0.8677,
+      "step": 3201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.75828709405721e-06,
+      "loss": 0.8047,
+      "step": 3202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.752383987879008e-06,
+      "loss": 0.8301,
+      "step": 3203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7464821465477685e-06,
+      "loss": 0.8608,
+      "step": 3204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740581572362079e-06,
+      "loss": 0.8608,
+      "step": 3205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734682267620033e-06,
+      "loss": 0.7949,
+      "step": 3206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.728784234619228e-06,
+      "loss": 0.9053,
+      "step": 3207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.722887475656771e-06,
+      "loss": 0.7905,
+      "step": 3208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716991993029268e-06,
+      "loss": 0.8218,
+      "step": 3209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.711097789032825e-06,
+      "loss": 0.7915,
+      "step": 3210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.705204865963059e-06,
+      "loss": 0.8496,
+      "step": 3211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699313226115076e-06,
+      "loss": 0.8301,
+      "step": 3212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693422871783497e-06,
+      "loss": 0.7891,
+      "step": 3213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.687533805262432e-06,
+      "loss": 0.8198,
+      "step": 3214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.681646028845492e-06,
+      "loss": 0.8003,
+      "step": 3215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.675759544825784e-06,
+      "loss": 0.7546,
+      "step": 3216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.669874355495911e-06,
+      "loss": 0.4285,
+      "step": 3217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66399046314798e-06,
+      "loss": 0.8218,
+      "step": 3218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.658107870073585e-06,
+      "loss": 0.7983,
+      "step": 3219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6522265785638115e-06,
+      "loss": 0.7891,
+      "step": 3220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.646346590909252e-06,
+      "loss": 0.7764,
+      "step": 3221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.64046790939997e-06,
+      "loss": 0.8228,
+      "step": 3222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.634590536325538e-06,
+      "loss": 0.7974,
+      "step": 3223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.628714473975013e-06,
+      "loss": 0.7847,
+      "step": 3224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.62283972463694e-06,
+      "loss": 0.8242,
+      "step": 3225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6169662905993536e-06,
+      "loss": 0.8252,
+      "step": 3226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6110941741497795e-06,
+      "loss": 0.8306,
+      "step": 3227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.605223377575224e-06,
+      "loss": 0.8208,
+      "step": 3228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.599353903162182e-06,
+      "loss": 0.7646,
+      "step": 3229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593485753196637e-06,
+      "loss": 0.8062,
+      "step": 3230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.587618929964048e-06,
+      "loss": 0.8374,
+      "step": 3231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.581753435749372e-06,
+      "loss": 0.8276,
+      "step": 3232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.575889272837034e-06,
+      "loss": 0.7378,
+      "step": 3233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.570026443510944e-06,
+      "loss": 0.7915,
+      "step": 3234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564164950054498e-06,
+      "loss": 0.772,
+      "step": 3235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.558304794750564e-06,
+      "loss": 0.7808,
+      "step": 3236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.552445979881499e-06,
+      "loss": 0.7705,
+      "step": 3237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.546588507729128e-06,
+      "loss": 0.8389,
+      "step": 3238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.540732380574759e-06,
+      "loss": 0.7617,
+      "step": 3239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.534877600699172e-06,
+      "loss": 0.8301,
+      "step": 3240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.529024170382623e-06,
+      "loss": 0.4355,
+      "step": 3241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523172091904848e-06,
+      "loss": 0.8589,
+      "step": 3242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.51732136754505e-06,
+      "loss": 0.8237,
+      "step": 3243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.511471999581906e-06,
+      "loss": 0.8037,
+      "step": 3244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.505623990293572e-06,
+      "loss": 0.7827,
+      "step": 3245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4997773419576596e-06,
+      "loss": 0.7974,
+      "step": 3246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.493932056851264e-06,
+      "loss": 0.7603,
+      "step": 3247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.488088137250944e-06,
+      "loss": 0.8086,
+      "step": 3248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.482245585432727e-06,
+      "loss": 0.7739,
+      "step": 3249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4764044036721074e-06,
+      "loss": 0.9033,
+      "step": 3250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.470564594244051e-06,
+      "loss": 0.8315,
+      "step": 3251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4647261594229784e-06,
+      "loss": 0.8711,
+      "step": 3252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458889101482784e-06,
+      "loss": 0.7739,
+      "step": 3253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4530534226968226e-06,
+      "loss": 0.812,
+      "step": 3254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447219125337912e-06,
+      "loss": 0.8159,
+      "step": 3255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4413862116783335e-06,
+      "loss": 0.8296,
+      "step": 3256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43555468398983e-06,
+      "loss": 0.8169,
+      "step": 3257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429724544543598e-06,
+      "loss": 0.877,
+      "step": 3258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.423895795610302e-06,
+      "loss": 0.8135,
+      "step": 3259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.418068439460059e-06,
+      "loss": 0.832,
+      "step": 3260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4122424783624474e-06,
+      "loss": 0.8457,
+      "step": 3261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4064179145865014e-06,
+      "loss": 0.8618,
+      "step": 3262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.40059475040071e-06,
+      "loss": 0.8184,
+      "step": 3263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.394772988073015e-06,
+      "loss": 0.8062,
+      "step": 3264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.388952629870813e-06,
+      "loss": 0.835,
+      "step": 3265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.383133678060959e-06,
+      "loss": 0.7588,
+      "step": 3266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.377316134909757e-06,
+      "loss": 0.7974,
+      "step": 3267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3715000026829564e-06,
+      "loss": 0.8335,
+      "step": 3268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365685283645774e-06,
+      "loss": 0.7974,
+      "step": 3269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.35987198006285e-06,
+      "loss": 0.7588,
+      "step": 3270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.354060094198297e-06,
+      "loss": 0.7456,
+      "step": 3271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.348249628315667e-06,
+      "loss": 0.8184,
+      "step": 3272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.342440584677956e-06,
+      "loss": 0.9131,
+      "step": 3273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.336632965547605e-06,
+      "loss": 0.7925,
+      "step": 3274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.330826773186512e-06,
+      "loss": 0.731,
+      "step": 3275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.325022009856006e-06,
+      "loss": 0.7593,
+      "step": 3276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.319218677816865e-06,
+      "loss": 0.8516,
+      "step": 3277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.313416779329309e-06,
+      "loss": 0.8594,
+      "step": 3278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.307616316652998e-06,
+      "loss": 0.7769,
+      "step": 3279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.301817292047039e-06,
+      "loss": 0.8027,
+      "step": 3280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2960197077699715e-06,
+      "loss": 0.4663,
+      "step": 3281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290223566079779e-06,
+      "loss": 0.8076,
+      "step": 3282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2844288692338784e-06,
+      "loss": 0.8481,
+      "step": 3283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.278635619489126e-06,
+      "loss": 0.8257,
+      "step": 3284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2728438191018194e-06,
+      "loss": 0.7754,
+      "step": 3285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267053470327686e-06,
+      "loss": 0.8096,
+      "step": 3286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2612645754218855e-06,
+      "loss": 0.7612,
+      "step": 3287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.255477136639023e-06,
+      "loss": 0.791,
+      "step": 3288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.24969115623312e-06,
+      "loss": 0.9375,
+      "step": 3289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243906636457643e-06,
+      "loss": 0.8403,
+      "step": 3290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238123579565485e-06,
+      "loss": 0.8589,
+      "step": 3291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.232341987808968e-06,
+      "loss": 0.7866,
+      "step": 3292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2265618634398465e-06,
+      "loss": 0.7705,
+      "step": 3293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.220783208709304e-06,
+      "loss": 0.8311,
+      "step": 3294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.215006025867945e-06,
+      "loss": 0.7427,
+      "step": 3295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.209230317165809e-06,
+      "loss": 0.8145,
+      "step": 3296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.203456084852354e-06,
+      "loss": 0.7402,
+      "step": 3297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.197683331176467e-06,
+      "loss": 0.8091,
+      "step": 3298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1919120583864625e-06,
+      "loss": 0.8018,
+      "step": 3299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186142268730074e-06,
+      "loss": 0.8149,
+      "step": 3300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1803739644544545e-06,
+      "loss": 0.8647,
+      "step": 3301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.174607147806184e-06,
+      "loss": 0.8594,
+      "step": 3302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.168841821031257e-06,
+      "loss": 0.8042,
+      "step": 3303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.163077986375097e-06,
+      "loss": 0.7559,
+      "step": 3304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.157315646082541e-06,
+      "loss": 0.832,
+      "step": 3305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.151554802397844e-06,
+      "loss": 0.8076,
+      "step": 3306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.145795457564672e-06,
+      "loss": 0.771,
+      "step": 3307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1400376138261185e-06,
+      "loss": 0.8218,
+      "step": 3308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.134281273424687e-06,
+      "loss": 0.8345,
+      "step": 3309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.128526438602296e-06,
+      "loss": 0.8208,
+      "step": 3310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.122773111600275e-06,
+      "loss": 0.8242,
+      "step": 3311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.117021294659376e-06,
+      "loss": 0.8379,
+      "step": 3312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.111270990019745e-06,
+      "loss": 0.7852,
+      "step": 3313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105522199920954e-06,
+      "loss": 0.8374,
+      "step": 3314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.099774926601982e-06,
+      "loss": 0.8418,
+      "step": 3315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.094029172301213e-06,
+      "loss": 0.4819,
+      "step": 3316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.088284939256448e-06,
+      "loss": 0.8032,
+      "step": 3317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.082542229704886e-06,
+      "loss": 0.8569,
+      "step": 3318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.076801045883136e-06,
+      "loss": 0.8159,
+      "step": 3319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.071061390027213e-06,
+      "loss": 0.8311,
+      "step": 3320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.065323264372539e-06,
+      "loss": 0.8188,
+      "step": 3321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059586671153934e-06,
+      "loss": 0.8516,
+      "step": 3322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.053851612605633e-06,
+      "loss": 0.8052,
+      "step": 3323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.04811809096126e-06,
+      "loss": 0.8408,
+      "step": 3324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.042386108453847e-06,
+      "loss": 0.499,
+      "step": 3325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.036655667315826e-06,
+      "loss": 0.7939,
+      "step": 3326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0309267697790265e-06,
+      "loss": 0.7832,
+      "step": 3327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.025199418074681e-06,
+      "loss": 0.8047,
+      "step": 3328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.019473614433417e-06,
+      "loss": 0.7739,
+      "step": 3329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0137493610852635e-06,
+      "loss": 0.7842,
+      "step": 3330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.008026660259634e-06,
+      "loss": 0.8247,
+      "step": 3331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.002305514185347e-06,
+      "loss": 0.8555,
+      "step": 3332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.996585925090619e-06,
+      "loss": 0.8164,
+      "step": 3333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.990867895203052e-06,
+      "loss": 0.7852,
+      "step": 3334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.985151426749639e-06,
+      "loss": 0.8154,
+      "step": 3335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.979436521956781e-06,
+      "loss": 0.7866,
+      "step": 3336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.973723183050244e-06,
+      "loss": 0.8335,
+      "step": 3337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9680114122552065e-06,
+      "loss": 0.7837,
+      "step": 3338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.962301211796229e-06,
+      "loss": 0.8403,
+      "step": 3339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.956592583897254e-06,
+      "loss": 0.4702,
+      "step": 3340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.950885530781622e-06,
+      "loss": 0.8877,
+      "step": 3341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94518005467206e-06,
+      "loss": 0.4353,
+      "step": 3342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.939476157790666e-06,
+      "loss": 0.8569,
+      "step": 3343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9337738423589374e-06,
+      "loss": 0.7954,
+      "step": 3344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.928073110597752e-06,
+      "loss": 0.7778,
+      "step": 3345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922373964727368e-06,
+      "loss": 0.8604,
+      "step": 3346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.916676406967432e-06,
+      "loss": 0.8584,
+      "step": 3347
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.910980439536969e-06,
+      "loss": 0.8267,
+      "step": 3348
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.905286064654379e-06,
+      "loss": 0.8179,
+      "step": 3349
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.899593284537448e-06,
+      "loss": 0.8247,
+      "step": 3350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.893902101403337e-06,
+      "loss": 0.8755,
+      "step": 3351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.888212517468594e-06,
+      "loss": 0.834,
+      "step": 3352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8825245349491346e-06,
+      "loss": 0.8208,
+      "step": 3353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.876838156060256e-06,
+      "loss": 0.7705,
+      "step": 3354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.871153383016624e-06,
+      "loss": 0.7578,
+      "step": 3355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.865470218032283e-06,
+      "loss": 0.8354,
+      "step": 3356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.859788663320656e-06,
+      "loss": 0.8506,
+      "step": 3357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.854108721094533e-06,
+      "loss": 0.7598,
+      "step": 3358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.848430393566077e-06,
+      "loss": 0.8584,
+      "step": 3359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.842753682946829e-06,
+      "loss": 0.835,
+      "step": 3360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.83707859144768e-06,
+      "loss": 0.8765,
+      "step": 3361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.831405121278917e-06,
+      "loss": 0.8335,
+      "step": 3362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.825733274650176e-06,
+      "loss": 0.8745,
+      "step": 3363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.820063053770471e-06,
+      "loss": 0.8604,
+      "step": 3364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8143944608481786e-06,
+      "loss": 0.7339,
+      "step": 3365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.80872749809104e-06,
+      "loss": 0.8257,
+      "step": 3366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.803062167706166e-06,
+      "loss": 0.7773,
+      "step": 3367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.797398471900029e-06,
+      "loss": 0.8315,
+      "step": 3368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.791736412878463e-06,
+      "loss": 0.7832,
+      "step": 3369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.78607599284667e-06,
+      "loss": 0.8213,
+      "step": 3370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7804172140092044e-06,
+      "loss": 0.8472,
+      "step": 3371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.774760078569998e-06,
+      "loss": 0.4839,
+      "step": 3372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7691045887323215e-06,
+      "loss": 0.8218,
+      "step": 3373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.763450746698818e-06,
+      "loss": 0.8223,
+      "step": 3374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.757798554671487e-06,
+      "loss": 0.7983,
+      "step": 3375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.752148014851678e-06,
+      "loss": 0.8047,
+      "step": 3376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.746499129440114e-06,
+      "loss": 0.835,
+      "step": 3377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.740851900636861e-06,
+      "loss": 0.7629,
+      "step": 3378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.735206330641334e-06,
+      "loss": 0.8018,
+      "step": 3379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.729562421652314e-06,
+      "loss": 0.7773,
+      "step": 3380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.72392017586793e-06,
+      "loss": 0.8652,
+      "step": 3381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7182795954856605e-06,
+      "loss": 0.8013,
+      "step": 3382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.712640682702346e-06,
+      "loss": 0.7969,
+      "step": 3383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.707003439714172e-06,
+      "loss": 0.7949,
+      "step": 3384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.701367868716661e-06,
+      "loss": 0.8228,
+      "step": 3385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.695733971904702e-06,
+      "loss": 0.7861,
+      "step": 3386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.690101751472521e-06,
+      "loss": 0.8193,
+      "step": 3387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.684471209613702e-06,
+      "loss": 0.8071,
+      "step": 3388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.678842348521166e-06,
+      "loss": 0.7803,
+      "step": 3389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6732151703871855e-06,
+      "loss": 0.855,
+      "step": 3390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.667589677403366e-06,
+      "loss": 0.811,
+      "step": 3391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.661965871760664e-06,
+      "loss": 0.7944,
+      "step": 3392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.656343755649387e-06,
+      "loss": 0.4963,
+      "step": 3393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.650723331259176e-06,
+      "loss": 0.7344,
+      "step": 3394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.645104600779012e-06,
+      "loss": 0.8154,
+      "step": 3395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.639487566397217e-06,
+      "loss": 0.8418,
+      "step": 3396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.633872230301458e-06,
+      "loss": 0.8057,
+      "step": 3397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.628258594678733e-06,
+      "loss": 0.8418,
+      "step": 3398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622646661715381e-06,
+      "loss": 0.8574,
+      "step": 3399
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.61703643359708e-06,
+      "loss": 0.8022,
+      "step": 3400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611427912508841e-06,
+      "loss": 0.8018,
+      "step": 3401
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605821100635012e-06,
+      "loss": 0.7949,
+      "step": 3402
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.600216000159273e-06,
+      "loss": 0.8198,
+      "step": 3403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.594612613264641e-06,
+      "loss": 0.7998,
+      "step": 3404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.58901094213346e-06,
+      "loss": 0.8071,
+      "step": 3405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.583410988947413e-06,
+      "loss": 0.7993,
+      "step": 3406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.577812755887508e-06,
+      "loss": 0.8521,
+      "step": 3407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572216245134088e-06,
+      "loss": 0.8159,
+      "step": 3408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.566621458866819e-06,
+      "loss": 0.7896,
+      "step": 3409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5610283992647e-06,
+      "loss": 0.7959,
+      "step": 3410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55543706850606e-06,
+      "loss": 0.7778,
+      "step": 3411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.549847468768546e-06,
+      "loss": 0.8628,
+      "step": 3412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.544259602229139e-06,
+      "loss": 0.7783,
+      "step": 3413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.538673471064136e-06,
+      "loss": 0.8169,
+      "step": 3414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.533089077449176e-06,
+      "loss": 0.814,
+      "step": 3415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.527506423559199e-06,
+      "loss": 0.8096,
+      "step": 3416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.521925511568482e-06,
+      "loss": 0.7886,
+      "step": 3417
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.516346343650616e-06,
+      "loss": 0.4973,
+      "step": 3418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.510768921978515e-06,
+      "loss": 0.8188,
+      "step": 3419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.505193248724421e-06,
+      "loss": 0.8066,
+      "step": 3420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4996193260598885e-06,
+      "loss": 0.4456,
+      "step": 3421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.494047156155781e-06,
+      "loss": 0.8394,
+      "step": 3422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.488476741182291e-06,
+      "loss": 0.7764,
+      "step": 3423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4829080833089225e-06,
+      "loss": 0.8086,
+      "step": 3424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.477341184704503e-06,
+      "loss": 0.832,
+      "step": 3425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.471776047537169e-06,
+      "loss": 0.7458,
+      "step": 3426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.466212673974369e-06,
+      "loss": 0.8325,
+      "step": 3427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4606510661828645e-06,
+      "loss": 0.8013,
+      "step": 3428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4550912263287324e-06,
+      "loss": 0.7715,
+      "step": 3429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.449533156577356e-06,
+      "loss": 0.7959,
+      "step": 3430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.44397685909344e-06,
+      "loss": 0.8477,
+      "step": 3431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.438422336040991e-06,
+      "loss": 0.7344,
+      "step": 3432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432869589583329e-06,
+      "loss": 0.8623,
+      "step": 3433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4273186218830685e-06,
+      "loss": 0.7734,
+      "step": 3434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421769435102144e-06,
+      "loss": 0.7876,
+      "step": 3435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.416222031401801e-06,
+      "loss": 0.8296,
+      "step": 3436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410676412942578e-06,
+      "loss": 0.8628,
+      "step": 3437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.405132581884325e-06,
+      "loss": 0.8198,
+      "step": 3438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399590540386197e-06,
+      "loss": 0.8135,
+      "step": 3439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.39405029060664e-06,
+      "loss": 0.791,
+      "step": 3440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.388511834703421e-06,
+      "loss": 0.8394,
+      "step": 3441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3829751748335955e-06,
+      "loss": 0.8608,
+      "step": 3442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3774403131535234e-06,
+      "loss": 0.7881,
+      "step": 3443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.371907251818864e-06,
+      "loss": 0.8071,
+      "step": 3444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.366375992984575e-06,
+      "loss": 0.8213,
+      "step": 3445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360846538804911e-06,
+      "loss": 0.8135,
+      "step": 3446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.355318891433428e-06,
+      "loss": 0.8247,
+      "step": 3447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3497930530229725e-06,
+      "loss": 0.8262,
+      "step": 3448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.34426902572569e-06,
+      "loss": 0.4922,
+      "step": 3449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.338746811693021e-06,
+      "loss": 0.8516,
+      "step": 3450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3332264130756985e-06,
+      "loss": 0.8096,
+      "step": 3451
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.327707832023748e-06,
+      "loss": 0.7827,
+      "step": 3452
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.322191070686487e-06,
+      "loss": 0.7773,
+      "step": 3453
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.316676131212527e-06,
+      "loss": 0.8555,
+      "step": 3454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.311163015749768e-06,
+      "loss": 0.812,
+      "step": 3455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.305651726445396e-06,
+      "loss": 0.8271,
+      "step": 3456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.300142265445893e-06,
+      "loss": 0.5044,
+      "step": 3457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294634634897026e-06,
+      "loss": 0.813,
+      "step": 3458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.289128836943845e-06,
+      "loss": 0.8394,
+      "step": 3459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2836248737306926e-06,
+      "loss": 0.8213,
+      "step": 3460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.278122747401193e-06,
+      "loss": 0.7568,
+      "step": 3461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.27262246009825e-06,
+      "loss": 0.8223,
+      "step": 3462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.267124013964071e-06,
+      "loss": 0.7998,
+      "step": 3463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2616274111401214e-06,
+      "loss": 0.7549,
+      "step": 3464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.256132653767161e-06,
+      "loss": 0.8091,
+      "step": 3465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.250639743985231e-06,
+      "loss": 0.8262,
+      "step": 3466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.245148683933648e-06,
+      "loss": 0.7886,
+      "step": 3467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239659475751018e-06,
+      "loss": 0.8403,
+      "step": 3468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.234172121575221e-06,
+      "loss": 0.8555,
+      "step": 3469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2286866235434045e-06,
+      "loss": 0.8389,
+      "step": 3470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.223202983792006e-06,
+      "loss": 0.8345,
+      "step": 3471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2177212044567315e-06,
+      "loss": 0.7827,
+      "step": 3472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.212241287672575e-06,
+      "loss": 0.8169,
+      "step": 3473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.206763235573791e-06,
+      "loss": 0.4546,
+      "step": 3474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.201287050293917e-06,
+      "loss": 0.813,
+      "step": 3475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195812733965753e-06,
+      "loss": 0.7612,
+      "step": 3476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190340288721379e-06,
+      "loss": 0.8091,
+      "step": 3477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.184869716692143e-06,
+      "loss": 0.8247,
+      "step": 3478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.179401020008671e-06,
+      "loss": 0.7798,
+      "step": 3479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173934200800851e-06,
+      "loss": 0.7842,
+      "step": 3480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.168469261197845e-06,
+      "loss": 0.7739,
+      "step": 3481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.163006203328071e-06,
+      "loss": 0.8008,
+      "step": 3482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.157545029319224e-06,
+      "loss": 0.7974,
+      "step": 3483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.152085741298272e-06,
+      "loss": 0.8101,
+      "step": 3484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.146628341391433e-06,
+      "loss": 0.8013,
+      "step": 3485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1411728317242015e-06,
+      "loss": 0.8535,
+      "step": 3486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135719214421334e-06,
+      "loss": 0.8442,
+      "step": 3487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130267491606835e-06,
+      "loss": 0.7256,
+      "step": 3488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1248176654039945e-06,
+      "loss": 0.4561,
+      "step": 3489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.119369737935352e-06,
+      "loss": 0.7866,
+      "step": 3490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113923711322705e-06,
+      "loss": 0.8408,
+      "step": 3491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.108479587687115e-06,
+      "loss": 0.8462,
+      "step": 3492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.103037369148902e-06,
+      "loss": 0.8208,
+      "step": 3493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.097597057827644e-06,
+      "loss": 0.5083,
+      "step": 3494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092158655842175e-06,
+      "loss": 0.8477,
+      "step": 3495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0867221653105856e-06,
+      "loss": 0.8384,
+      "step": 3496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.081287588350223e-06,
+      "loss": 0.8169,
+      "step": 3497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.075854927077688e-06,
+      "loss": 0.8521,
+      "step": 3498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.070424183608835e-06,
+      "loss": 0.8394,
+      "step": 3499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0649953600587755e-06,
+      "loss": 0.79,
+      "step": 3500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.059568458541866e-06,
+      "loss": 0.7632,
+      "step": 3501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.054143481171719e-06,
+      "loss": 0.7803,
+      "step": 3502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0487204300612e-06,
+      "loss": 0.8623,
+      "step": 3503
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.043299307322418e-06,
+      "loss": 0.814,
+      "step": 3504
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0378801150667365e-06,
+      "loss": 0.8057,
+      "step": 3505
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.032462855404764e-06,
+      "loss": 0.7676,
+      "step": 3506
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.027047530446357e-06,
+      "loss": 0.7998,
+      "step": 3507
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0216341423006195e-06,
+      "loss": 0.7998,
+      "step": 3508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.016222693075899e-06,
+      "loss": 0.7617,
+      "step": 3509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.010813184879787e-06,
+      "loss": 0.8354,
+      "step": 3510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.005405619819131e-06,
+      "loss": 0.8633,
+      "step": 3511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.8286,
+      "step": 3512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.994596327527727e-06,
+      "loss": 0.7974,
+      "step": 3513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.98919460450687e-06,
+      "loss": 0.77,
+      "step": 3514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.983794833041233e-06,
+      "loss": 0.8003,
+      "step": 3515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.978397015233871e-06,
+      "loss": 0.8145,
+      "step": 3516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.973001153187063e-06,
+      "loss": 0.8081,
+      "step": 3517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.967607249002336e-06,
+      "loss": 0.8735,
+      "step": 3518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.962215304780444e-06,
+      "loss": 0.7949,
+      "step": 3519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956825322621386e-06,
+      "loss": 0.8057,
+      "step": 3520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.951437304624392e-06,
+      "loss": 0.7998,
+      "step": 3521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.946051252887938e-06,
+      "loss": 0.8179,
+      "step": 3522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940667169509723e-06,
+      "loss": 0.8525,
+      "step": 3523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935285056586685e-06,
+      "loss": 0.7856,
+      "step": 3524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.929904916214986e-06,
+      "loss": 0.7905,
+      "step": 3525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.924526750490024e-06,
+      "loss": 0.8286,
+      "step": 3526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9191505615064384e-06,
+      "loss": 0.8516,
+      "step": 3527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.913776351358087e-06,
+      "loss": 0.8047,
+      "step": 3528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9084041221380585e-06,
+      "loss": 0.7749,
+      "step": 3529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903033875938677e-06,
+      "loss": 0.8169,
+      "step": 3530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.897665614851477e-06,
+      "loss": 0.7817,
+      "step": 3531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8922993409672425e-06,
+      "loss": 0.812,
+      "step": 3532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.88693505637597e-06,
+      "loss": 0.8237,
+      "step": 3533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881572763166883e-06,
+      "loss": 0.8301,
+      "step": 3534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.876212463428431e-06,
+      "loss": 0.7207,
+      "step": 3535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.870854159248287e-06,
+      "loss": 0.833,
+      "step": 3536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.865497852713347e-06,
+      "loss": 0.8091,
+      "step": 3537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.860143545909725e-06,
+      "loss": 0.7925,
+      "step": 3538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854791240922764e-06,
+      "loss": 0.4734,
+      "step": 3539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.84944093983702e-06,
+      "loss": 0.8447,
+      "step": 3540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8440926447362714e-06,
+      "loss": 0.8149,
+      "step": 3541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.838746357703516e-06,
+      "loss": 0.8184,
+      "step": 3542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.833402080820969e-06,
+      "loss": 0.7778,
+      "step": 3543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.828059816170063e-06,
+      "loss": 0.8135,
+      "step": 3544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.822719565831445e-06,
+      "loss": 0.7661,
+      "step": 3545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.817381331884981e-06,
+      "loss": 0.7798,
+      "step": 3546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.812045116409747e-06,
+      "loss": 0.8657,
+      "step": 3547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.806710921484037e-06,
+      "loss": 0.8257,
+      "step": 3548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801378749185357e-06,
+      "loss": 0.7915,
+      "step": 3549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.796048601590424e-06,
+      "loss": 0.8301,
+      "step": 3550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.790720480775168e-06,
+      "loss": 0.7783,
+      "step": 3551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.785394388814726e-06,
+      "loss": 0.8379,
+      "step": 3552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7800703277834485e-06,
+      "loss": 0.7954,
+      "step": 3553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774748299754902e-06,
+      "loss": 0.4319,
+      "step": 3554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.769428306801844e-06,
+      "loss": 0.7549,
+      "step": 3555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764110350996252e-06,
+      "loss": 0.7578,
+      "step": 3556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.758794434409306e-06,
+      "loss": 0.4746,
+      "step": 3557
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7534805591113905e-06,
+      "loss": 0.7891,
+      "step": 3558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.748168727172103e-06,
+      "loss": 0.8745,
+      "step": 3559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742858940660241e-06,
+      "loss": 0.7676,
+      "step": 3560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7375512016437955e-06,
+      "loss": 0.7915,
+      "step": 3561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732245512189971e-06,
+      "loss": 0.7524,
+      "step": 3562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.72694187436517e-06,
+      "loss": 0.7554,
+      "step": 3563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.721640290235001e-06,
+      "loss": 0.813,
+      "step": 3564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7163407618642675e-06,
+      "loss": 0.4968,
+      "step": 3565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7110432913169766e-06,
+      "loss": 0.8047,
+      "step": 3566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705747880656323e-06,
+      "loss": 0.7935,
+      "step": 3567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700454531944711e-06,
+      "loss": 0.7998,
+      "step": 3568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695163247243733e-06,
+      "loss": 0.7246,
+      "step": 3569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6898740286141895e-06,
+      "loss": 0.9004,
+      "step": 3570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684586878116069e-06,
+      "loss": 0.8823,
+      "step": 3571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.679301797808554e-06,
+      "loss": 0.8164,
+      "step": 3572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674018789750018e-06,
+      "loss": 0.8364,
+      "step": 3573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.668737855998026e-06,
+      "loss": 0.7866,
+      "step": 3574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663458998609353e-06,
+      "loss": 0.7939,
+      "step": 3575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658182219639944e-06,
+      "loss": 0.8179,
+      "step": 3576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.652907521144945e-06,
+      "loss": 0.7881,
+      "step": 3577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.647634905178692e-06,
+      "loss": 0.7681,
+      "step": 3578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.642364373794697e-06,
+      "loss": 0.8164,
+      "step": 3579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637095929045682e-06,
+      "loss": 0.8394,
+      "step": 3580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.63182957298354e-06,
+      "loss": 0.8008,
+      "step": 3581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.626565307659358e-06,
+      "loss": 0.8267,
+      "step": 3582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.621303135123403e-06,
+      "loss": 0.8096,
+      "step": 3583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.616043057425131e-06,
+      "loss": 0.4683,
+      "step": 3584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610785076613182e-06,
+      "loss": 0.8203,
+      "step": 3585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.605529194735379e-06,
+      "loss": 0.5156,
+      "step": 3586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600275413838724e-06,
+      "loss": 0.8164,
+      "step": 3587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595023735969406e-06,
+      "loss": 0.7959,
+      "step": 3588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.589774163172793e-06,
+      "loss": 0.7939,
+      "step": 3589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.584526697493432e-06,
+      "loss": 0.8418,
+      "step": 3590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57928134097505e-06,
+      "loss": 0.7432,
+      "step": 3591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.574038095660553e-06,
+      "loss": 0.8291,
+      "step": 3592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568796963592025e-06,
+      "loss": 0.8237,
+      "step": 3593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.563557946810725e-06,
+      "loss": 0.8081,
+      "step": 3594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.558321047357092e-06,
+      "loss": 0.834,
+      "step": 3595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5530862672707345e-06,
+      "loss": 0.8071,
+      "step": 3596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.547853608590442e-06,
+      "loss": 0.7979,
+      "step": 3597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.542623073354173e-06,
+      "loss": 0.7822,
+      "step": 3598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.537394663599062e-06,
+      "loss": 0.8076,
+      "step": 3599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532168381361414e-06,
+      "loss": 0.7769,
+      "step": 3600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.526944228676702e-06,
+      "loss": 0.7847,
+      "step": 3601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.521722207579584e-06,
+      "loss": 0.8159,
+      "step": 3602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.516502320103867e-06,
+      "loss": 0.4917,
+      "step": 3603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.51128456828254e-06,
+      "loss": 0.7803,
+      "step": 3604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.50606895414776e-06,
+      "loss": 0.8091,
+      "step": 3605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.500855479730842e-06,
+      "loss": 0.8066,
+      "step": 3606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.495644147062285e-06,
+      "loss": 0.7734,
+      "step": 3607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.490434958171741e-06,
+      "loss": 0.8154,
+      "step": 3608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485227915088023e-06,
+      "loss": 0.7598,
+      "step": 3609
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.480023019839119e-06,
+      "loss": 0.4658,
+      "step": 3610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.474820274452172e-06,
+      "loss": 0.752,
+      "step": 3611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4696196809535e-06,
+      "loss": 0.8013,
+      "step": 3612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.46442124136857e-06,
+      "loss": 0.8535,
+      "step": 3613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.45922495772202e-06,
+      "loss": 0.7822,
+      "step": 3614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454030832037635e-06,
+      "loss": 0.8696,
+      "step": 3615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.448838866338372e-06,
+      "loss": 0.792,
+      "step": 3616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.443649062646339e-06,
+      "loss": 0.7871,
+      "step": 3617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.438461422982812e-06,
+      "loss": 0.8354,
+      "step": 3618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4332759493682165e-06,
+      "loss": 0.8267,
+      "step": 3619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428092643822136e-06,
+      "loss": 0.7485,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4229115083633046e-06,
+      "loss": 0.751,
+      "step": 3621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4177325450096145e-06,
+      "loss": 0.8804,
+      "step": 3622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4125557557781205e-06,
+      "loss": 0.814,
+      "step": 3623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.407381142685021e-06,
+      "loss": 0.8452,
+      "step": 3624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402208707745669e-06,
+      "loss": 0.833,
+      "step": 3625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.397038452974571e-06,
+      "loss": 0.7778,
+      "step": 3626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3918703803853745e-06,
+      "loss": 0.8228,
+      "step": 3627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3867044919908966e-06,
+      "loss": 0.7515,
+      "step": 3628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.381540789803086e-06,
+      "loss": 0.7808,
+      "step": 3629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.376379275833051e-06,
+      "loss": 0.8477,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371219952091038e-06,
+      "loss": 0.7842,
+      "step": 3631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.366062820586449e-06,
+      "loss": 0.8257,
+      "step": 3632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.360907883327829e-06,
+      "loss": 0.7725,
+      "step": 3633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355755142322866e-06,
+      "loss": 0.7188,
+      "step": 3634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.350604599578396e-06,
+      "loss": 0.814,
+      "step": 3635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.345456257100398e-06,
+      "loss": 0.8413,
+      "step": 3636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.340310116893993e-06,
+      "loss": 0.8228,
+      "step": 3637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335166180963445e-06,
+      "loss": 0.8081,
+      "step": 3638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.33002445131216e-06,
+      "loss": 0.7983,
+      "step": 3639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324884929942684e-06,
+      "loss": 0.8647,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3197476188567044e-06,
+      "loss": 0.7974,
+      "step": 3641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3146125200550434e-06,
+      "loss": 0.467,
+      "step": 3642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.309479635537668e-06,
+      "loss": 0.7739,
+      "step": 3643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304348967303673e-06,
+      "loss": 0.7983,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29922051735131e-06,
+      "loss": 0.7776,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29409428767794e-06,
+      "loss": 0.8584,
+      "step": 3646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.288970280280078e-06,
+      "loss": 0.8423,
+      "step": 3647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.283848497153367e-06,
+      "loss": 0.7964,
+      "step": 3648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.278728940292581e-06,
+      "loss": 0.8081,
+      "step": 3649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.273611611691638e-06,
+      "loss": 0.8628,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268496513343582e-06,
+      "loss": 0.761,
+      "step": 3651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.263383647240576e-06,
+      "loss": 0.7686,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.258273015373934e-06,
+      "loss": 0.8433,
+      "step": 3653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253164619734082e-06,
+      "loss": 0.8179,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.248058462310594e-06,
+      "loss": 0.8081,
+      "step": 3655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.242954545092155e-06,
+      "loss": 0.8506,
+      "step": 3656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237852870066592e-06,
+      "loss": 0.7822,
+      "step": 3657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2327534392208415e-06,
+      "loss": 0.8062,
+      "step": 3658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.227656254540977e-06,
+      "loss": 0.8403,
+      "step": 3659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.222561318012201e-06,
+      "loss": 0.7856,
+      "step": 3660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.217468631618834e-06,
+      "loss": 0.8208,
+      "step": 3661
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.21237819734432e-06,
+      "loss": 0.8438,
+      "step": 3662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.20729001717123e-06,
+      "loss": 0.7646,
+      "step": 3663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2022040930812505e-06,
+      "loss": 0.7661,
+      "step": 3664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.197120427055189e-06,
+      "loss": 0.7993,
+      "step": 3665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192039021072987e-06,
+      "loss": 0.8071,
+      "step": 3666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.186959877113693e-06,
+      "loss": 0.4697,
+      "step": 3667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.181882997155478e-06,
+      "loss": 0.8608,
+      "step": 3668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176808383175633e-06,
+      "loss": 0.7549,
+      "step": 3669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.171736037150556e-06,
+      "loss": 0.8101,
+      "step": 3670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.166665961055781e-06,
+      "loss": 0.7593,
+      "step": 3671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.161598156865944e-06,
+      "loss": 0.8315,
+      "step": 3672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156532626554799e-06,
+      "loss": 0.7856,
+      "step": 3673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1514693720952135e-06,
+      "loss": 0.7583,
+      "step": 3674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.146408395459171e-06,
+      "loss": 0.7554,
+      "step": 3675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.141349698617769e-06,
+      "loss": 0.7842,
+      "step": 3676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1362932835412125e-06,
+      "loss": 0.832,
+      "step": 3677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.131239152198821e-06,
+      "loss": 0.7339,
+      "step": 3678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126187306559023e-06,
+      "loss": 0.7695,
+      "step": 3679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12113774858936e-06,
+      "loss": 0.7817,
+      "step": 3680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.116090480256477e-06,
+      "loss": 0.8042,
+      "step": 3681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.111045503526133e-06,
+      "loss": 0.8394,
+      "step": 3682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.106002820363191e-06,
+      "loss": 0.8774,
+      "step": 3683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.100962432731621e-06,
+      "loss": 0.8228,
+      "step": 3684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.095924342594501e-06,
+      "loss": 0.7437,
+      "step": 3685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.090888551914012e-06,
+      "loss": 0.7783,
+      "step": 3686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0858550626514394e-06,
+      "loss": 0.8486,
+      "step": 3687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080823876767174e-06,
+      "loss": 0.7935,
+      "step": 3688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.075794996220709e-06,
+      "loss": 0.8066,
+      "step": 3689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.070768422970639e-06,
+      "loss": 0.8745,
+      "step": 3690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065744158974659e-06,
+      "loss": 0.4609,
+      "step": 3691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0607222061895645e-06,
+      "loss": 0.8057,
+      "step": 3692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.055702566571262e-06,
+      "loss": 0.8066,
+      "step": 3693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.050685242074738e-06,
+      "loss": 0.7803,
+      "step": 3694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.045670234654087e-06,
+      "loss": 0.4973,
+      "step": 3695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.040657546262504e-06,
+      "loss": 0.7744,
+      "step": 3696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0356471788522725e-06,
+      "loss": 0.8521,
+      "step": 3697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.030639134374784e-06,
+      "loss": 0.8511,
+      "step": 3698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.02563341478052e-06,
+      "loss": 0.8438,
+      "step": 3699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.020630022019046e-06,
+      "loss": 0.7856,
+      "step": 3700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.015628958039035e-06,
+      "loss": 0.4568,
+      "step": 3701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.010630224788244e-06,
+      "loss": 0.8433,
+      "step": 3702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0056338242135354e-06,
+      "loss": 0.8071,
+      "step": 3703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.000639758260847e-06,
+      "loss": 0.4282,
+      "step": 3704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.995648028875222e-06,
+      "loss": 0.7754,
+      "step": 3705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990658638000776e-06,
+      "loss": 0.4019,
+      "step": 3706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985671587580726e-06,
+      "loss": 0.7549,
+      "step": 3707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.980686879557382e-06,
+      "loss": 0.7773,
+      "step": 3708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.97570451587213e-06,
+      "loss": 0.7886,
+      "step": 3709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.97072449846545e-06,
+      "loss": 0.7671,
+      "step": 3710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.96574682927691e-06,
+      "loss": 0.7837,
+      "step": 3711
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.96077151024515e-06,
+      "loss": 0.8486,
+      "step": 3712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9557985433079064e-06,
+      "loss": 0.7734,
+      "step": 3713
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.950827930402005e-06,
+      "loss": 0.8306,
+      "step": 3714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945859673463343e-06,
+      "loss": 0.7627,
+      "step": 3715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940893774426902e-06,
+      "loss": 0.7773,
+      "step": 3716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935930235226756e-06,
+      "loss": 0.7393,
+      "step": 3717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9309690577960365e-06,
+      "loss": 0.814,
+      "step": 3718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.926010244066983e-06,
+      "loss": 0.8135,
+      "step": 3719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.921053795970899e-06,
+      "loss": 0.7236,
+      "step": 3720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916099715438166e-06,
+      "loss": 0.8115,
+      "step": 3721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.91114800439825e-06,
+      "loss": 0.8384,
+      "step": 3722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.906198664779688e-06,
+      "loss": 0.7705,
+      "step": 3723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.901251698510097e-06,
+      "loss": 0.8379,
+      "step": 3724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.89630710751617e-06,
+      "loss": 0.7871,
+      "step": 3725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.891364893723673e-06,
+      "loss": 0.8008,
+      "step": 3726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.886425059057446e-06,
+      "loss": 0.7759,
+      "step": 3727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.881487605441404e-06,
+      "loss": 0.8062,
+      "step": 3728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.876552534798535e-06,
+      "loss": 0.853,
+      "step": 3729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.871619849050896e-06,
+      "loss": 0.8057,
+      "step": 3730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866689550119619e-06,
+      "loss": 0.752,
+      "step": 3731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8617616399249035e-06,
+      "loss": 0.7969,
+      "step": 3732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.856836120386018e-06,
+      "loss": 0.8213,
+      "step": 3733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851912993421304e-06,
+      "loss": 0.7935,
+      "step": 3734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846992260948168e-06,
+      "loss": 0.8003,
+      "step": 3735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8420739248830825e-06,
+      "loss": 0.8164,
+      "step": 3736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.837157987141592e-06,
+      "loss": 0.7969,
+      "step": 3737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832244449638303e-06,
+      "loss": 0.7515,
+      "step": 3738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.827333314286885e-06,
+      "loss": 0.833,
+      "step": 3739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.822424583000074e-06,
+      "loss": 0.8125,
+      "step": 3740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81751825768968e-06,
+      "loss": 0.7637,
+      "step": 3741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.812614340266557e-06,
+      "loss": 0.8428,
+      "step": 3742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807712832640632e-06,
+      "loss": 0.8281,
+      "step": 3743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802813736720895e-06,
+      "loss": 0.772,
+      "step": 3744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7979170544153874e-06,
+      "loss": 0.833,
+      "step": 3745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7930227876312264e-06,
+      "loss": 0.8506,
+      "step": 3746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7881309382745767e-06,
+      "loss": 0.8003,
+      "step": 3747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.783241508250659e-06,
+      "loss": 0.7578,
+      "step": 3748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.778354499463759e-06,
+      "loss": 0.5039,
+      "step": 3749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7734699138172127e-06,
+      "loss": 0.7642,
+      "step": 3750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7685877532134254e-06,
+      "loss": 0.791,
+      "step": 3751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7637080195538435e-06,
+      "loss": 0.4695,
+      "step": 3752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.758830714738976e-06,
+      "loss": 0.79,
+      "step": 3753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.753955840668385e-06,
+      "loss": 0.834,
+      "step": 3754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.749083399240675e-06,
+      "loss": 0.8418,
+      "step": 3755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7442133923535253e-06,
+      "loss": 0.7583,
+      "step": 3756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.739345821903647e-06,
+      "loss": 0.7393,
+      "step": 3757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7344806897868137e-06,
+      "loss": 0.8203,
+      "step": 3758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.729617997897842e-06,
+      "loss": 0.8433,
+      "step": 3759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7247577481306065e-06,
+      "loss": 0.8193,
+      "step": 3760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7198999423780157e-06,
+      "loss": 0.8257,
+      "step": 3761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.715044582532045e-06,
+      "loss": 0.8247,
+      "step": 3762
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.710191670483706e-06,
+      "loss": 0.8521,
+      "step": 3763
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.70534120812306e-06,
+      "loss": 0.7905,
+      "step": 3764
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7004931973392123e-06,
+      "loss": 0.8135,
+      "step": 3765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6956476400203147e-06,
+      "loss": 0.7876,
+      "step": 3766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.690804538053563e-06,
+      "loss": 0.8223,
+      "step": 3767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6859638933251985e-06,
+      "loss": 0.8711,
+      "step": 3768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.681125707720502e-06,
+      "loss": 0.8413,
+      "step": 3769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6762899831238007e-06,
+      "loss": 0.8477,
+      "step": 3770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6714567214184605e-06,
+      "loss": 0.8242,
+      "step": 3771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6666259244868883e-06,
+      "loss": 0.8213,
+      "step": 3772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6617975942105324e-06,
+      "loss": 0.813,
+      "step": 3773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6569717324698782e-06,
+      "loss": 0.7905,
+      "step": 3774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6521483411444514e-06,
+      "loss": 0.4563,
+      "step": 3775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6473274221128163e-06,
+      "loss": 0.7881,
+      "step": 3776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6425089772525735e-06,
+      "loss": 0.8179,
+      "step": 3777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.637693008440357e-06,
+      "loss": 0.8037,
+      "step": 3778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6328795175518417e-06,
+      "loss": 0.7617,
+      "step": 3779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.628068506461735e-06,
+      "loss": 0.8184,
+      "step": 3780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.623259977043776e-06,
+      "loss": 0.8608,
+      "step": 3781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.618453931170741e-06,
+      "loss": 0.8022,
+      "step": 3782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6136503707144354e-06,
+      "loss": 0.8286,
+      "step": 3783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.608849297545707e-06,
+      "loss": 0.8115,
+      "step": 3784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.604050713534417e-06,
+      "loss": 0.7725,
+      "step": 3785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5992546205494725e-06,
+      "loss": 0.7798,
+      "step": 3786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5944610204588014e-06,
+      "loss": 0.8208,
+      "step": 3787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5896699151293635e-06,
+      "loss": 0.8008,
+      "step": 3788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5848813064271527e-06,
+      "loss": 0.8237,
+      "step": 3789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.580095196217187e-06,
+      "loss": 0.7783,
+      "step": 3790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5753115863635014e-06,
+      "loss": 0.772,
+      "step": 3791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5705304787291685e-06,
+      "loss": 0.812,
+      "step": 3792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.565751875176282e-06,
+      "loss": 0.7881,
+      "step": 3793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.560975777565966e-06,
+      "loss": 0.7827,
+      "step": 3794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.556202187758363e-06,
+      "loss": 0.7896,
+      "step": 3795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.551431107612642e-06,
+      "loss": 0.8013,
+      "step": 3796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5466625389869858e-06,
+      "loss": 0.4956,
+      "step": 3797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.541896483738607e-06,
+      "loss": 0.7812,
+      "step": 3798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5371329437237432e-06,
+      "loss": 0.8018,
+      "step": 3799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.532371920797646e-06,
+      "loss": 0.7944,
+      "step": 3800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.527613416814586e-06,
+      "loss": 0.8057,
+      "step": 3801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5228574336278596e-06,
+      "loss": 0.7866,
+      "step": 3802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5181039730897657e-06,
+      "loss": 0.7422,
+      "step": 3803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.513353037051641e-06,
+      "loss": 0.75,
+      "step": 3804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5086046273638287e-06,
+      "loss": 0.8315,
+      "step": 3805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5038587458756857e-06,
+      "loss": 0.7451,
+      "step": 3806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4991153944355903e-06,
+      "loss": 0.49,
+      "step": 3807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.494374574890934e-06,
+      "loss": 0.8413,
+      "step": 3808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.489636289088112e-06,
+      "loss": 0.7842,
+      "step": 3809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.484900538872551e-06,
+      "loss": 0.853,
+      "step": 3810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.480167326088677e-06,
+      "loss": 0.4314,
+      "step": 3811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4754366525799323e-06,
+      "loss": 0.8296,
+      "step": 3812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.470708520188768e-06,
+      "loss": 0.7231,
+      "step": 3813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4659829307566482e-06,
+      "loss": 0.7134,
+      "step": 3814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4612598861240443e-06,
+      "loss": 0.8174,
+      "step": 3815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4565393881304376e-06,
+      "loss": 0.7891,
+      "step": 3816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.451821438614319e-06,
+      "loss": 0.7461,
+      "step": 3817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4471060394131827e-06,
+      "loss": 0.8018,
+      "step": 3818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4423931923635344e-06,
+      "loss": 0.7964,
+      "step": 3819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4376828993008827e-06,
+      "loss": 0.8384,
+      "step": 3820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.432975162059743e-06,
+      "loss": 0.7842,
+      "step": 3821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.428269982473634e-06,
+      "loss": 0.7866,
+      "step": 3822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.42356736237508e-06,
+      "loss": 0.7034,
+      "step": 3823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4188673035956065e-06,
+      "loss": 0.8008,
+      "step": 3824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.414169807965744e-06,
+      "loss": 0.8062,
+      "step": 3825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.409474877315021e-06,
+      "loss": 0.793,
+      "step": 3826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.40478251347197e-06,
+      "loss": 0.8901,
+      "step": 3827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4000927182641243e-06,
+      "loss": 0.7949,
+      "step": 3828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3954054935180136e-06,
+      "loss": 0.8159,
+      "step": 3829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3907208410591696e-06,
+      "loss": 0.8252,
+      "step": 3830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.386038762712117e-06,
+      "loss": 0.4717,
+      "step": 3831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.381359260300393e-06,
+      "loss": 0.8247,
+      "step": 3832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3766823356465095e-06,
+      "loss": 0.8076,
+      "step": 3833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3720079905719883e-06,
+      "loss": 0.8208,
+      "step": 3834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.367336226897344e-06,
+      "loss": 0.7822,
+      "step": 3835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.362667046442083e-06,
+      "loss": 0.8096,
+      "step": 3836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3580004510247134e-06,
+      "loss": 0.8384,
+      "step": 3837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3533364424627324e-06,
+      "loss": 0.833,
+      "step": 3838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3486750225726206e-06,
+      "loss": 0.8306,
+      "step": 3839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.344016193169862e-06,
+      "loss": 0.8003,
+      "step": 3840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.339359956068926e-06,
+      "loss": 0.8291,
+      "step": 3841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.334706313083279e-06,
+      "loss": 0.835,
+      "step": 3842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3300552660253706e-06,
+      "loss": 0.4468,
+      "step": 3843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3254068167066433e-06,
+      "loss": 0.7915,
+      "step": 3844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3207609669375195e-06,
+      "loss": 0.8052,
+      "step": 3845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.316117718527416e-06,
+      "loss": 0.7983,
+      "step": 3846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3114770732847413e-06,
+      "loss": 0.8403,
+      "step": 3847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3068390330168833e-06,
+      "loss": 0.8311,
+      "step": 3848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3022035995302138e-06,
+      "loss": 0.7593,
+      "step": 3849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2975707746300977e-06,
+      "loss": 0.8086,
+      "step": 3850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2929405601208674e-06,
+      "loss": 0.8081,
+      "step": 3851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.28831295780586e-06,
+      "loss": 0.8032,
+      "step": 3852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2836879694873814e-06,
+      "loss": 0.8154,
+      "step": 3853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2790655969667227e-06,
+      "loss": 0.8184,
+      "step": 3854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.274445842044157e-06,
+      "loss": 0.8501,
+      "step": 3855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2698287065189404e-06,
+      "loss": 0.7676,
+      "step": 3856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2652141921892957e-06,
+      "loss": 0.8127,
+      "step": 3857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2606023008524447e-06,
+      "loss": 0.7886,
+      "step": 3858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.255993034304574e-06,
+      "loss": 0.8145,
+      "step": 3859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.251386394340853e-06,
+      "loss": 0.8223,
+      "step": 3860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.246782382755427e-06,
+      "loss": 0.8027,
+      "step": 3861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2421810013414158e-06,
+      "loss": 0.7817,
+      "step": 3862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.237582251890917e-06,
+      "loss": 0.7729,
+      "step": 3863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2329861361950042e-06,
+      "loss": 0.7271,
+      "step": 3864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.228392656043723e-06,
+      "loss": 0.7495,
+      "step": 3865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.223801813226092e-06,
+      "loss": 0.8071,
+      "step": 3866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2192136095301043e-06,
+      "loss": 0.7617,
+      "step": 3867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.214628046742726e-06,
+      "loss": 0.7354,
+      "step": 3868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.210045126649892e-06,
+      "loss": 0.8359,
+      "step": 3869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2054648510365094e-06,
+      "loss": 0.8184,
+      "step": 3870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.200887221686456e-06,
+      "loss": 0.7979,
+      "step": 3871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1963122403825787e-06,
+      "loss": 0.8091,
+      "step": 3872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.19173990890669e-06,
+      "loss": 0.8442,
+      "step": 3873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1871702290395734e-06,
+      "loss": 0.7339,
+      "step": 3874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1826032025609855e-06,
+      "loss": 0.8271,
+      "step": 3875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1780388312496357e-06,
+      "loss": 0.8022,
+      "step": 3876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1734771168832103e-06,
+      "loss": 0.8257,
+      "step": 3877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.168918061238355e-06,
+      "loss": 0.8198,
+      "step": 3878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1643616660906806e-06,
+      "loss": 0.8188,
+      "step": 3879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1598079332147704e-06,
+      "loss": 0.8042,
+      "step": 3880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.155256864384163e-06,
+      "loss": 0.8481,
+      "step": 3881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1507084613713548e-06,
+      "loss": 0.769,
+      "step": 3882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.146162725947811e-06,
+      "loss": 0.7837,
+      "step": 3883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1416196598839555e-06,
+      "loss": 0.7998,
+      "step": 3884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1370792649491777e-06,
+      "loss": 0.813,
+      "step": 3885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.13254154291182e-06,
+      "loss": 0.8154,
+      "step": 3886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1280064955391885e-06,
+      "loss": 0.7515,
+      "step": 3887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1234741245975406e-06,
+      "loss": 0.8125,
+      "step": 3888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1189444318520946e-06,
+      "loss": 0.7993,
+      "step": 3889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.114417419067034e-06,
+      "loss": 0.8423,
+      "step": 3890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1098930880054888e-06,
+      "loss": 0.8125,
+      "step": 3891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1053714404295453e-06,
+      "loss": 0.8276,
+      "step": 3892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1008524781002536e-06,
+      "loss": 0.7515,
+      "step": 3893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.096336202777598e-06,
+      "loss": 0.7905,
+      "step": 3894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.09182261622054e-06,
+      "loss": 0.7563,
+      "step": 3895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.087311720186982e-06,
+      "loss": 0.793,
+      "step": 3896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0828035164337766e-06,
+      "loss": 0.771,
+      "step": 3897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0782980067167347e-06,
+      "loss": 0.8164,
+      "step": 3898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.073795192790614e-06,
+      "loss": 0.8179,
+      "step": 3899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0692950764091156e-06,
+      "loss": 0.7715,
+      "step": 3900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0647976593249053e-06,
+      "loss": 0.8066,
+      "step": 3901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.060302943289587e-06,
+      "loss": 0.8462,
+      "step": 3902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0558109300537143e-06,
+      "loss": 0.8071,
+      "step": 3903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0513216213667884e-06,
+      "loss": 0.7593,
+      "step": 3904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.046835018977258e-06,
+      "loss": 0.8252,
+      "step": 3905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0423511246325176e-06,
+      "loss": 0.4536,
+      "step": 3906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0378699400789057e-06,
+      "loss": 0.8223,
+      "step": 3907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.033391467061708e-06,
+      "loss": 0.856,
+      "step": 3908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0289157073251516e-06,
+      "loss": 0.8101,
+      "step": 3909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.024442662612408e-06,
+      "loss": 0.8115,
+      "step": 3910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0199723346655906e-06,
+      "loss": 0.4556,
+      "step": 3911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.015504725225756e-06,
+      "loss": 0.7783,
+      "step": 3912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0110398360329006e-06,
+      "loss": 0.752,
+      "step": 3913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.006577668825963e-06,
+      "loss": 0.769,
+      "step": 3914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.002118225342819e-06,
+      "loss": 0.812,
+      "step": 3915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9976615073202843e-06,
+      "loss": 0.8335,
+      "step": 3916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.993207516494121e-06,
+      "loss": 0.7788,
+      "step": 3917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9887562545990145e-06,
+      "loss": 0.811,
+      "step": 3918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9843077233685967e-06,
+      "loss": 0.8345,
+      "step": 3919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9798619245354354e-06,
+      "loss": 0.7993,
+      "step": 3920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.975418859831034e-06,
+      "loss": 0.7935,
+      "step": 3921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9709785309858242e-06,
+      "loss": 0.79,
+      "step": 3922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.96654093972919e-06,
+      "loss": 0.8101,
+      "step": 3923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.962106087789426e-06,
+      "loss": 0.7769,
+      "step": 3924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9576739768937757e-06,
+      "loss": 0.8198,
+      "step": 3925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.953244608768411e-06,
+      "loss": 0.834,
+      "step": 3926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9488179851384293e-06,
+      "loss": 0.8198,
+      "step": 3927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9443941077278737e-06,
+      "loss": 0.8208,
+      "step": 3928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9399729782597086e-06,
+      "loss": 0.4744,
+      "step": 3929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.935554598455822e-06,
+      "loss": 0.8208,
+      "step": 3930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.93113897003704e-06,
+      "loss": 0.7905,
+      "step": 3931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9267260947231115e-06,
+      "loss": 0.854,
+      "step": 3932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9223159742327232e-06,
+      "loss": 0.7617,
+      "step": 3933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9179086102834788e-06,
+      "loss": 0.8232,
+      "step": 3934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.913504004591915e-06,
+      "loss": 0.8164,
+      "step": 3935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9091021588734826e-06,
+      "loss": 0.8022,
+      "step": 3936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9047030748425685e-06,
+      "loss": 0.8218,
+      "step": 3937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.900306754212485e-06,
+      "loss": 0.7905,
+      "step": 3938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8959131986954635e-06,
+      "loss": 0.8125,
+      "step": 3939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8915224100026575e-06,
+      "loss": 0.7939,
+      "step": 3940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8871343898441484e-06,
+      "loss": 0.8228,
+      "step": 3941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8827491399289266e-06,
+      "loss": 0.8623,
+      "step": 3942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.878366661964921e-06,
+      "loss": 0.7568,
+      "step": 3943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.873986957658971e-06,
+      "loss": 0.8369,
+      "step": 3944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.869610028716835e-06,
+      "loss": 0.7866,
+      "step": 3945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8652358768431953e-06,
+      "loss": 0.8345,
+      "step": 3946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8608645037416516e-06,
+      "loss": 0.7183,
+      "step": 3947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8564959111147117e-06,
+      "loss": 0.7642,
+      "step": 3948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8521301006638156e-06,
+      "loss": 0.5112,
+      "step": 3949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8477670740893127e-06,
+      "loss": 0.7412,
+      "step": 3950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8434068330904664e-06,
+      "loss": 0.8047,
+      "step": 3951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8390493793654574e-06,
+      "loss": 0.772,
+      "step": 3952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.834694714611379e-06,
+      "loss": 0.7754,
+      "step": 3953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.830342840524243e-06,
+      "loss": 0.7993,
+      "step": 3954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.825993758798968e-06,
+      "loss": 0.8252,
+      "step": 3955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.821647471129391e-06,
+      "loss": 0.7236,
+      "step": 3956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8173039792082547e-06,
+      "loss": 0.8076,
+      "step": 3957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8129632847272182e-06,
+      "loss": 0.7832,
+      "step": 3958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8086253893768477e-06,
+      "loss": 0.8076,
+      "step": 3959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8042902948466223e-06,
+      "loss": 0.7817,
+      "step": 3960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7999580028249273e-06,
+      "loss": 0.7529,
+      "step": 3961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.795628514999059e-06,
+      "loss": 0.7637,
+      "step": 3962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.791301833055219e-06,
+      "loss": 0.8311,
+      "step": 3963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.786977958678515e-06,
+      "loss": 0.8237,
+      "step": 3964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7826568935529718e-06,
+      "loss": 0.7827,
+      "step": 3965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.778338639361504e-06,
+      "loss": 0.7461,
+      "step": 3966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.774023197785942e-06,
+      "loss": 0.4829,
+      "step": 3967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.769710570507018e-06,
+      "loss": 0.8403,
+      "step": 3968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7654007592043676e-06,
+      "loss": 0.6887,
+      "step": 3969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7610937655565284e-06,
+      "loss": 0.7559,
+      "step": 3970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7567895912409513e-06,
+      "loss": 0.7891,
+      "step": 3971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.752488237933971e-06,
+      "loss": 0.8062,
+      "step": 3972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7481897073108364e-06,
+      "loss": 0.7671,
+      "step": 3973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7438940010456926e-06,
+      "loss": 0.8594,
+      "step": 3974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7396011208115836e-06,
+      "loss": 0.8174,
+      "step": 3975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7353110682804596e-06,
+      "loss": 0.4604,
+      "step": 3976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7310238451231664e-06,
+      "loss": 0.7178,
+      "step": 3977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.726739453009438e-06,
+      "loss": 0.7861,
+      "step": 3978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7224578936079183e-06,
+      "loss": 0.7573,
+      "step": 3979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7181791685861403e-06,
+      "loss": 0.856,
+      "step": 3980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7139032796105415e-06,
+      "loss": 0.4031,
+      "step": 3981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.709630228346448e-06,
+      "loss": 0.7656,
+      "step": 3982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.705360016458085e-06,
+      "loss": 0.7622,
+      "step": 3983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.701092645608563e-06,
+      "loss": 0.8203,
+      "step": 3984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.696828117459892e-06,
+      "loss": 0.772,
+      "step": 3985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6925664336729816e-06,
+      "loss": 0.8022,
+      "step": 3986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6883075959076245e-06,
+      "loss": 0.7622,
+      "step": 3987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.684051605822506e-06,
+      "loss": 0.8188,
+      "step": 3988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6797984650752085e-06,
+      "loss": 0.8853,
+      "step": 3989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.67554817532219e-06,
+      "loss": 0.4146,
+      "step": 3990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6713007382188183e-06,
+      "loss": 0.7681,
+      "step": 3991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6670561554193352e-06,
+      "loss": 0.8262,
+      "step": 3992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6628144285768777e-06,
+      "loss": 0.8174,
+      "step": 3993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.658575559343467e-06,
+      "loss": 0.7598,
+      "step": 3994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.654339549370013e-06,
+      "loss": 0.7686,
+      "step": 3995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6501064003063136e-06,
+      "loss": 0.8433,
+      "step": 3996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.645876113801048e-06,
+      "loss": 0.7734,
+      "step": 3997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6416486915017857e-06,
+      "loss": 0.7725,
+      "step": 3998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6374241350549766e-06,
+      "loss": 0.8188,
+      "step": 3999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.633202446105957e-06,
+      "loss": 0.4727,
+      "step": 4000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6289836262989444e-06,
+      "loss": 0.832,
+      "step": 4001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.62476767727704e-06,
+      "loss": 0.8413,
+      "step": 4002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.620554600682228e-06,
+      "loss": 0.7256,
+      "step": 4003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6163443981553727e-06,
+      "loss": 0.8584,
+      "step": 4004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.612137071336217e-06,
+      "loss": 0.853,
+      "step": 4005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.607932621863387e-06,
+      "loss": 0.8081,
+      "step": 4006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.603731051374385e-06,
+      "loss": 0.811,
+      "step": 4007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5995323615056023e-06,
+      "loss": 0.8262,
+      "step": 4008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5953365538922902e-06,
+      "loss": 0.8335,
+      "step": 4009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.591143630168591e-06,
+      "loss": 0.5076,
+      "step": 4010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5869535919675203e-06,
+      "loss": 0.7666,
+      "step": 4011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5827664409209686e-06,
+      "loss": 0.7974,
+      "step": 4012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.578582178659701e-06,
+      "loss": 0.8145,
+      "step": 4013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5744008068133685e-06,
+      "loss": 0.813,
+      "step": 4014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5702223270104776e-06,
+      "loss": 0.8213,
+      "step": 4015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5660467408784217e-06,
+      "loss": 0.7729,
+      "step": 4016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5618740500434647e-06,
+      "loss": 0.4409,
+      "step": 4017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5577042561307364e-06,
+      "loss": 0.7671,
+      "step": 4018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.553537360764252e-06,
+      "loss": 0.7949,
+      "step": 4019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5493733655668886e-06,
+      "loss": 0.8257,
+      "step": 4020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.545212272160389e-06,
+      "loss": 0.8096,
+      "step": 4021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.541054082165374e-06,
+      "loss": 0.7651,
+      "step": 4022
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.536898797201329e-06,
+      "loss": 0.8276,
+      "step": 4023
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.532746418886617e-06,
+      "loss": 0.8164,
+      "step": 4024
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5285969488384578e-06,
+      "loss": 0.8013,
+      "step": 4025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5244503886729467e-06,
+      "loss": 0.8408,
+      "step": 4026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5203067400050353e-06,
+      "loss": 0.8491,
+      "step": 4027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.516166004448548e-06,
+      "loss": 0.4365,
+      "step": 4028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.512028183616181e-06,
+      "loss": 0.7847,
+      "step": 4029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.507893279119484e-06,
+      "loss": 0.79,
+      "step": 4030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.503761292568878e-06,
+      "loss": 0.7607,
+      "step": 4031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.499632225573647e-06,
+      "loss": 0.4749,
+      "step": 4032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4955060797419274e-06,
+      "loss": 0.8413,
+      "step": 4033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4913828566807363e-06,
+      "loss": 0.4353,
+      "step": 4034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4872625579959374e-06,
+      "loss": 0.771,
+      "step": 4035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.483145185292264e-06,
+      "loss": 0.7734,
+      "step": 4036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.479030740173304e-06,
+      "loss": 0.8154,
+      "step": 4037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.47491922424151e-06,
+      "loss": 0.8027,
+      "step": 4038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4708106390981888e-06,
+      "loss": 0.7861,
+      "step": 4039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.466704986343511e-06,
+      "loss": 0.7861,
+      "step": 4040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4626022675765014e-06,
+      "loss": 0.8135,
+      "step": 4041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4585024843950443e-06,
+      "loss": 0.8335,
+      "step": 4042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4544056383958783e-06,
+      "loss": 0.8086,
+      "step": 4043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4503117311746006e-06,
+      "loss": 0.7104,
+      "step": 4044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.446220764325662e-06,
+      "loss": 0.7563,
+      "step": 4045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4421327394423678e-06,
+      "loss": 0.5002,
+      "step": 4046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.438047658116881e-06,
+      "loss": 0.7803,
+      "step": 4047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4339655219402134e-06,
+      "loss": 0.7759,
+      "step": 4048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4298863325022327e-06,
+      "loss": 0.8062,
+      "step": 4049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.425810091391659e-06,
+      "loss": 0.7803,
+      "step": 4050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.421736800196063e-06,
+      "loss": 0.8315,
+      "step": 4051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4176664605018662e-06,
+      "loss": 0.7715,
+      "step": 4052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4135990738943414e-06,
+      "loss": 0.7395,
+      "step": 4053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4095346419576116e-06,
+      "loss": 0.7539,
+      "step": 4054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4054731662746457e-06,
+      "loss": 0.4419,
+      "step": 4055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4014146484272716e-06,
+      "loss": 0.8579,
+      "step": 4056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.397359089996151e-06,
+      "loss": 0.4487,
+      "step": 4057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.393306492560801e-06,
+      "loss": 0.8091,
+      "step": 4058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3892568576995834e-06,
+      "loss": 0.73,
+      "step": 4059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3852101869897092e-06,
+      "loss": 0.7383,
+      "step": 4060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.38116648200723e-06,
+      "loss": 0.7832,
+      "step": 4061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.377125744327051e-06,
+      "loss": 0.7964,
+      "step": 4062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3730879755229085e-06,
+      "loss": 0.7759,
+      "step": 4063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3690531771673942e-06,
+      "loss": 0.7646,
+      "step": 4064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3650213508319354e-06,
+      "loss": 0.8428,
+      "step": 4065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.360992498086805e-06,
+      "loss": 0.7656,
+      "step": 4066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.356966620501121e-06,
+      "loss": 0.7998,
+      "step": 4067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352943719642842e-06,
+      "loss": 0.8101,
+      "step": 4068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3489237970787557e-06,
+      "loss": 0.7866,
+      "step": 4069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3449068543745045e-06,
+      "loss": 0.7939,
+      "step": 4070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3408928930945605e-06,
+      "loss": 0.8521,
+      "step": 4071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.336881914802246e-06,
+      "loss": 0.7554,
+      "step": 4072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3328739210597085e-06,
+      "loss": 0.7661,
+      "step": 4073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.328868913427943e-06,
+      "loss": 0.7896,
+      "step": 4074
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.324866893466772e-06,
+      "loss": 0.7817,
+      "step": 4075
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.32086786273486e-06,
+      "loss": 0.4578,
+      "step": 4076
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3168718227897115e-06,
+      "loss": 0.7935,
+      "step": 4077
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.312878775187659e-06,
+      "loss": 0.7988,
+      "step": 4078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3088887214838727e-06,
+      "loss": 0.8433,
+      "step": 4079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3049016632323597e-06,
+      "loss": 0.7803,
+      "step": 4080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.300917601985948e-06,
+      "loss": 0.7817,
+      "step": 4081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.296936539296315e-06,
+      "loss": 0.7949,
+      "step": 4082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.292958476713961e-06,
+      "loss": 0.8032,
+      "step": 4083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2889834157882197e-06,
+      "loss": 0.7983,
+      "step": 4084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.285011358067255e-06,
+      "loss": 0.8174,
+      "step": 4085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2810423050980613e-06,
+      "loss": 0.8354,
+      "step": 4086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.277076258426464e-06,
+      "loss": 0.8066,
+      "step": 4087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.273113219597116e-06,
+      "loss": 0.8018,
+      "step": 4088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2691531901535015e-06,
+      "loss": 0.7856,
+      "step": 4089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2651961716379277e-06,
+      "loss": 0.8296,
+      "step": 4090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.261242165591534e-06,
+      "loss": 0.7822,
+      "step": 4091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2572911735542844e-06,
+      "loss": 0.8418,
+      "step": 4092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2533431970649687e-06,
+      "loss": 0.4429,
+      "step": 4093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2493982376612035e-06,
+      "loss": 0.7842,
+      "step": 4094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.24545629687943e-06,
+      "loss": 0.791,
+      "step": 4095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2415173762549124e-06,
+      "loss": 0.7979,
+      "step": 4096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2375814773217395e-06,
+      "loss": 0.4468,
+      "step": 4097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2336486016128246e-06,
+      "loss": 0.4783,
+      "step": 4098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.229718750659902e-06,
+      "loss": 0.8145,
+      "step": 4099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.225791925993528e-06,
+      "loss": 0.7671,
+      "step": 4100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.221868129143079e-06,
+      "loss": 0.7881,
+      "step": 4101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2179473616367563e-06,
+      "loss": 0.8506,
+      "step": 4102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2140296250015747e-06,
+      "loss": 0.7485,
+      "step": 4103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.210114920763381e-06,
+      "loss": 0.7769,
+      "step": 4104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2062032504468246e-06,
+      "loss": 0.8398,
+      "step": 4105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2022946155753833e-06,
+      "loss": 0.8286,
+      "step": 4106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.198389017671352e-06,
+      "loss": 0.8301,
+      "step": 4107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.19448645825584e-06,
+      "loss": 0.8042,
+      "step": 4108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.190586938848772e-06,
+      "loss": 0.7915,
+      "step": 4109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.186690460968901e-06,
+      "loss": 0.7676,
+      "step": 4110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1827970261337763e-06,
+      "loss": 0.855,
+      "step": 4111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1789066358597753e-06,
+      "loss": 0.8359,
+      "step": 4112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.175019291662085e-06,
+      "loss": 0.8237,
+      "step": 4113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1711349950547033e-06,
+      "loss": 0.7993,
+      "step": 4114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.167253747550452e-06,
+      "loss": 0.8237,
+      "step": 4115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1633755506609554e-06,
+      "loss": 0.8042,
+      "step": 4116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.159500405896654e-06,
+      "loss": 0.8438,
+      "step": 4117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1556283147667924e-06,
+      "loss": 0.7876,
+      "step": 4118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1517592787794327e-06,
+      "loss": 0.8442,
+      "step": 4119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1478932994414504e-06,
+      "loss": 0.8154,
+      "step": 4120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1440303782585238e-06,
+      "loss": 0.7949,
+      "step": 4121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1401705167351416e-06,
+      "loss": 0.8071,
+      "step": 4122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1363137163746062e-06,
+      "loss": 0.4338,
+      "step": 4123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1324599786790123e-06,
+      "loss": 0.7549,
+      "step": 4124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.128609305149283e-06,
+      "loss": 0.834,
+      "step": 4125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.124761697285135e-06,
+      "loss": 0.8115,
+      "step": 4126
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.120917156585093e-06,
+      "loss": 0.7925,
+      "step": 4127
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.117075684546487e-06,
+      "loss": 0.8242,
+      "step": 4128
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.113237282665455e-06,
+      "loss": 0.7969,
+      "step": 4129
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1094019524369334e-06,
+      "loss": 0.7915,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.10556969535467e-06,
+      "loss": 0.7686,
+      "step": 4131
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.101740512911209e-06,
+      "loss": 0.7188,
+      "step": 4132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0979144065978996e-06,
+      "loss": 0.8252,
+      "step": 4133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0940913779048946e-06,
+      "loss": 0.791,
+      "step": 4134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0902714283211456e-06,
+      "loss": 0.7466,
+      "step": 4135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0864545593344055e-06,
+      "loss": 0.8096,
+      "step": 4136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.082640772431228e-06,
+      "loss": 0.7471,
+      "step": 4137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0788300690969675e-06,
+      "loss": 0.8604,
+      "step": 4138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0750224508157757e-06,
+      "loss": 0.8545,
+      "step": 4139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0712179190706015e-06,
+      "loss": 0.832,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.067416475343197e-06,
+      "loss": 0.7998,
+      "step": 4141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0636181211141047e-06,
+      "loss": 0.7861,
+      "step": 4142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0598228578626687e-06,
+      "loss": 0.8521,
+      "step": 4143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0560306870670278e-06,
+      "loss": 0.8115,
+      "step": 4144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0522416102041167e-06,
+      "loss": 0.8013,
+      "step": 4145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0484556287496614e-06,
+      "loss": 0.7983,
+      "step": 4146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.044672744178193e-06,
+      "loss": 0.7988,
+      "step": 4147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0408929579630244e-06,
+      "loss": 0.7554,
+      "step": 4148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.037116271576265e-06,
+      "loss": 0.7983,
+      "step": 4149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0333426864888216e-06,
+      "loss": 0.7729,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0295722041703857e-06,
+      "loss": 0.7515,
+      "step": 4151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.02580482608945e-06,
+      "loss": 0.7744,
+      "step": 4152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.022040553713295e-06,
+      "loss": 0.7754,
+      "step": 4153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.018279388507981e-06,
+      "loss": 0.8042,
+      "step": 4154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.014521331938372e-06,
+      "loss": 0.8486,
+      "step": 4155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.010766385468114e-06,
+      "loss": 0.8096,
+      "step": 4156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0070145505596415e-06,
+      "loss": 0.8521,
+      "step": 4157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0032658286741858e-06,
+      "loss": 0.8613,
+      "step": 4158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999520221271757e-06,
+      "loss": 0.8408,
+      "step": 4159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9957777298111514e-06,
+      "loss": 0.7881,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9920383557499567e-06,
+      "loss": 0.8149,
+      "step": 4161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9883021005445415e-06,
+      "loss": 0.7773,
+      "step": 4162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9845689656500677e-06,
+      "loss": 0.7822,
+      "step": 4163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9808389525204763e-06,
+      "loss": 0.7983,
+      "step": 4164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977112062608493e-06,
+      "loss": 0.7065,
+      "step": 4165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733882973656237e-06,
+      "loss": 0.7832,
+      "step": 4166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9696676582421614e-06,
+      "loss": 0.4688,
+      "step": 4167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9659501466871857e-06,
+      "loss": 0.7988,
+      "step": 4168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9622357641485503e-06,
+      "loss": 0.8096,
+      "step": 4169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9585245120728935e-06,
+      "loss": 0.8608,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9548163919056372e-06,
+      "loss": 0.7617,
+      "step": 4171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951111405090973e-06,
+      "loss": 0.8262,
+      "step": 4172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9474095530718885e-06,
+      "loss": 0.8027,
+      "step": 4173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.943710837290137e-06,
+      "loss": 0.4658,
+      "step": 4174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9400152591862562e-06,
+      "loss": 0.8115,
+      "step": 4175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.936322820199561e-06,
+      "loss": 0.8018,
+      "step": 4176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9326335217681423e-06,
+      "loss": 0.4919,
+      "step": 4177
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9289473653288695e-06,
+      "loss": 0.8267,
+      "step": 4178
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9252643523173874e-06,
+      "loss": 0.8179,
+      "step": 4179
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9215844841681166e-06,
+      "loss": 0.6963,
+      "step": 4180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.917907762314253e-06,
+      "loss": 0.8003,
+      "step": 4181
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914234188187768e-06,
+      "loss": 0.8369,
+      "step": 4182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910563763219405e-06,
+      "loss": 0.7119,
+      "step": 4183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9068964888386832e-06,
+      "loss": 0.8506,
+      "step": 4184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9032323664738928e-06,
+      "loss": 0.7637,
+      "step": 4185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8995713975520968e-06,
+      "loss": 0.7959,
+      "step": 4186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8959135834991328e-06,
+      "loss": 0.7603,
+      "step": 4187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.892258925739604e-06,
+      "loss": 0.7734,
+      "step": 4188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8886074256968912e-06,
+      "loss": 0.8262,
+      "step": 4189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8849590847931388e-06,
+      "loss": 0.7754,
+      "step": 4190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8813139044492656e-06,
+      "loss": 0.8174,
+      "step": 4191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8776718860849563e-06,
+      "loss": 0.8262,
+      "step": 4192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8740330311186672e-06,
+      "loss": 0.7739,
+      "step": 4193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8703973409676168e-06,
+      "loss": 0.8267,
+      "step": 4194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8667648170478036e-06,
+      "loss": 0.7705,
+      "step": 4195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.863135460773977e-06,
+      "loss": 0.7964,
+      "step": 4196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.859509273559662e-06,
+      "loss": 0.8135,
+      "step": 4197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8558862568171477e-06,
+      "loss": 0.834,
+      "step": 4198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.852266411957485e-06,
+      "loss": 0.8008,
+      "step": 4199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8486497403904978e-06,
+      "loss": 0.73,
+      "step": 4200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8450362435247704e-06,
+      "loss": 0.8115,
+      "step": 4201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8414259227676423e-06,
+      "loss": 0.8027,
+      "step": 4202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8378187795252268e-06,
+      "loss": 0.7886,
+      "step": 4203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8342148152023943e-06,
+      "loss": 0.8262,
+      "step": 4204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8306140312027766e-06,
+      "loss": 0.8252,
+      "step": 4205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8270164289287739e-06,
+      "loss": 0.8262,
+      "step": 4206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8234220097815426e-06,
+      "loss": 0.7954,
+      "step": 4207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8198307751609933e-06,
+      "loss": 0.7588,
+      "step": 4208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8162427264658044e-06,
+      "loss": 0.7397,
+      "step": 4209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.812657865093408e-06,
+      "loss": 0.7959,
+      "step": 4210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8090761924400035e-06,
+      "loss": 0.8203,
+      "step": 4211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.80549770990054e-06,
+      "loss": 0.7983,
+      "step": 4212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8019224188687302e-06,
+      "loss": 0.7905,
+      "step": 4213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7983503207370334e-06,
+      "loss": 0.8496,
+      "step": 4214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7947814168966738e-06,
+      "loss": 0.7788,
+      "step": 4215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7912157087376347e-06,
+      "loss": 0.7646,
+      "step": 4216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.787653197648649e-06,
+      "loss": 0.4673,
+      "step": 4217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.784093885017204e-06,
+      "loss": 0.8447,
+      "step": 4218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7805377722295459e-06,
+      "loss": 0.8047,
+      "step": 4219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7769848606706642e-06,
+      "loss": 0.7588,
+      "step": 4220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7734351517243165e-06,
+      "loss": 0.7886,
+      "step": 4221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7698886467730036e-06,
+      "loss": 0.8003,
+      "step": 4222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7663453471979819e-06,
+      "loss": 0.8203,
+      "step": 4223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.762805254379255e-06,
+      "loss": 0.8271,
+      "step": 4224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7592683696955826e-06,
+      "loss": 0.7808,
+      "step": 4225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7557346945244725e-06,
+      "loss": 0.8345,
+      "step": 4226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7522042302421837e-06,
+      "loss": 0.7305,
+      "step": 4227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7486769782237224e-06,
+      "loss": 0.8125,
+      "step": 4228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7451529398428468e-06,
+      "loss": 0.8262,
+      "step": 4229
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7416321164720606e-06,
+      "loss": 0.7959,
+      "step": 4230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7381145094826169e-06,
+      "loss": 0.7778,
+      "step": 4231
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7346001202445163e-06,
+      "loss": 0.8477,
+      "step": 4232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.731088950126506e-06,
+      "loss": 0.4485,
+      "step": 4233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7275810004960791e-06,
+      "loss": 0.7935,
+      "step": 4234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7240762727194739e-06,
+      "loss": 0.7627,
+      "step": 4235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7205747681616758e-06,
+      "loss": 0.7988,
+      "step": 4236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7170764881864088e-06,
+      "loss": 0.8281,
+      "step": 4237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7135814341561552e-06,
+      "loss": 0.7793,
+      "step": 4238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7100896074321238e-06,
+      "loss": 0.8213,
+      "step": 4239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7066010093742768e-06,
+      "loss": 0.8003,
+      "step": 4240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7031156413413152e-06,
+      "loss": 0.7764,
+      "step": 4241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6996335046906808e-06,
+      "loss": 0.7773,
+      "step": 4242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6961546007785656e-06,
+      "loss": 0.7632,
+      "step": 4243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6926789309598945e-06,
+      "loss": 0.8018,
+      "step": 4244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6892064965883304e-06,
+      "loss": 0.8145,
+      "step": 4245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6857372990162824e-06,
+      "loss": 0.8101,
+      "step": 4246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6822713395948942e-06,
+      "loss": 0.77,
+      "step": 4247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.678808619674055e-06,
+      "loss": 0.8003,
+      "step": 4248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.675349140602387e-06,
+      "loss": 0.7876,
+      "step": 4249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6718929037272523e-06,
+      "loss": 0.8672,
+      "step": 4250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6684399103947456e-06,
+      "loss": 0.8613,
+      "step": 4251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6649901619497056e-06,
+      "loss": 0.8174,
+      "step": 4252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6615436597356983e-06,
+      "loss": 0.4475,
+      "step": 4253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6581004050950366e-06,
+      "loss": 0.8511,
+      "step": 4254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6546603993687626e-06,
+      "loss": 0.7896,
+      "step": 4255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6512236438966522e-06,
+      "loss": 0.8545,
+      "step": 4256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6477901400172136e-06,
+      "loss": 0.791,
+      "step": 4257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6443598890676914e-06,
+      "loss": 0.7612,
+      "step": 4258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6409328923840673e-06,
+      "loss": 0.7866,
+      "step": 4259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.637509151301049e-06,
+      "loss": 0.8564,
+      "step": 4260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6340886671520795e-06,
+      "loss": 0.7388,
+      "step": 4261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6306714412693348e-06,
+      "loss": 0.8315,
+      "step": 4262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6272574749837123e-06,
+      "loss": 0.4548,
+      "step": 4263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6238467696248549e-06,
+      "loss": 0.7959,
+      "step": 4264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.620439326521126e-06,
+      "loss": 0.7861,
+      "step": 4265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6170351469996183e-06,
+      "loss": 0.8008,
+      "step": 4266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6136342323861576e-06,
+      "loss": 0.4639,
+      "step": 4267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6102365840052948e-06,
+      "loss": 0.8115,
+      "step": 4268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.60684220318031e-06,
+      "loss": 0.7979,
+      "step": 4269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.60345109123321e-06,
+      "loss": 0.437,
+      "step": 4270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6000632494847313e-06,
+      "loss": 0.77,
+      "step": 4271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.596678679254332e-06,
+      "loss": 0.7817,
+      "step": 4272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5932973818601994e-06,
+      "loss": 0.7979,
+      "step": 4273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5899193586192452e-06,
+      "loss": 0.7539,
+      "step": 4274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5865446108471073e-06,
+      "loss": 0.8027,
+      "step": 4275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5831731398581441e-06,
+      "loss": 0.4678,
+      "step": 4276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.579804946965442e-06,
+      "loss": 0.772,
+      "step": 4277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5764400334808095e-06,
+      "loss": 0.8218,
+      "step": 4278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5730784007147747e-06,
+      "loss": 0.4736,
+      "step": 4279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.569720049976594e-06,
+      "loss": 0.8047,
+      "step": 4280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.56636498257424e-06,
+      "loss": 0.8101,
+      "step": 4281
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5630131998144082e-06,
+      "loss": 0.7505,
+      "step": 4282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.559664703002517e-06,
+      "loss": 0.7998,
+      "step": 4283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5563194934427027e-06,
+      "loss": 0.7896,
+      "step": 4284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5529775724378194e-06,
+      "loss": 0.7725,
+      "step": 4285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5496389412894507e-06,
+      "loss": 0.7197,
+      "step": 4286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5463036012978828e-06,
+      "loss": 0.791,
+      "step": 4287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5429715537621326e-06,
+      "loss": 0.7856,
+      "step": 4288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.539642799979929e-06,
+      "loss": 0.8179,
+      "step": 4289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5363173412477172e-06,
+      "loss": 0.7793,
+      "step": 4290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.532995178860668e-06,
+      "loss": 0.8149,
+      "step": 4291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5296763141126602e-06,
+      "loss": 0.8032,
+      "step": 4292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.526360748296285e-06,
+      "loss": 0.8164,
+      "step": 4293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.523048482702858e-06,
+      "loss": 0.8052,
+      "step": 4294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5197395186224017e-06,
+      "loss": 0.7998,
+      "step": 4295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.516433857343661e-06,
+      "loss": 0.8066,
+      "step": 4296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.513131500154087e-06,
+      "loss": 0.8091,
+      "step": 4297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5098324483398496e-06,
+      "loss": 0.8042,
+      "step": 4298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5065367031858235e-06,
+      "loss": 0.8447,
+      "step": 4299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.503244265975603e-06,
+      "loss": 0.8105,
+      "step": 4300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.499955137991489e-06,
+      "loss": 0.7939,
+      "step": 4301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.496669320514501e-06,
+      "loss": 0.8091,
+      "step": 4302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4933868148243613e-06,
+      "loss": 0.8164,
+      "step": 4303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4901076221995091e-06,
+      "loss": 0.7578,
+      "step": 4304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4868317439170832e-06,
+      "loss": 0.8594,
+      "step": 4305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4835591812529392e-06,
+      "loss": 0.8081,
+      "step": 4306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4802899354816435e-06,
+      "loss": 0.8042,
+      "step": 4307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4770240078764652e-06,
+      "loss": 0.7695,
+      "step": 4308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4737613997093836e-06,
+      "loss": 0.4395,
+      "step": 4309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4705021122510866e-06,
+      "loss": 0.7695,
+      "step": 4310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4672461467709599e-06,
+      "loss": 0.8052,
+      "step": 4311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.463993504537109e-06,
+      "loss": 0.7881,
+      "step": 4312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4607441868163363e-06,
+      "loss": 0.814,
+      "step": 4313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4574981948741518e-06,
+      "loss": 0.8262,
+      "step": 4314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4542555299747707e-06,
+      "loss": 0.4075,
+      "step": 4315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4510161933811095e-06,
+      "loss": 0.6968,
+      "step": 4316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4477801863547925e-06,
+      "loss": 0.8657,
+      "step": 4317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4445475101561434e-06,
+      "loss": 0.8521,
+      "step": 4318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4413181660441932e-06,
+      "loss": 0.7656,
+      "step": 4319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4380921552766703e-06,
+      "loss": 0.8208,
+      "step": 4320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4348694791100081e-06,
+      "loss": 0.8096,
+      "step": 4321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.431650138799341e-06,
+      "loss": 0.7764,
+      "step": 4322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.428434135598501e-06,
+      "loss": 0.7979,
+      "step": 4323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4252214707600254e-06,
+      "loss": 0.7529,
+      "step": 4324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4220121455351465e-06,
+      "loss": 0.8022,
+      "step": 4325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4188061611738002e-06,
+      "loss": 0.7427,
+      "step": 4326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4156035189246187e-06,
+      "loss": 0.7651,
+      "step": 4327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.412404220034932e-06,
+      "loss": 0.4592,
+      "step": 4328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.40920826575077e-06,
+      "loss": 0.8447,
+      "step": 4329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4060156573168593e-06,
+      "loss": 0.7759,
+      "step": 4330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4028263959766219e-06,
+      "loss": 0.7773,
+      "step": 4331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.399640482972179e-06,
+      "loss": 0.8135,
+      "step": 4332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3964579195443417e-06,
+      "loss": 0.7778,
+      "step": 4333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3932787069326291e-06,
+      "loss": 0.7993,
+      "step": 4334
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3901028463752409e-06,
+      "loss": 0.7974,
+      "step": 4335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3869303391090793e-06,
+      "loss": 0.8364,
+      "step": 4336
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3837611863697386e-06,
+      "loss": 0.7832,
+      "step": 4337
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3805953893915036e-06,
+      "loss": 0.75,
+      "step": 4338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3774329494073614e-06,
+      "loss": 0.7759,
+      "step": 4339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3742738676489842e-06,
+      "loss": 0.7715,
+      "step": 4340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3711181453467337e-06,
+      "loss": 0.8662,
+      "step": 4341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3679657837296688e-06,
+      "loss": 0.7866,
+      "step": 4342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3648167840255367e-06,
+      "loss": 0.8169,
+      "step": 4343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.361671147460779e-06,
+      "loss": 0.8022,
+      "step": 4344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.358528875260524e-06,
+      "loss": 0.832,
+      "step": 4345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3553899686485928e-06,
+      "loss": 0.5024,
+      "step": 4346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.352254428847488e-06,
+      "loss": 0.7793,
+      "step": 4347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.349122257078409e-06,
+      "loss": 0.791,
+      "step": 4348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3459934545612386e-06,
+      "loss": 0.8252,
+      "step": 4349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.342868022514554e-06,
+      "loss": 0.4565,
+      "step": 4350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.7964,
+      "step": 4351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3366272747003627e-06,
+      "loss": 0.4514,
+      "step": 4352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3335119613634384e-06,
+      "loss": 0.7871,
+      "step": 4353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3304000233581516e-06,
+      "loss": 0.7949,
+      "step": 4354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.327291461896515e-06,
+      "loss": 0.7998,
+      "step": 4355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3241862781892157e-06,
+      "loss": 0.7656,
+      "step": 4356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3210844734456262e-06,
+      "loss": 0.769,
+      "step": 4357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.317986048873805e-06,
+      "loss": 0.7671,
+      "step": 4358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3148910056804952e-06,
+      "loss": 0.79,
+      "step": 4359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3117993450711198e-06,
+      "loss": 0.8374,
+      "step": 4360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3087110682497862e-06,
+      "loss": 0.8398,
+      "step": 4361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3056261764192836e-06,
+      "loss": 0.854,
+      "step": 4362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3025446707810818e-06,
+      "loss": 0.7275,
+      "step": 4363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2994665525353356e-06,
+      "loss": 0.751,
+      "step": 4364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2963918228808746e-06,
+      "loss": 0.7095,
+      "step": 4365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2933204830152147e-06,
+      "loss": 0.7656,
+      "step": 4366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2902525341345473e-06,
+      "loss": 0.8022,
+      "step": 4367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2871879774337436e-06,
+      "loss": 0.8188,
+      "step": 4368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2841268141063568e-06,
+      "loss": 0.7437,
+      "step": 4369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.281069045344614e-06,
+      "loss": 0.811,
+      "step": 4370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2780146723394238e-06,
+      "loss": 0.8052,
+      "step": 4371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2749636962803714e-06,
+      "loss": 0.8135,
+      "step": 4372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.271916118355717e-06,
+      "loss": 0.7437,
+      "step": 4373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2688719397523997e-06,
+      "loss": 0.7539,
+      "step": 4374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2658311616560337e-06,
+      "loss": 0.8174,
+      "step": 4375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2627937852509064e-06,
+      "loss": 0.8374,
+      "step": 4376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2597598117199905e-06,
+      "loss": 0.7881,
+      "step": 4377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2567292422449185e-06,
+      "loss": 0.7983,
+      "step": 4378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2537020780060072e-06,
+      "loss": 0.7859,
+      "step": 4379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2506783201822426e-06,
+      "loss": 0.7871,
+      "step": 4380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2476579699512847e-06,
+      "loss": 0.7988,
+      "step": 4381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2446410284894727e-06,
+      "loss": 0.8198,
+      "step": 4382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2416274969718112e-06,
+      "loss": 0.8242,
+      "step": 4383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2386173765719755e-06,
+      "loss": 0.792,
+      "step": 4384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.235610668462316e-06,
+      "loss": 0.8179,
+      "step": 4385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2326073738138534e-06,
+      "loss": 0.8125,
+      "step": 4386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2296074937962809e-06,
+      "loss": 0.7979,
+      "step": 4387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2266110295779587e-06,
+      "loss": 0.8354,
+      "step": 4388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.223617982325921e-06,
+      "loss": 0.7505,
+      "step": 4389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2206283532058617e-06,
+      "loss": 0.7803,
+      "step": 4390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2176421433821538e-06,
+      "loss": 0.7686,
+      "step": 4391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2146593540178319e-06,
+      "loss": 0.7954,
+      "step": 4392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2116799862746043e-06,
+      "loss": 0.7524,
+      "step": 4393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2087040413128426e-06,
+      "loss": 0.7866,
+      "step": 4394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2057315202915876e-06,
+      "loss": 0.811,
+      "step": 4395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2027624243685421e-06,
+      "loss": 0.793,
+      "step": 4396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1997967547000766e-06,
+      "loss": 0.7905,
+      "step": 4397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1968345124412328e-06,
+      "loss": 0.814,
+      "step": 4398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1938756987457122e-06,
+      "loss": 0.4834,
+      "step": 4399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.190920314765882e-06,
+      "loss": 0.7871,
+      "step": 4400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1879683616527759e-06,
+      "loss": 0.6924,
+      "step": 4401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.185019840556082e-06,
+      "loss": 0.8286,
+      "step": 4402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.182074752624166e-06,
+      "loss": 0.8022,
+      "step": 4403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1791330990040484e-06,
+      "loss": 0.7334,
+      "step": 4404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1761948808414103e-06,
+      "loss": 0.8086,
+      "step": 4405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1732600992805998e-06,
+      "loss": 0.8462,
+      "step": 4406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1703287554646247e-06,
+      "loss": 0.7988,
+      "step": 4407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.167400850535152e-06,
+      "loss": 0.8237,
+      "step": 4408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1644763856325126e-06,
+      "loss": 0.8149,
+      "step": 4409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1615553618956943e-06,
+      "loss": 0.7671,
+      "step": 4410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1586377804623483e-06,
+      "loss": 0.8364,
+      "step": 4411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1557236424687823e-06,
+      "loss": 0.7441,
+      "step": 4412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.152812949049963e-06,
+      "loss": 0.79,
+      "step": 4413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1499057013395177e-06,
+      "loss": 0.7783,
+      "step": 4414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1470019004697296e-06,
+      "loss": 0.8022,
+      "step": 4415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1441015475715412e-06,
+      "loss": 0.8203,
+      "step": 4416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1412046437745494e-06,
+      "loss": 0.7959,
+      "step": 4417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1383111902070099e-06,
+      "loss": 0.7974,
+      "step": 4418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1354211879958355e-06,
+      "loss": 0.7759,
+      "step": 4419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1325346382665924e-06,
+      "loss": 0.8042,
+      "step": 4420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1296515421435038e-06,
+      "loss": 0.7812,
+      "step": 4421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1267719007494483e-06,
+      "loss": 0.8145,
+      "step": 4422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1238957152059548e-06,
+      "loss": 0.792,
+      "step": 4423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1210229866332112e-06,
+      "loss": 0.8369,
+      "step": 4424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1181537161500612e-06,
+      "loss": 0.7363,
+      "step": 4425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1152879048739918e-06,
+      "loss": 0.8384,
+      "step": 4426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1124255539211515e-06,
+      "loss": 0.7993,
+      "step": 4427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1095666644063386e-06,
+      "loss": 0.7852,
+      "step": 4428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.106711237442999e-06,
+      "loss": 0.77,
+      "step": 4429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.103859274143241e-06,
+      "loss": 0.8008,
+      "step": 4430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1010107756178145e-06,
+      "loss": 0.7476,
+      "step": 4431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0981657429761195e-06,
+      "loss": 0.8076,
+      "step": 4432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0953241773262102e-06,
+      "loss": 0.8511,
+      "step": 4433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0924860797747894e-06,
+      "loss": 0.7686,
+      "step": 4434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0896514514272105e-06,
+      "loss": 0.8247,
+      "step": 4435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.086820293387475e-06,
+      "loss": 0.814,
+      "step": 4436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.083992606758232e-06,
+      "loss": 0.7896,
+      "step": 4437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0811683926407756e-06,
+      "loss": 0.7915,
+      "step": 4438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0783476521350532e-06,
+      "loss": 0.7905,
+      "step": 4439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.075530386339655e-06,
+      "loss": 0.7769,
+      "step": 4440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0727165963518238e-06,
+      "loss": 0.7676,
+      "step": 4441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0699062832674413e-06,
+      "loss": 0.7427,
+      "step": 4442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0670994481810415e-06,
+      "loss": 0.7891,
+      "step": 4443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.064296092185796e-06,
+      "loss": 0.7793,
+      "step": 4444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0614962163735264e-06,
+      "loss": 0.8354,
+      "step": 4445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0586998218347033e-06,
+      "loss": 0.729,
+      "step": 4446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.055906909658433e-06,
+      "loss": 0.7739,
+      "step": 4447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0531174809324706e-06,
+      "loss": 0.7705,
+      "step": 4448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0503315367432155e-06,
+      "loss": 0.7671,
+      "step": 4449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0475490781757003e-06,
+      "loss": 0.7915,
+      "step": 4450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0447701063136151e-06,
+      "loss": 0.8096,
+      "step": 4451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0419946222392808e-06,
+      "loss": 0.7554,
+      "step": 4452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0392226270336636e-06,
+      "loss": 0.8008,
+      "step": 4453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0364541217763712e-06,
+      "loss": 0.4744,
+      "step": 4454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0336891075456511e-06,
+      "loss": 0.8213,
+      "step": 4455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.030927585418392e-06,
+      "loss": 0.7539,
+      "step": 4456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0281695564701232e-06,
+      "loss": 0.8667,
+      "step": 4457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0254150217750103e-06,
+      "loss": 0.7871,
+      "step": 4458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0226639824058626e-06,
+      "loss": 0.7729,
+      "step": 4459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0199164394341232e-06,
+      "loss": 0.8242,
+      "step": 4460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0171723939298784e-06,
+      "loss": 0.7939,
+      "step": 4461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0144318469618475e-06,
+      "loss": 0.8047,
+      "step": 4462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0116947995973914e-06,
+      "loss": 0.7654,
+      "step": 4463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.008961252902505e-06,
+      "loss": 0.8252,
+      "step": 4464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0062312079418223e-06,
+      "loss": 0.7759,
+      "step": 4465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0035046657786118e-06,
+      "loss": 0.7773,
+      "step": 4466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0007816274747772e-06,
+      "loss": 0.8564,
+      "step": 4467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.980620940908593e-07,
+      "loss": 0.6765,
+      "step": 4468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.953460666860326e-07,
+      "loss": 0.7993,
+      "step": 4469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.92633546318107e-07,
+      "loss": 0.7686,
+      "step": 4470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.89924534043527e-07,
+      "loss": 0.77,
+      "step": 4471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.872190309173657e-07,
+      "loss": 0.7969,
+      "step": 4472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.8451703799334e-07,
+      "loss": 0.7715,
+      "step": 4473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.818185563237915e-07,
+      "loss": 0.8315,
+      "step": 4474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.791235869596927e-07,
+      "loss": 0.8184,
+      "step": 4475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.76432130950654e-07,
+      "loss": 0.7427,
+      "step": 4476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.73744189344914e-07,
+      "loss": 0.8047,
+      "step": 4477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.710597631893448e-07,
+      "loss": 0.7876,
+      "step": 4478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.683788535294491e-07,
+      "loss": 0.8101,
+      "step": 4479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.657014614093596e-07,
+      "loss": 0.8135,
+      "step": 4480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.630275878718353e-07,
+      "loss": 0.793,
+      "step": 4481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.603572339582678e-07,
+      "loss": 0.8145,
+      "step": 4482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.576904007086818e-07,
+      "loss": 0.8145,
+      "step": 4483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.55027089161724e-07,
+      "loss": 0.7642,
+      "step": 4484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.523673003546752e-07,
+      "loss": 0.7759,
+      "step": 4485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.497110353234429e-07,
+      "loss": 0.7993,
+      "step": 4486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.470582951025553e-07,
+      "loss": 0.7705,
+      "step": 4487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.444090807251761e-07,
+      "loss": 0.7583,
+      "step": 4488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.417633932230951e-07,
+      "loss": 0.7168,
+      "step": 4489
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.391212336267253e-07,
+      "loss": 0.7437,
+      "step": 4490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.364826029651064e-07,
+      "loss": 0.7798,
+      "step": 4491
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.338475022659066e-07,
+      "loss": 0.7305,
+      "step": 4492
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.312159325554115e-07,
+      "loss": 0.4912,
+      "step": 4493
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.285878948585426e-07,
+      "loss": 0.7881,
+      "step": 4494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.259633901988374e-07,
+      "loss": 0.8364,
+      "step": 4495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.233424195984608e-07,
+      "loss": 0.77,
+      "step": 4496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.207249840782007e-07,
+      "loss": 0.7939,
+      "step": 4497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.18111084657467e-07,
+      "loss": 0.8027,
+      "step": 4498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.155007223542944e-07,
+      "loss": 0.5007,
+      "step": 4499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.128938981853397e-07,
+      "loss": 0.4631,
+      "step": 4500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.102906131658784e-07,
+      "loss": 0.7651,
+      "step": 4501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.076908683098129e-07,
+      "loss": 0.7998,
+      "step": 4502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.050946646296643e-07,
+      "loss": 0.8408,
+      "step": 4503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.025020031365727e-07,
+      "loss": 0.7905,
+      "step": 4504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.999128848403005e-07,
+      "loss": 0.7964,
+      "step": 4505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.973273107492309e-07,
+      "loss": 0.7837,
+      "step": 4506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.947452818703662e-07,
+      "loss": 0.7983,
+      "step": 4507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.921667992093264e-07,
+      "loss": 0.7915,
+      "step": 4508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.895918637703515e-07,
+      "loss": 0.8198,
+      "step": 4509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.870204765563006e-07,
+      "loss": 0.8154,
+      "step": 4510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.844526385686513e-07,
+      "loss": 0.7656,
+      "step": 4511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.818883508074949e-07,
+      "loss": 0.8125,
+      "step": 4512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.793276142715456e-07,
+      "loss": 0.7837,
+      "step": 4513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.767704299581315e-07,
+      "loss": 0.7979,
+      "step": 4514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.742167988631944e-07,
+      "loss": 0.7954,
+      "step": 4515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.716667219813024e-07,
+      "loss": 0.8516,
+      "step": 4516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.691202003056253e-07,
+      "loss": 0.8208,
+      "step": 4517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.665772348279588e-07,
+      "loss": 0.4402,
+      "step": 4518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.640378265387084e-07,
+      "loss": 0.7725,
+      "step": 4519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.615019764268951e-07,
+      "loss": 0.8325,
+      "step": 4520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.589696854801588e-07,
+      "loss": 0.7778,
+      "step": 4521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.564409546847474e-07,
+      "loss": 0.814,
+      "step": 4522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.539157850255231e-07,
+      "loss": 0.7397,
+      "step": 4523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.513941774859625e-07,
+      "loss": 0.8188,
+      "step": 4524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.488761330481521e-07,
+      "loss": 0.7222,
+      "step": 4525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.463616526927975e-07,
+      "loss": 0.4836,
+      "step": 4526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.4385073739921e-07,
+      "loss": 0.7871,
+      "step": 4527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.413433881453159e-07,
+      "loss": 0.8076,
+      "step": 4528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.388396059076476e-07,
+      "loss": 0.7969,
+      "step": 4529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.363393916613505e-07,
+      "loss": 0.8252,
+      "step": 4530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.338427463801868e-07,
+      "loss": 0.7783,
+      "step": 4531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.313496710365199e-07,
+      "loss": 0.77,
+      "step": 4532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.288601666013274e-07,
+      "loss": 0.8076,
+      "step": 4533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.26374234044196e-07,
+      "loss": 0.8149,
+      "step": 4534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.23891874333318e-07,
+      "loss": 0.8394,
+      "step": 4535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.214130884354953e-07,
+      "loss": 0.7432,
+      "step": 4536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.189378773161427e-07,
+      "loss": 0.4641,
+      "step": 4537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.164662419392777e-07,
+      "loss": 0.8276,
+      "step": 4538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.139981832675259e-07,
+      "loss": 0.8281,
+      "step": 4539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.115337022621239e-07,
+      "loss": 0.749,
+      "step": 4540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.090727998829051e-07,
+      "loss": 0.8198,
+      "step": 4541
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.066154770883205e-07,
+      "loss": 0.7612,
+      "step": 4542
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.041617348354214e-07,
+      "loss": 0.8652,
+      "step": 4543
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.017115740798642e-07,
+      "loss": 0.7905,
+      "step": 4544
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.992649957759124e-07,
+      "loss": 0.8311,
+      "step": 4545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.968220008764327e-07,
+      "loss": 0.769,
+      "step": 4546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.943825903328972e-07,
+      "loss": 0.7339,
+      "step": 4547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.919467650953816e-07,
+      "loss": 0.7744,
+      "step": 4548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.895145261125647e-07,
+      "loss": 0.7563,
+      "step": 4549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.870858743317289e-07,
+      "loss": 0.4646,
+      "step": 4550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.846608106987608e-07,
+      "loss": 0.7764,
+      "step": 4551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.822393361581482e-07,
+      "loss": 0.749,
+      "step": 4552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.798214516529801e-07,
+      "loss": 0.7666,
+      "step": 4553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.774071581249487e-07,
+      "loss": 0.7744,
+      "step": 4554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.749964565143486e-07,
+      "loss": 0.7651,
+      "step": 4555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.725893477600743e-07,
+      "loss": 0.7769,
+      "step": 4556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.701858327996192e-07,
+      "loss": 0.7837,
+      "step": 4557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.677859125690801e-07,
+      "loss": 0.7388,
+      "step": 4558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.653895880031514e-07,
+      "loss": 0.8159,
+      "step": 4559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.62996860035129e-07,
+      "loss": 0.4482,
+      "step": 4560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.606077295969072e-07,
+      "loss": 0.7896,
+      "step": 4561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.582221976189785e-07,
+      "loss": 0.8462,
+      "step": 4562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.55840265030433e-07,
+      "loss": 0.8008,
+      "step": 4563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.534619327589666e-07,
+      "loss": 0.8037,
+      "step": 4564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.510872017308601e-07,
+      "loss": 0.7661,
+      "step": 4565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.487160728710008e-07,
+      "loss": 0.7793,
+      "step": 4566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.463485471028719e-07,
+      "loss": 0.7354,
+      "step": 4567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.439846253485494e-07,
+      "loss": 0.7749,
+      "step": 4568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.416243085287123e-07,
+      "loss": 0.8154,
+      "step": 4569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.392675975626307e-07,
+      "loss": 0.8228,
+      "step": 4570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.369144933681682e-07,
+      "loss": 0.7534,
+      "step": 4571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.345649968617896e-07,
+      "loss": 0.7725,
+      "step": 4572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.322191089585495e-07,
+      "loss": 0.7422,
+      "step": 4573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.298768305721027e-07,
+      "loss": 0.7456,
+      "step": 4574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.275381626146938e-07,
+      "loss": 0.8105,
+      "step": 4575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.25203105997162e-07,
+      "loss": 0.8208,
+      "step": 4576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2287166162894e-07,
+      "loss": 0.7578,
+      "step": 4577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.205438304180512e-07,
+      "loss": 0.77,
+      "step": 4578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.182196132711194e-07,
+      "loss": 0.7549,
+      "step": 4579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.15899011093355e-07,
+      "loss": 0.7842,
+      "step": 4580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.135820247885595e-07,
+      "loss": 0.79,
+      "step": 4581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.112686552591308e-07,
+      "loss": 0.832,
+      "step": 4582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.089589034060518e-07,
+      "loss": 0.8232,
+      "step": 4583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.066527701289016e-07,
+      "loss": 0.8032,
+      "step": 4584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.043502563258497e-07,
+      "loss": 0.7437,
+      "step": 4585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.020513628936543e-07,
+      "loss": 0.7866,
+      "step": 4586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.997560907276635e-07,
+      "loss": 0.731,
+      "step": 4587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.974644407218168e-07,
+      "loss": 0.7808,
+      "step": 4588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.951764137686378e-07,
+      "loss": 0.7935,
+      "step": 4589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.928920107592474e-07,
+      "loss": 0.856,
+      "step": 4590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.906112325833491e-07,
+      "loss": 0.7681,
+      "step": 4591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.883340801292349e-07,
+      "loss": 0.4656,
+      "step": 4592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.860605542837873e-07,
+      "loss": 0.8325,
+      "step": 4593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.837906559324748e-07,
+      "loss": 0.5022,
+      "step": 4594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.815243859593534e-07,
+      "loss": 0.7739,
+      "step": 4595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.792617452470652e-07,
+      "loss": 0.4126,
+      "step": 4596
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.770027346768393e-07,
+      "loss": 0.7998,
+      "step": 4597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.747473551284922e-07,
+      "loss": 0.8413,
+      "step": 4598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.724956074804246e-07,
+      "loss": 0.7983,
+      "step": 4599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.702474926096225e-07,
+      "loss": 0.8125,
+      "step": 4600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.680030113916592e-07,
+      "loss": 0.7676,
+      "step": 4601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.657621647006896e-07,
+      "loss": 0.8516,
+      "step": 4602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.635249534094568e-07,
+      "loss": 0.4277,
+      "step": 4603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.612913783892849e-07,
+      "loss": 0.8042,
+      "step": 4604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.590614405100837e-07,
+      "loss": 0.752,
+      "step": 4605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.568351406403428e-07,
+      "loss": 0.7397,
+      "step": 4606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.546124796471443e-07,
+      "loss": 0.7915,
+      "step": 4607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.523934583961411e-07,
+      "loss": 0.811,
+      "step": 4608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.501780777515743e-07,
+      "loss": 0.8628,
+      "step": 4609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.479663385762691e-07,
+      "loss": 0.7988,
+      "step": 4610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.457582417316266e-07,
+      "loss": 0.7773,
+      "step": 4611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.435537880776377e-07,
+      "loss": 0.8025,
+      "step": 4612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.413529784728678e-07,
+      "loss": 0.7998,
+      "step": 4613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.391558137744636e-07,
+      "loss": 0.7031,
+      "step": 4614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.369622948381527e-07,
+      "loss": 0.8525,
+      "step": 4615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.34772422518245e-07,
+      "loss": 0.8013,
+      "step": 4616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.325861976676296e-07,
+      "loss": 0.8706,
+      "step": 4617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.304036211377717e-07,
+      "loss": 0.791,
+      "step": 4618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.282246937787229e-07,
+      "loss": 0.791,
+      "step": 4619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.260494164391029e-07,
+      "loss": 0.8125,
+      "step": 4620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.238777899661163e-07,
+      "loss": 0.8677,
+      "step": 4621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.217098152055478e-07,
+      "loss": 0.749,
+      "step": 4622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.195454930017563e-07,
+      "loss": 0.7637,
+      "step": 4623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.173848241976798e-07,
+      "loss": 0.772,
+      "step": 4624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.152278096348318e-07,
+      "loss": 0.7515,
+      "step": 4625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.130744501533003e-07,
+      "loss": 0.8428,
+      "step": 4626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.109247465917578e-07,
+      "loss": 0.8042,
+      "step": 4627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.087786997874457e-07,
+      "loss": 0.7681,
+      "step": 4628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.066363105761841e-07,
+      "loss": 0.8057,
+      "step": 4629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.044975797923669e-07,
+      "loss": 0.4609,
+      "step": 4630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.023625082689666e-07,
+      "loss": 0.7764,
+      "step": 4631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.002310968375247e-07,
+      "loss": 0.8003,
+      "step": 4632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.981033463281627e-07,
+      "loss": 0.8481,
+      "step": 4633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.959792575695744e-07,
+      "loss": 0.8198,
+      "step": 4634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.938588313890259e-07,
+      "loss": 0.8188,
+      "step": 4635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.917420686123609e-07,
+      "loss": 0.7842,
+      "step": 4636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.896289700639912e-07,
+      "loss": 0.8579,
+      "step": 4637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.875195365669051e-07,
+      "loss": 0.7666,
+      "step": 4638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.85413768942662e-07,
+      "loss": 0.7852,
+      "step": 4639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.833116680113937e-07,
+      "loss": 0.7705,
+      "step": 4640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.812132345918042e-07,
+      "loss": 0.8062,
+      "step": 4641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.791184695011698e-07,
+      "loss": 0.75,
+      "step": 4642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.770273735553366e-07,
+      "loss": 0.8354,
+      "step": 4643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.749399475687223e-07,
+      "loss": 0.8101,
+      "step": 4644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.728561923543164e-07,
+      "loss": 0.833,
+      "step": 4645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.707761087236763e-07,
+      "loss": 0.8198,
+      "step": 4646
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.686996974869318e-07,
+      "loss": 0.8159,
+      "step": 4647
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.666269594527807e-07,
+      "loss": 0.8574,
+      "step": 4648
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.645578954284925e-07,
+      "loss": 0.71,
+      "step": 4649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.624925062199027e-07,
+      "loss": 0.686,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.604307926314179e-07,
+      "loss": 0.8096,
+      "step": 4651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.583727554660123e-07,
+      "loss": 0.7427,
+      "step": 4652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.563183955252283e-07,
+      "loss": 0.7856,
+      "step": 4653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.542677136091734e-07,
+      "loss": 0.7256,
+      "step": 4654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.522207105165323e-07,
+      "loss": 0.7695,
+      "step": 4655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.501773870445448e-07,
+      "loss": 0.7847,
+      "step": 4656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.481377439890224e-07,
+      "loss": 0.8301,
+      "step": 4657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.461017821443459e-07,
+      "loss": 0.8208,
+      "step": 4658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.440695023034582e-07,
+      "loss": 0.7837,
+      "step": 4659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.420409052578712e-07,
+      "loss": 0.7998,
+      "step": 4660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.400159917976633e-07,
+      "loss": 0.7749,
+      "step": 4661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.379947627114734e-07,
+      "loss": 0.4536,
+      "step": 4662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.359772187865086e-07,
+      "loss": 0.8008,
+      "step": 4663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.339633608085393e-07,
+      "loss": 0.8164,
+      "step": 4664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.319531895619046e-07,
+      "loss": 0.8208,
+      "step": 4665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.299467058295027e-07,
+      "loss": 0.8184,
+      "step": 4666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.279439103927997e-07,
+      "loss": 0.77,
+      "step": 4667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.259448040318205e-07,
+      "loss": 0.7808,
+      "step": 4668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.239493875251545e-07,
+      "loss": 0.811,
+      "step": 4669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2195766164996e-07,
+      "loss": 0.8359,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.199696271819499e-07,
+      "loss": 0.8196,
+      "step": 4671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.179852848954037e-07,
+      "loss": 0.8237,
+      "step": 4672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.160046355631632e-07,
+      "loss": 0.8838,
+      "step": 4673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.140276799566257e-07,
+      "loss": 0.8193,
+      "step": 4674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.12054418845761e-07,
+      "loss": 0.7749,
+      "step": 4675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.100848529990909e-07,
+      "loss": 0.8242,
+      "step": 4676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.081189831837008e-07,
+      "loss": 0.8032,
+      "step": 4677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.061568101652381e-07,
+      "loss": 0.8262,
+      "step": 4678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.041983347079093e-07,
+      "loss": 0.7563,
+      "step": 4679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.022435575744767e-07,
+      "loss": 0.4976,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.002924795262709e-07,
+      "loss": 0.8374,
+      "step": 4681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.983451013231744e-07,
+      "loss": 0.7024,
+      "step": 4682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.964014237236325e-07,
+      "loss": 0.7881,
+      "step": 4683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.944614474846488e-07,
+      "loss": 0.7949,
+      "step": 4684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.92525173361782e-07,
+      "loss": 0.813,
+      "step": 4685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.905926021091534e-07,
+      "loss": 0.7412,
+      "step": 4686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.886637344794409e-07,
+      "loss": 0.8115,
+      "step": 4687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.867385712238781e-07,
+      "loss": 0.4409,
+      "step": 4688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.848171130922563e-07,
+      "loss": 0.7515,
+      "step": 4689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.828993608329269e-07,
+      "loss": 0.7998,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.809853151927946e-07,
+      "loss": 0.7632,
+      "step": 4691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.790749769173219e-07,
+      "loss": 0.8076,
+      "step": 4692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.771683467505262e-07,
+      "loss": 0.7314,
+      "step": 4693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.75265425434982e-07,
+      "loss": 0.8813,
+      "step": 4694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.733662137118189e-07,
+      "loss": 0.8408,
+      "step": 4695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7147071232072073e-07,
+      "loss": 0.7651,
+      "step": 4696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.695789219999292e-07,
+      "loss": 0.4937,
+      "step": 4697
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6769084348623817e-07,
+      "loss": 0.7485,
+      "step": 4698
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6580647751499507e-07,
+      "loss": 0.8091,
+      "step": 4699
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6392582482010396e-07,
+      "loss": 0.6582,
+      "step": 4700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6204888613402e-07,
+      "loss": 0.7583,
+      "step": 4701
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.60175662187754e-07,
+      "loss": 0.7812,
+      "step": 4702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.583061537108724e-07,
+      "loss": 0.7974,
+      "step": 4703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.564403614314872e-07,
+      "loss": 0.8022,
+      "step": 4704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.545782860762682e-07,
+      "loss": 0.6768,
+      "step": 4705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5271992837043863e-07,
+      "loss": 0.8257,
+      "step": 4706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.508652890377685e-07,
+      "loss": 0.8203,
+      "step": 4707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.490143688005877e-07,
+      "loss": 0.7861,
+      "step": 4708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.471671683797707e-07,
+      "loss": 0.7583,
+      "step": 4709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4532368849474427e-07,
+      "loss": 0.8433,
+      "step": 4710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4348392986348853e-07,
+      "loss": 0.7656,
+      "step": 4711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.416478932025314e-07,
+      "loss": 0.7666,
+      "step": 4712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3981557922695537e-07,
+      "loss": 0.7524,
+      "step": 4713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.379869886503907e-07,
+      "loss": 0.7695,
+      "step": 4714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.361621221850154e-07,
+      "loss": 0.8491,
+      "step": 4715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3434098054156213e-07,
+      "loss": 0.7935,
+      "step": 4716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3252356442930464e-07,
+      "loss": 0.4514,
+      "step": 4717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.307098745560756e-07,
+      "loss": 0.7539,
+      "step": 4718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2889991162825e-07,
+      "loss": 0.8047,
+      "step": 4719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2709367635075273e-07,
+      "loss": 0.8018,
+      "step": 4720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2529116942705784e-07,
+      "loss": 0.7871,
+      "step": 4721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.234923915591871e-07,
+      "loss": 0.7837,
+      "step": 4722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2169734344770564e-07,
+      "loss": 0.8101,
+      "step": 4723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.199060257917331e-07,
+      "loss": 0.7632,
+      "step": 4724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1811843928893256e-07,
+      "loss": 0.791,
+      "step": 4725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.163345846355149e-07,
+      "loss": 0.7485,
+      "step": 4726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.145544625262343e-07,
+      "loss": 0.4302,
+      "step": 4727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.127780736543963e-07,
+      "loss": 0.73,
+      "step": 4728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.110054187118484e-07,
+      "loss": 0.7695,
+      "step": 4729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0923649838898625e-07,
+      "loss": 0.811,
+      "step": 4730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0747131337474986e-07,
+      "loss": 0.8384,
+      "step": 4731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0570986435662483e-07,
+      "loss": 0.811,
+      "step": 4732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0395215202064133e-07,
+      "loss": 0.8657,
+      "step": 4733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.02198177051375e-07,
+      "loss": 0.7744,
+      "step": 4734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.00447940131945e-07,
+      "loss": 0.8066,
+      "step": 4735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9870144194401385e-07,
+      "loss": 0.7778,
+      "step": 4736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9695868316779073e-07,
+      "loss": 0.7422,
+      "step": 4737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.952196644820272e-07,
+      "loss": 0.8315,
+      "step": 4738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9348438656401476e-07,
+      "loss": 0.8022,
+      "step": 4739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.917528500895962e-07,
+      "loss": 0.7759,
+      "step": 4740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9002505573314753e-07,
+      "loss": 0.7378,
+      "step": 4741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8830100416759277e-07,
+      "loss": 0.8257,
+      "step": 4742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8658069606439806e-07,
+      "loss": 0.8438,
+      "step": 4743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8486413209356974e-07,
+      "loss": 0.832,
+      "step": 4744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.831513129236564e-07,
+      "loss": 0.7798,
+      "step": 4745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.814422392217509e-07,
+      "loss": 0.7642,
+      "step": 4746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.797369116534833e-07,
+      "loss": 0.731,
+      "step": 4747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.780353308830276e-07,
+      "loss": 0.813,
+      "step": 4748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.76337497573096e-07,
+      "loss": 0.8042,
+      "step": 4749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7464341238494187e-07,
+      "loss": 0.7925,
+      "step": 4750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.729530759783628e-07,
+      "loss": 0.7485,
+      "step": 4751
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.712664890116924e-07,
+      "loss": 0.7759,
+      "step": 4752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.695836521418017e-07,
+      "loss": 0.7471,
+      "step": 4753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.679045660241065e-07,
+      "loss": 0.4819,
+      "step": 4754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6622923131255704e-07,
+      "loss": 0.7935,
+      "step": 4755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6455764865964694e-07,
+      "loss": 0.8018,
+      "step": 4756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.628898187164076e-07,
+      "loss": 0.791,
+      "step": 4757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6122574213240615e-07,
+      "loss": 0.7656,
+      "step": 4758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.595654195557485e-07,
+      "loss": 0.8315,
+      "step": 4759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5790885163307975e-07,
+      "loss": 0.8228,
+      "step": 4760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.562560390095837e-07,
+      "loss": 0.813,
+      "step": 4761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5460698232897995e-07,
+      "loss": 0.8022,
+      "step": 4762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.529616822335258e-07,
+      "loss": 0.4509,
+      "step": 4763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5132013936401424e-07,
+      "loss": 0.8086,
+      "step": 4764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.496823543597749e-07,
+      "loss": 0.7832,
+      "step": 4765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.480483278586777e-07,
+      "loss": 0.4487,
+      "step": 4766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4641806049712454e-07,
+      "loss": 0.7485,
+      "step": 4767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.447915529100543e-07,
+      "loss": 0.7563,
+      "step": 4768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4316880573094235e-07,
+      "loss": 0.7891,
+      "step": 4769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.415498195917999e-07,
+      "loss": 0.811,
+      "step": 4770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3993459512316806e-07,
+      "loss": 0.7881,
+      "step": 4771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3832313295413365e-07,
+      "loss": 0.7847,
+      "step": 4772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3671543371230687e-07,
+      "loss": 0.8047,
+      "step": 4773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3511149802384013e-07,
+      "loss": 0.7642,
+      "step": 4774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3351132651341713e-07,
+      "loss": 0.8208,
+      "step": 4775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.319149198042537e-07,
+      "loss": 0.7549,
+      "step": 4776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.303222785181037e-07,
+      "loss": 0.8662,
+      "step": 4777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2873340327524983e-07,
+      "loss": 0.8096,
+      "step": 4778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.271482946945126e-07,
+      "loss": 0.8276,
+      "step": 4779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2556695339324154e-07,
+      "loss": 0.8145,
+      "step": 4780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.239893799873217e-07,
+      "loss": 0.7725,
+      "step": 4781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2241557509116837e-07,
+      "loss": 0.7681,
+      "step": 4782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.208455393177312e-07,
+      "loss": 0.7725,
+      "step": 4783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.192792732784922e-07,
+      "loss": 0.8105,
+      "step": 4784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.177167775834622e-07,
+      "loss": 0.7979,
+      "step": 4785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1615805284118564e-07,
+      "loss": 0.7627,
+      "step": 4786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.146030996587368e-07,
+      "loss": 0.7549,
+      "step": 4787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.130519186417269e-07,
+      "loss": 0.4485,
+      "step": 4788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.115045103942893e-07,
+      "loss": 0.4731,
+      "step": 4789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0996087551909414e-07,
+      "loss": 0.752,
+      "step": 4790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.084210146173383e-07,
+      "loss": 0.7769,
+      "step": 4791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.068849282887509e-07,
+      "loss": 0.8462,
+      "step": 4792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0535261713159015e-07,
+      "loss": 0.8105,
+      "step": 4793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.038240817426474e-07,
+      "loss": 0.4436,
+      "step": 4794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.022993227172366e-07,
+      "loss": 0.7842,
+      "step": 4795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0077834064920595e-07,
+      "loss": 0.8037,
+      "step": 4796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9926113613093164e-07,
+      "loss": 0.7344,
+      "step": 4797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9774770975331656e-07,
+      "loss": 0.7544,
+      "step": 4798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.962380621057981e-07,
+      "loss": 0.7915,
+      "step": 4799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9473219377633477e-07,
+      "loss": 0.8413,
+      "step": 4800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.932301053514164e-07,
+      "loss": 0.7087,
+      "step": 4801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.917317974160605e-07,
+      "loss": 0.7769,
+      "step": 4802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9023727055381146e-07,
+      "loss": 0.7778,
+      "step": 4803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.887465253467436e-07,
+      "loss": 0.8091,
+      "step": 4804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.872595623754548e-07,
+      "loss": 0.7993,
+      "step": 4805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.85776382219074e-07,
+      "loss": 0.7817,
+      "step": 4806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.842969854552513e-07,
+      "loss": 0.7773,
+      "step": 4807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.82821372660167e-07,
+      "loss": 0.7832,
+      "step": 4808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.81349544408529e-07,
+      "loss": 0.7573,
+      "step": 4809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7988150127356785e-07,
+      "loss": 0.7554,
+      "step": 4810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.784172438270416e-07,
+      "loss": 0.8296,
+      "step": 4811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.769567726392353e-07,
+      "loss": 0.7822,
+      "step": 4812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.755000882789538e-07,
+      "loss": 0.7656,
+      "step": 4813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.740471913135356e-07,
+      "loss": 0.8311,
+      "step": 4814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.72598082308837e-07,
+      "loss": 0.7695,
+      "step": 4815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7115276182924197e-07,
+      "loss": 0.8037,
+      "step": 4816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.697112304376581e-07,
+      "loss": 0.8525,
+      "step": 4817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.682734886955196e-07,
+      "loss": 0.7544,
+      "step": 4818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6683953716278077e-07,
+      "loss": 0.7866,
+      "step": 4819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.654093763979215e-07,
+      "loss": 0.7534,
+      "step": 4820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6398300695794853e-07,
+      "loss": 0.8408,
+      "step": 4821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6256042939838634e-07,
+      "loss": 0.7754,
+      "step": 4822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6114164427328613e-07,
+      "loss": 0.8359,
+      "step": 4823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5972665213522155e-07,
+      "loss": 0.7798,
+      "step": 4824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.583154535352883e-07,
+      "loss": 0.7832,
+      "step": 4825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5690804902310463e-07,
+      "loss": 0.8403,
+      "step": 4826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.555044391468131e-07,
+      "loss": 0.8179,
+      "step": 4827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5410462445307535e-07,
+      "loss": 0.7832,
+      "step": 4828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.527086054870764e-07,
+      "loss": 0.792,
+      "step": 4829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5131638279252247e-07,
+      "loss": 0.749,
+      "step": 4830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4992795691164304e-07,
+      "loss": 0.812,
+      "step": 4831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.485433283851868e-07,
+      "loss": 0.8477,
+      "step": 4832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.471624977524245e-07,
+      "loss": 0.7734,
+      "step": 4833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.457854655511471e-07,
+      "loss": 0.7852,
+      "step": 4834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.444122323176656e-07,
+      "loss": 0.8667,
+      "step": 4835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.430427985868122e-07,
+      "loss": 0.772,
+      "step": 4836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.416771648919425e-07,
+      "loss": 0.7041,
+      "step": 4837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4031533176492536e-07,
+      "loss": 0.8037,
+      "step": 4838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3895729973615444e-07,
+      "loss": 0.8369,
+      "step": 4839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3760306933454215e-07,
+      "loss": 0.8169,
+      "step": 4840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3625264108751765e-07,
+      "loss": 0.7842,
+      "step": 4841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3490601552103366e-07,
+      "loss": 0.752,
+      "step": 4842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3356319315956054e-07,
+      "loss": 0.439,
+      "step": 4843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3222417452608337e-07,
+      "loss": 0.7466,
+      "step": 4844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.308889601421105e-07,
+      "loss": 0.8179,
+      "step": 4845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2955755052766593e-07,
+      "loss": 0.7886,
+      "step": 4846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.282299462012949e-07,
+      "loss": 0.8442,
+      "step": 4847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2690614768005714e-07,
+      "loss": 0.8389,
+      "step": 4848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2558615547953354e-07,
+      "loss": 0.8032,
+      "step": 4849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2426997011381847e-07,
+      "loss": 0.7744,
+      "step": 4850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2295759209552515e-07,
+      "loss": 0.8271,
+      "step": 4851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2164902193578696e-07,
+      "loss": 0.8296,
+      "step": 4852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.203442601442507e-07,
+      "loss": 0.7964,
+      "step": 4853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1904330722908206e-07,
+      "loss": 0.8213,
+      "step": 4854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1774616369696243e-07,
+      "loss": 0.8066,
+      "step": 4855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.164528300530866e-07,
+      "loss": 0.7373,
+      "step": 4856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1516330680117048e-07,
+      "loss": 0.8081,
+      "step": 4857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.138775944434457e-07,
+      "loss": 0.7397,
+      "step": 4858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1259569348065502e-07,
+      "loss": 0.7441,
+      "step": 4859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1131760441206017e-07,
+      "loss": 0.8242,
+      "step": 4860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1004332773543857e-07,
+      "loss": 0.7822,
+      "step": 4861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0877286394708206e-07,
+      "loss": 0.7393,
+      "step": 4862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.07506213541796e-07,
+      "loss": 0.8101,
+      "step": 4863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0624337701290353e-07,
+      "loss": 0.4382,
+      "step": 4864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.049843548522401e-07,
+      "loss": 0.7295,
+      "step": 4865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0372914755015572e-07,
+      "loss": 0.8525,
+      "step": 4866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0247775559551597e-07,
+      "loss": 0.7842,
+      "step": 4867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.012301794756999e-07,
+      "loss": 0.79,
+      "step": 4868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9998641967659881e-07,
+      "loss": 0.8311,
+      "step": 4869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.987464766826197e-07,
+      "loss": 0.8003,
+      "step": 4870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9751035097668402e-07,
+      "loss": 0.7271,
+      "step": 4871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9627804304022224e-07,
+      "loss": 0.7876,
+      "step": 4872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9504955335318265e-07,
+      "loss": 0.7725,
+      "step": 4873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9382488239402252e-07,
+      "loss": 0.4258,
+      "step": 4874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.926040306397159e-07,
+      "loss": 0.7935,
+      "step": 4875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9138699856574462e-07,
+      "loss": 0.4307,
+      "step": 4876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.901737866461073e-07,
+      "loss": 0.7773,
+      "step": 4877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8896439535331158e-07,
+      "loss": 0.8164,
+      "step": 4878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.877588251583806e-07,
+      "loss": 0.73,
+      "step": 4879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8655707653084442e-07,
+      "loss": 0.7666,
+      "step": 4880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8535914993874859e-07,
+      "loss": 0.7803,
+      "step": 4881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.841650458486477e-07,
+      "loss": 0.7822,
+      "step": 4882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8297476472561083e-07,
+      "loss": 0.7593,
+      "step": 4883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.817883070332127e-07,
+      "loss": 0.7529,
+      "step": 4884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8060567323354594e-07,
+      "loss": 0.8257,
+      "step": 4885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7942686378720875e-07,
+      "loss": 0.7944,
+      "step": 4886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7825187915331054e-07,
+      "loss": 0.8413,
+      "step": 4887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7708071978947193e-07,
+      "loss": 0.8232,
+      "step": 4888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7591338615182363e-07,
+      "loss": 0.7744,
+      "step": 4889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7474987869500748e-07,
+      "loss": 0.7651,
+      "step": 4890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7359019787217435e-07,
+      "loss": 0.8081,
+      "step": 4891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7243434413498184e-07,
+      "loss": 0.792,
+      "step": 4892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7128231793360207e-07,
+      "loss": 0.7549,
+      "step": 4893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7013411971671168e-07,
+      "loss": 0.7666,
+      "step": 4894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6898974993150074e-07,
+      "loss": 0.7148,
+      "step": 4895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6784920902366607e-07,
+      "loss": 0.8052,
+      "step": 4896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6671249743741236e-07,
+      "loss": 0.7402,
+      "step": 4897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6557961561545434e-07,
+      "loss": 0.8037,
+      "step": 4898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.644505639990146e-07,
+      "loss": 0.7698,
+      "step": 4899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6332534302782476e-07,
+      "loss": 0.7354,
+      "step": 4900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6220395314012315e-07,
+      "loss": 0.7734,
+      "step": 4901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610863947726582e-07,
+      "loss": 0.7983,
+      "step": 4902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5997266836068393e-07,
+      "loss": 0.7524,
+      "step": 4903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5886277433796116e-07,
+      "loss": 0.4597,
+      "step": 4904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5775671313676077e-07,
+      "loss": 0.8389,
+      "step": 4905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5665448518786043e-07,
+      "loss": 0.811,
+      "step": 4906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5555609092054448e-07,
+      "loss": 0.876,
+      "step": 4907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5446153076260185e-07,
+      "loss": 0.4468,
+      "step": 4908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5337080514033153e-07,
+      "loss": 0.7793,
+      "step": 4909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5228391447853708e-07,
+      "loss": 0.7612,
+      "step": 4910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5120085920052986e-07,
+      "loss": 0.7588,
+      "step": 4911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5012163972812687e-07,
+      "loss": 0.8052,
+      "step": 4912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4904625648164972e-07,
+      "loss": 0.8174,
+      "step": 4913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.479747098799289e-07,
+      "loss": 0.7744,
+      "step": 4914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4690700034029727e-07,
+      "loss": 0.7847,
+      "step": 4915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4584312827859658e-07,
+      "loss": 0.7769,
+      "step": 4916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4478309410917212e-07,
+      "loss": 0.7471,
+      "step": 4917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4372689824487363e-07,
+      "loss": 0.835,
+      "step": 4918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.426745410970587e-07,
+      "loss": 0.8276,
+      "step": 4919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4162602307558615e-07,
+      "loss": 0.772,
+      "step": 4920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4058134458882267e-07,
+      "loss": 0.7686,
+      "step": 4921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3954050604363946e-07,
+      "loss": 0.7417,
+      "step": 4922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3850350784541e-07,
+      "loss": 0.8159,
+      "step": 4923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3747035039801459e-07,
+      "loss": 0.8008,
+      "step": 4924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.364410341038347e-07,
+      "loss": 0.7383,
+      "step": 4925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.354155593637574e-07,
+      "loss": 0.8125,
+      "step": 4926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3439392657717654e-07,
+      "loss": 0.7808,
+      "step": 4927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3337613614198385e-07,
+      "loss": 0.8306,
+      "step": 4928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3236218845457894e-07,
+      "loss": 0.7725,
+      "step": 4929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3135208390986363e-07,
+      "loss": 0.8281,
+      "step": 4930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3034582290124108e-07,
+      "loss": 0.7891,
+      "step": 4931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2934340582061998e-07,
+      "loss": 0.7197,
+      "step": 4932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.283448330584125e-07,
+      "loss": 0.7622,
+      "step": 4933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2735010500353085e-07,
+      "loss": 0.7388,
+      "step": 4934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2635922204339068e-07,
+      "loss": 0.7803,
+      "step": 4935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2537218456390999e-07,
+      "loss": 0.8398,
+      "step": 4936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2438899294951124e-07,
+      "loss": 0.7812,
+      "step": 4937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2340964758311814e-07,
+      "loss": 0.7339,
+      "step": 4938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2243414884615445e-07,
+      "loss": 0.8076,
+      "step": 4939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.214624971185463e-07,
+      "loss": 0.7788,
+      "step": 4940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2049469277872317e-07,
+      "loss": 0.8242,
+      "step": 4941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1953073620361355e-07,
+      "loss": 0.7583,
+      "step": 4942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.185706277686527e-07,
+      "loss": 0.8755,
+      "step": 4943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1761436784777036e-07,
+      "loss": 0.7656,
+      "step": 4944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.166619568134031e-07,
+      "loss": 0.7583,
+      "step": 4945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1571339503648305e-07,
+      "loss": 0.8521,
+      "step": 4946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1476868288644583e-07,
+      "loss": 0.8003,
+      "step": 4947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1382782073123044e-07,
+      "loss": 0.7866,
+      "step": 4948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1289080893727267e-07,
+      "loss": 0.7549,
+      "step": 4949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1195764786950946e-07,
+      "loss": 0.8071,
+      "step": 4950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1102833789137901e-07,
+      "loss": 0.8296,
+      "step": 4951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1010287936481845e-07,
+      "loss": 0.7812,
+      "step": 4952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0918127265026723e-07,
+      "loss": 0.7515,
+      "step": 4953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0826351810666047e-07,
+      "loss": 0.4441,
+      "step": 4954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0734961609143668e-07,
+      "loss": 0.8154,
+      "step": 4955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0643956696053337e-07,
+      "loss": 0.8179,
+      "step": 4956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.055333710683859e-07,
+      "loss": 0.7612,
+      "step": 4957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0463102876793086e-07,
+      "loss": 0.8374,
+      "step": 4958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0373254041060154e-07,
+      "loss": 0.6855,
+      "step": 4959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0283790634633139e-07,
+      "loss": 0.7729,
+      "step": 4960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0194712692355502e-07,
+      "loss": 0.7261,
+      "step": 4961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0106020248920156e-07,
+      "loss": 0.4312,
+      "step": 4962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.001771333887025e-07,
+      "loss": 0.7461,
+      "step": 4963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.929791996598492e-08,
+      "loss": 0.4126,
+      "step": 4964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.842256256347715e-08,
+      "loss": 0.7798,
+      "step": 4965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.755106152210314e-08,
+      "loss": 0.813,
+      "step": 4966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.668341718128471e-08,
+      "loss": 0.7871,
+      "step": 4967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.581962987894489e-08,
+      "loss": 0.8477,
+      "step": 4968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.495969995150012e-08,
+      "loss": 0.7959,
+      "step": 4969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.410362773386916e-08,
+      "loss": 0.4873,
+      "step": 4970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.325141355946532e-08,
+      "loss": 0.729,
+      "step": 4971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.240305776019864e-08,
+      "loss": 0.813,
+      "step": 4972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.155856066647817e-08,
+      "loss": 0.6675,
+      "step": 4973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.071792260720969e-08,
+      "loss": 0.8618,
+      "step": 4974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.988114390979686e-08,
+      "loss": 0.8242,
+      "step": 4975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.904822490013898e-08,
+      "loss": 0.7866,
+      "step": 4976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.821916590263102e-08,
+      "loss": 0.811,
+      "step": 4977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.739396724016913e-08,
+      "loss": 0.7451,
+      "step": 4978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.657262923413956e-08,
+      "loss": 0.8071,
+      "step": 4979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.575515220442976e-08,
+      "loss": 0.832,
+      "step": 4980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.494153646942282e-08,
+      "loss": 0.8188,
+      "step": 4981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.41317823459964e-08,
+      "loss": 0.793,
+      "step": 4982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.332589014952486e-08,
+      "loss": 0.7788,
+      "step": 4983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.252386019387826e-08,
+      "loss": 0.7803,
+      "step": 4984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.172569279142229e-08,
+      "loss": 0.8174,
+      "step": 4985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.09313882530205e-08,
+      "loss": 0.4487,
+      "step": 4986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.014094688802876e-08,
+      "loss": 0.7559,
+      "step": 4987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.935436900429971e-08,
+      "loss": 0.7729,
+      "step": 4988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.857165490818274e-08,
+      "loss": 0.8003,
+      "step": 4989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.779280490451846e-08,
+      "loss": 0.7983,
+      "step": 4990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.701781929664754e-08,
+      "loss": 0.4783,
+      "step": 4991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.624669838640297e-08,
+      "loss": 0.8135,
+      "step": 4992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.54794424741112e-08,
+      "loss": 0.8042,
+      "step": 4993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.471605185859654e-08,
+      "loss": 0.8042,
+      "step": 4994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.395652683717447e-08,
+      "loss": 0.7412,
+      "step": 4995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.320086770565837e-08,
+      "loss": 0.7979,
+      "step": 4996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.244907475835284e-08,
+      "loss": 0.791,
+      "step": 4997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.170114828805919e-08,
+      "loss": 0.7783,
+      "step": 4998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.09570885860722e-08,
+      "loss": 0.8184,
+      "step": 4999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.021689594218007e-08,
+      "loss": 0.7456,
+      "step": 5000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.94805706446644e-08,
+      "loss": 0.772,
+      "step": 5001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.874811298030138e-08,
+      "loss": 0.7935,
+      "step": 5002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.801952323436167e-08,
+      "loss": 0.8159,
+      "step": 5003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.729480169060832e-08,
+      "loss": 0.7837,
+      "step": 5004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.657394863129885e-08,
+      "loss": 0.7896,
+      "step": 5005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.585696433718203e-08,
+      "loss": 0.7954,
+      "step": 5006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.514384908750226e-08,
+      "loss": 0.8379,
+      "step": 5007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.443460315999516e-08,
+      "loss": 0.77,
+      "step": 5008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.3729226830892e-08,
+      "loss": 0.8096,
+      "step": 5009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.302772037491411e-08,
+      "loss": 0.7866,
+      "step": 5010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.233008406527629e-08,
+      "loss": 0.7905,
+      "step": 5011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.163631817368677e-08,
+      "loss": 0.792,
+      "step": 5012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.094642297034714e-08,
+      "loss": 0.7754,
+      "step": 5013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.026039872395028e-08,
+      "loss": 0.7671,
+      "step": 5014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.957824570168025e-08,
+      "loss": 0.8403,
+      "step": 5015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8899964169215616e-08,
+      "loss": 0.8179,
+      "step": 5016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.822555439072619e-08,
+      "loss": 0.7095,
+      "step": 5017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7555016628874086e-08,
+      "loss": 0.7886,
+      "step": 5018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6888351144813725e-08,
+      "loss": 0.8345,
+      "step": 5019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6225558198189644e-08,
+      "loss": 0.7749,
+      "step": 5020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.55666380471398e-08,
+      "loss": 0.8149,
+      "step": 5021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4911590948294456e-08,
+      "loss": 0.8076,
+      "step": 5022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.426041715677288e-08,
+      "loss": 0.7856,
+      "step": 5023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3613116926188867e-08,
+      "loss": 0.8374,
+      "step": 5024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.296969050864409e-08,
+      "loss": 0.8267,
+      "step": 5025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2330138154735863e-08,
+      "loss": 0.8467,
+      "step": 5026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.169446011354828e-08,
+      "loss": 0.7559,
+      "step": 5027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.106265663265775e-08,
+      "loss": 0.4412,
+      "step": 5028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.04347279581352e-08,
+      "loss": 0.8345,
+      "step": 5029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9810674334538346e-08,
+      "loss": 0.7283,
+      "step": 5030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.91904960049161e-08,
+      "loss": 0.7891,
+      "step": 5031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.857419321080969e-08,
+      "loss": 0.8042,
+      "step": 5032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.796176619225046e-08,
+      "loss": 0.7827,
+      "step": 5033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.73532151877587e-08,
+      "loss": 0.7798,
+      "step": 5034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6748540434348175e-08,
+      "loss": 0.8276,
+      "step": 5035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.614774216752049e-08,
+      "loss": 0.4722,
+      "step": 5036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5550820621267364e-08,
+      "loss": 0.7944,
+      "step": 5037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4957776028071723e-08,
+      "loss": 0.7847,
+      "step": 5038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.436860861890657e-08,
+      "loss": 0.8247,
+      "step": 5039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.378331862323504e-08,
+      "loss": 0.7671,
+      "step": 5040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.320190626900922e-08,
+      "loss": 0.4678,
+      "step": 5041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2624371782671315e-08,
+      "loss": 0.812,
+      "step": 5042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.205071538915362e-08,
+      "loss": 0.8335,
+      "step": 5043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1480937311877414e-08,
+      "loss": 0.8633,
+      "step": 5044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0915037772754075e-08,
+      "loss": 0.7896,
+      "step": 5045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.035301699218286e-08,
+      "loss": 0.4834,
+      "step": 5046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.979487518905645e-08,
+      "loss": 0.7119,
+      "step": 5047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.924061258075096e-08,
+      "loss": 0.7734,
+      "step": 5048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.869022938313593e-08,
+      "loss": 0.8105,
+      "step": 5049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.814372581056991e-08,
+      "loss": 0.7788,
+      "step": 5050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7601102075898175e-08,
+      "loss": 0.7988,
+      "step": 5051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7062358390455024e-08,
+      "loss": 0.8203,
+      "step": 5052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6527494964067045e-08,
+      "loss": 0.7842,
+      "step": 5053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.599651200504539e-08,
+      "loss": 0.7642,
+      "step": 5054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5469409720191305e-08,
+      "loss": 0.8115,
+      "step": 5055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.494618831479612e-08,
+      "loss": 0.7681,
+      "step": 5056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4426847992639064e-08,
+      "loss": 0.7925,
+      "step": 5057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391138895598611e-08,
+      "loss": 0.7544,
+      "step": 5058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.339981140559223e-08,
+      "loss": 0.7627,
+      "step": 5059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2892115540703596e-08,
+      "loss": 0.8315,
+      "step": 5060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2388301559048705e-08,
+      "loss": 0.748,
+      "step": 5061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.188836965685171e-08,
+      "loss": 0.8091,
+      "step": 5062
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.139232002881687e-08,
+      "loss": 0.7871,
+      "step": 5063
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.090015286814296e-08,
+      "loss": 0.8032,
+      "step": 5064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.041186836651333e-08,
+      "loss": 0.8477,
+      "step": 5065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9927466714100295e-08,
+      "loss": 0.7905,
+      "step": 5066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9446948099561833e-08,
+      "loss": 0.8096,
+      "step": 5067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8970312710047134e-08,
+      "loss": 0.7378,
+      "step": 5068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.849756073118881e-08,
+      "loss": 0.8057,
+      "step": 5069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.802869234711181e-08,
+      "loss": 0.7944,
+      "step": 5070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7563707740424494e-08,
+      "loss": 0.7822,
+      "step": 5071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7102607092225343e-08,
+      "loss": 0.7168,
+      "step": 5072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6645390582097363e-08,
+      "loss": 0.4351,
+      "step": 5073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6192058388113672e-08,
+      "loss": 0.8066,
+      "step": 5074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5742610686831927e-08,
+      "loss": 0.478,
+      "step": 5075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5297047653298767e-08,
+      "loss": 0.7744,
+      "step": 5076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4855369461048717e-08,
+      "loss": 0.8188,
+      "step": 5077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4417576282099732e-08,
+      "loss": 0.7251,
+      "step": 5078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.398366828696097e-08,
+      "loss": 0.7329,
+      "step": 5079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3553645644625034e-08,
+      "loss": 0.832,
+      "step": 5080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3127508522572397e-08,
+      "loss": 0.7666,
+      "step": 5081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2705257086771405e-08,
+      "loss": 0.8267,
+      "step": 5082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2286891501674958e-08,
+      "loss": 0.7671,
+      "step": 5083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1872411930223826e-08,
+      "loss": 0.7764,
+      "step": 5084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1461818533846656e-08,
+      "loss": 0.8169,
+      "step": 5085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1055111472454425e-08,
+      "loss": 0.7935,
+      "step": 5086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0652290904449312e-08,
+      "loss": 0.7583,
+      "step": 5087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0253356986716933e-08,
+      "loss": 0.7759,
+      "step": 5088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9858309874628557e-08,
+      "loss": 0.5156,
+      "step": 5089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9467149722044444e-08,
+      "loss": 0.8594,
+      "step": 5090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9079876681309396e-08,
+      "loss": 0.7979,
+      "step": 5091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.869649090325387e-08,
+      "loss": 0.7754,
+      "step": 5092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.83169925371951e-08,
+      "loss": 0.8042,
+      "step": 5093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.794138173093596e-08,
+      "loss": 0.772,
+      "step": 5094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7569658630764986e-08,
+      "loss": 0.8291,
+      "step": 5095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7201823381457483e-08,
+      "loss": 0.7485,
+      "step": 5096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.683787612627441e-08,
+      "loss": 0.8286,
+      "step": 5097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6477817006961272e-08,
+      "loss": 0.7979,
+      "step": 5098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.612164616375034e-08,
+      "loss": 0.7969,
+      "step": 5099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576936373535953e-08,
+      "loss": 0.8018,
+      "step": 5100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.542096985899133e-08,
+      "loss": 0.7441,
+      "step": 5101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5076464670334967e-08,
+      "loss": 0.7681,
+      "step": 5102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4735848303565337e-08,
+      "loss": 0.8381,
+      "step": 5103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4399120891340768e-08,
+      "loss": 0.7886,
+      "step": 5104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4066282564807466e-08,
+      "loss": 0.8374,
+      "step": 5105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.373733345359396e-08,
+      "loss": 0.7964,
+      "step": 5106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3412273685818878e-08,
+      "loss": 0.7104,
+      "step": 5107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.309110338807984e-08,
+      "loss": 0.4502,
+      "step": 5108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2773822685464566e-08,
+      "loss": 0.7832,
+      "step": 5109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2460431701544206e-08,
+      "loss": 0.7329,
+      "step": 5110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2150930558374463e-08,
+      "loss": 0.8164,
+      "step": 5111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1845319376496689e-08,
+      "loss": 0.7676,
+      "step": 5112
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1543598274936785e-08,
+      "loss": 0.7524,
+      "step": 5113
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1245767371206307e-08,
+      "loss": 0.7212,
+      "step": 5114
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0951826781301356e-08,
+      "loss": 0.7349,
+      "step": 5115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.066177661970258e-08,
+      "loss": 0.7568,
+      "step": 5116
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0375616999375171e-08,
+      "loss": 0.7646,
+      "step": 5117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0093348031769979e-08,
+      "loss": 0.8306,
+      "step": 5118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.814969826822396e-09,
+      "loss": 0.8452,
+      "step": 5119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.540482492952363e-09,
+      "loss": 0.7954,
+      "step": 5120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.269886137063255e-09,
+      "loss": 0.4897,
+      "step": 5121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.003180864545214e-09,
+      "loss": 0.7856,
+      "step": 5122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.740366779270704e-09,
+      "loss": 0.8276,
+      "step": 5123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.481443983598958e-09,
+      "loss": 0.8076,
+      "step": 5124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.226412578371534e-09,
+      "loss": 0.811,
+      "step": 5125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.975272662915645e-09,
+      "loss": 0.4065,
+      "step": 5126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.72802433504194e-09,
+      "loss": 0.7944,
+      "step": 5127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.484667691047832e-09,
+      "loss": 0.7764,
+      "step": 5128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.2452028257108396e-09,
+      "loss": 0.7593,
+      "step": 5129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.009629832297471e-09,
+      "loss": 0.7729,
+      "step": 5130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.777948802554335e-09,
+      "loss": 0.7666,
+      "step": 5131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.550159826715918e-09,
+      "loss": 0.4541,
+      "step": 5132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.326262993497922e-09,
+      "loss": 0.9272,
+      "step": 5133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.106258390101704e-09,
+      "loss": 0.4468,
+      "step": 5134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8901461022120575e-09,
+      "loss": 0.7412,
+      "step": 5135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.677926213997209e-09,
+      "loss": 0.8232,
+      "step": 5136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.4695988081121535e-09,
+      "loss": 0.7969,
+      "step": 5137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.265163965693099e-09,
+      "loss": 0.8076,
+      "step": 5138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0646217663608e-09,
+      "loss": 0.7725,
+      "step": 5139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.867972288221667e-09,
+      "loss": 0.8022,
+      "step": 5140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.675215607862216e-09,
+      "loss": 0.7744,
+      "step": 5141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.486351800357947e-09,
+      "loss": 0.7871,
+      "step": 5142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.301380939263355e-09,
+      "loss": 0.4497,
+      "step": 5143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.120303096620815e-09,
+      "loss": 0.7686,
+      "step": 5144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.943118342952801e-09,
+      "loss": 0.7939,
+      "step": 5145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.769826747268557e-09,
+      "loss": 0.791,
+      "step": 5146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6004283770596504e-09,
+      "loss": 0.8145,
+      "step": 5147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.434923298302195e-09,
+      "loss": 0.7534,
+      "step": 5148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2733115754535194e-09,
+      "loss": 0.79,
+      "step": 5149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1155932714577175e-09,
+      "loss": 0.7935,
+      "step": 5150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.961768447741209e-09,
+      "loss": 0.5356,
+      "step": 5151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8118371642138485e-09,
+      "loss": 0.79,
+      "step": 5152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.665799479268927e-09,
+      "loss": 0.7485,
+      "step": 5153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5236554497842793e-09,
+      "loss": 0.8335,
+      "step": 5154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3854051311200666e-09,
+      "loss": 0.7954,
+      "step": 5155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2510485771209957e-09,
+      "loss": 0.7925,
+      "step": 5156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.120585840115208e-09,
+      "loss": 0.8237,
+      "step": 5157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9940169709131706e-09,
+      "loss": 0.7896,
+      "step": 5158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.871342018809896e-09,
+      "loss": 0.8042,
+      "step": 5159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7525610315827225e-09,
+      "loss": 0.771,
+      "step": 5160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6376740554946424e-09,
+      "loss": 0.835,
+      "step": 5161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5266811352898647e-09,
+      "loss": 0.7886,
+      "step": 5162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4195823141971433e-09,
+      "loss": 0.4966,
+      "step": 5163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3163776339275569e-09,
+      "loss": 0.8003,
+      "step": 5164
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2170671346756203e-09,
+      "loss": 0.8232,
+      "step": 5165
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1216508551215032e-09,
+      "loss": 0.7261,
+      "step": 5166
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.030128832425481e-09,
+      "loss": 0.811,
+      "step": 5167
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.42501102233484e-10,
+      "loss": 0.7075,
+      "step": 5168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.587676986737681e-10,
+      "loss": 0.7803,
+      "step": 5169
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.789286543569141e-10,
+      "loss": 0.8022,
+      "step": 5170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.029840003791588e-10,
+      "loss": 0.8218,
+      "step": 5171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.309337663168435e-10,
+      "loss": 0.7656,
+      "step": 5172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.627779802330757e-10,
+      "loss": 0.7593,
+      "step": 5173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.985166686721776e-10,
+      "loss": 0.7129,
+      "step": 5174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.381498566607967e-10,
+      "loss": 0.4875,
+      "step": 5175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.816775677112361e-10,
+      "loss": 0.7456,
+      "step": 5176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2909982381590375e-10,
+      "loss": 0.77,
+      "step": 5177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8041664545508385e-10,
+      "loss": 0.7568,
+      "step": 5178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.356280515869447e-10,
+      "loss": 0.7983,
+      "step": 5179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9473405965642066e-10,
+      "loss": 0.7925,
+      "step": 5180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5773468559077132e-10,
+      "loss": 0.7812,
+      "step": 5181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2462994379958125e-10,
+      "loss": 0.8198,
+      "step": 5182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.541984717476027e-11,
+      "loss": 0.8257,
+      "step": 5183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.010440709498412e-11,
+      "loss": 0.7969,
+      "step": 5184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.868363341903326e-11,
+      "loss": 0.4014,
+      "step": 5185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1157534489123506e-11,
+      "loss": 0.8013,
+      "step": 5186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.752611713201624e-11,
+      "loss": 0.4514,
+      "step": 5187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.789386655687737e-12,
+      "loss": 0.7734,
+      "step": 5188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.947346853770071e-12,
+      "loss": 0.6841,
+      "step": 5189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.511,
+      "step": 5190
+    },
+    {
+      "epoch": 1.0,
+      "step": 5190,
+      "total_flos": 1746566985744384.0,
+      "train_loss": 0.8219152902592124,
+      "train_runtime": 37703.9628,
+      "train_samples_per_second": 17.619,
+      "train_steps_per_second": 0.138
+    }
+  ],
+  "max_steps": 5190,
+  "num_train_epochs": 1,
+  "total_flos": 1746566985744384.0,
+  "trial_name": null,
+  "trial_params": null
+}