diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,124776 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 20791,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.41025641025641e-08,
+      "loss": 2.9297,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-07,
+      "loss": 3.0078,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-07,
+      "loss": 3.1641,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-07,
+      "loss": 3.0977,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.205128205128205e-07,
+      "loss": 2.9766,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 2.6445,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4871794871794876e-07,
+      "loss": 2.8594,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 2.6953,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.76923076923077e-07,
+      "loss": 3.0938,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.41025641025641e-07,
+      "loss": 2.7227,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.051282051282052e-07,
+      "loss": 2.8672,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 2.0801,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 3.0078,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.974358974358975e-07,
+      "loss": 3.0547,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.615384615384617e-07,
+      "loss": 3.0039,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 2.6406,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0897435897435899e-06,
+      "loss": 2.6562,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 2.7656,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.217948717948718e-06,
+      "loss": 2.7773,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 2.7617,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3461538461538462e-06,
+      "loss": 2.8477,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4102564102564104e-06,
+      "loss": 1.8809,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4743589743589745e-06,
+      "loss": 2.8164,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 2.6602,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.602564102564103e-06,
+      "loss": 2.375,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 2.2109,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7307692307692308e-06,
+      "loss": 2.4844,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.794871794871795e-06,
+      "loss": 2.25,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8589743589743592e-06,
+      "loss": 2.2734,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 2.1562,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.987179487179487e-06,
+      "loss": 2.0723,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 2.2305,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1153846153846155e-06,
+      "loss": 2.1191,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1794871794871797e-06,
+      "loss": 2.0918,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.243589743589744e-06,
+      "loss": 1.9453,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.8535,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.371794871794872e-06,
+      "loss": 1.8848,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.435897435897436e-06,
+      "loss": 1.7715,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.8359,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 1.8125,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6282051282051286e-06,
+      "loss": 1.8105,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 1.709,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.756410256410257e-06,
+      "loss": 1.8281,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8205128205128207e-06,
+      "loss": 1.8438,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8846153846153845e-06,
+      "loss": 1.8086,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.948717948717949e-06,
+      "loss": 1.6973,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.012820512820513e-06,
+      "loss": 1.6348,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.5566,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.141025641025641e-06,
+      "loss": 1.6992,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.205128205128206e-06,
+      "loss": 1.5918,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2692307692307696e-06,
+      "loss": 1.7285,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.7031,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.397435897435898e-06,
+      "loss": 1.7402,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 1.6875,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5256410256410263e-06,
+      "loss": 1.4248,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 1.627,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.653846153846154e-06,
+      "loss": 1.5605,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7179487179487184e-06,
+      "loss": 1.5898,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.782051282051282e-06,
+      "loss": 1.6953,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.7812,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.910256410256411e-06,
+      "loss": 1.5547,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.974358974358974e-06,
+      "loss": 1.4863,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0384615384615385e-06,
+      "loss": 1.6602,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 1.6133,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.5977,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 1.5898,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.294871794871795e-06,
+      "loss": 1.6289,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.358974358974359e-06,
+      "loss": 1.4316,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.423076923076924e-06,
+      "loss": 1.4902,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.487179487179488e-06,
+      "loss": 1.4883,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.551282051282052e-06,
+      "loss": 1.5039,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.6211,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6794871794871795e-06,
+      "loss": 1.5098,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.743589743589744e-06,
+      "loss": 1.4824,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.807692307692308e-06,
+      "loss": 1.2852,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 1.4805,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.935897435897436e-06,
+      "loss": 1.2988,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 1.4961,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.064102564102565e-06,
+      "loss": 1.7812,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 1.2754,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.192307692307693e-06,
+      "loss": 1.5039,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.256410256410257e-06,
+      "loss": 1.2344,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.320512820512821e-06,
+      "loss": 1.5898,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 1.3281,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.448717948717949e-06,
+      "loss": 1.21,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.512820512820514e-06,
+      "loss": 1.3789,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.576923076923077e-06,
+      "loss": 1.5586,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 1.4453,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.705128205128206e-06,
+      "loss": 1.4297,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 1.5508,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 1.3438,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.897435897435898e-06,
+      "loss": 1.3789,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.961538461538462e-06,
+      "loss": 1.6211,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.025641025641026e-06,
+      "loss": 1.5195,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.08974358974359e-06,
+      "loss": 1.4805,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.248,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.217948717948718e-06,
+      "loss": 1.4082,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.282051282051282e-06,
+      "loss": 1.334,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.3461538461538466e-06,
+      "loss": 1.5078,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.410256410256412e-06,
+      "loss": 1.2109,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.474358974358975e-06,
+      "loss": 1.3418,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 1.3145,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.602564102564103e-06,
+      "loss": 1.377,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.3711,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.730769230769232e-06,
+      "loss": 1.5859,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.794871794871796e-06,
+      "loss": 1.3066,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.858974358974359e-06,
+      "loss": 1.3496,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 1.4023,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.9871794871794876e-06,
+      "loss": 1.4023,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.051282051282053e-06,
+      "loss": 1.4473,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.115384615384616e-06,
+      "loss": 1.2988,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.17948717948718e-06,
+      "loss": 1.4121,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.243589743589744e-06,
+      "loss": 1.4297,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 1.2559,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.371794871794873e-06,
+      "loss": 1.1514,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.435897435897437e-06,
+      "loss": 1.5469,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.1758,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.564102564102564e-06,
+      "loss": 1.3496,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.6282051282051286e-06,
+      "loss": 1.3125,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.4512,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.756410256410258e-06,
+      "loss": 1.373,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.820512820512822e-06,
+      "loss": 1.4277,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.884615384615384e-06,
+      "loss": 1.3828,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.948717948717949e-06,
+      "loss": 1.2637,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.012820512820515e-06,
+      "loss": 1.4492,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 1.3223,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.141025641025641e-06,
+      "loss": 1.4004,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 1.4648,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.26923076923077e-06,
+      "loss": 1.1699,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.2295,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.397435897435898e-06,
+      "loss": 1.3203,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 1.5098,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.525641025641026e-06,
+      "loss": 1.1055,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.58974358974359e-06,
+      "loss": 1.3105,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.653846153846155e-06,
+      "loss": 1.1777,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 1.3887,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.782051282051283e-06,
+      "loss": 1.3535,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 1.2559,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.910256410256411e-06,
+      "loss": 1.5234,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 1.1973,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.03846153846154e-06,
+      "loss": 1.3613,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.102564102564104e-06,
+      "loss": 1.3203,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 0.9912,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.2842,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.294871794871796e-06,
+      "loss": 1.0244,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.358974358974359e-06,
+      "loss": 1.3164,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.423076923076923e-06,
+      "loss": 1.3008,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.487179487179487e-06,
+      "loss": 1.5488,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.551282051282053e-06,
+      "loss": 1.1641,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 1.2578,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.67948717948718e-06,
+      "loss": 1.3086,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 1.2871,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.807692307692308e-06,
+      "loss": 1.2422,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.871794871794872e-06,
+      "loss": 1.084,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.935897435897437e-06,
+      "loss": 1.207,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 1.332,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0064102564102565e-05,
+      "loss": 1.2793,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.012820512820513e-05,
+      "loss": 1.4785,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0192307692307692e-05,
+      "loss": 1.1748,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 1.3457,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0320512820512822e-05,
+      "loss": 1.3477,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 1.2852,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.044871794871795e-05,
+      "loss": 1.0762,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0512820512820514e-05,
+      "loss": 1.1309,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0576923076923078e-05,
+      "loss": 1.1758,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0641025641025643e-05,
+      "loss": 1.1992,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0705128205128205e-05,
+      "loss": 1.4023,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 1.1797,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 1.4336,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0897435897435898e-05,
+      "loss": 1.4199,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0961538461538464e-05,
+      "loss": 1.2988,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1025641025641028e-05,
+      "loss": 1.4355,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1089743589743592e-05,
+      "loss": 1.248,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 1.2227,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1217948717948719e-05,
+      "loss": 1.2148,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 1.1621,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1346153846153847e-05,
+      "loss": 1.4629,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1410256410256411e-05,
+      "loss": 1.1387,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1474358974358974e-05,
+      "loss": 1.2051,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 1.0859,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1602564102564104e-05,
+      "loss": 1.2734,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 1.0752,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1730769230769232e-05,
+      "loss": 1.2637,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1794871794871796e-05,
+      "loss": 1.3398,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.185897435897436e-05,
+      "loss": 1.2852,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 1.4238,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1987179487179487e-05,
+      "loss": 1.3359,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2051282051282051e-05,
+      "loss": 1.3809,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2115384615384615e-05,
+      "loss": 1.3613,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.217948717948718e-05,
+      "loss": 1.3203,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2243589743589746e-05,
+      "loss": 1.1816,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 1.1934,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2371794871794874e-05,
+      "loss": 1.3027,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2435897435897436e-05,
+      "loss": 1.1084,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.25e-05,
+      "loss": 1.0996,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2564102564102565e-05,
+      "loss": 1.1406,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2628205128205129e-05,
+      "loss": 1.4414,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 1.293,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2756410256410257e-05,
+      "loss": 1.125,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2820512820512823e-05,
+      "loss": 1.3027,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2884615384615386e-05,
+      "loss": 0.9961,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.294871794871795e-05,
+      "loss": 1.2002,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3012820512820514e-05,
+      "loss": 1.3359,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 1.2832,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3141025641025642e-05,
+      "loss": 1.125,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3205128205128207e-05,
+      "loss": 1.3633,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3269230769230769e-05,
+      "loss": 1.248,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.3496,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3397435897435897e-05,
+      "loss": 1.2871,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3461538461538463e-05,
+      "loss": 1.2344,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3525641025641028e-05,
+      "loss": 1.2324,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3589743589743592e-05,
+      "loss": 1.4199,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3653846153846156e-05,
+      "loss": 1.2031,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3717948717948718e-05,
+      "loss": 1.2705,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3782051282051283e-05,
+      "loss": 1.127,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.3164,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3910256410256411e-05,
+      "loss": 1.2617,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3974358974358975e-05,
+      "loss": 1.3672,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.403846153846154e-05,
+      "loss": 1.1406,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4102564102564105e-05,
+      "loss": 1.1836,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.416666666666667e-05,
+      "loss": 1.3301,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4230769230769232e-05,
+      "loss": 1.3379,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4294871794871796e-05,
+      "loss": 1.2422,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.435897435897436e-05,
+      "loss": 1.2656,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4423076923076924e-05,
+      "loss": 1.1611,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4487179487179489e-05,
+      "loss": 1.2246,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4551282051282051e-05,
+      "loss": 1.0693,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 1.2852,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.467948717948718e-05,
+      "loss": 1.2617,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4743589743589745e-05,
+      "loss": 1.1055,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.480769230769231e-05,
+      "loss": 1.3359,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4871794871794874e-05,
+      "loss": 1.1992,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4935897435897438e-05,
+      "loss": 1.2715,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.207,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5064102564102565e-05,
+      "loss": 1.2129,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5128205128205129e-05,
+      "loss": 1.126,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5192307692307693e-05,
+      "loss": 1.3428,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5256410256410257e-05,
+      "loss": 1.1895,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5320512820512823e-05,
+      "loss": 1.1953,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.1543,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.544871794871795e-05,
+      "loss": 1.2598,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5512820512820516e-05,
+      "loss": 1.1338,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.557692307692308e-05,
+      "loss": 1.2168,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5641025641025644e-05,
+      "loss": 1.2637,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5705128205128205e-05,
+      "loss": 1.1504,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 1.1738,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 1.3516,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 1.2812,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5961538461538465e-05,
+      "loss": 1.252,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.602564102564103e-05,
+      "loss": 1.2344,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6089743589743593e-05,
+      "loss": 1.209,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 1.2402,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6217948717948718e-05,
+      "loss": 1.4004,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6282051282051282e-05,
+      "loss": 1.3555,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6346153846153847e-05,
+      "loss": 1.2559,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 1.166,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6474358974358975e-05,
+      "loss": 1.2656,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 1.291,
+      "step": 258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6602564102564103e-05,
+      "loss": 1.1045,
+      "step": 259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.1396,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.673076923076923e-05,
+      "loss": 1.1699,
+      "step": 261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6794871794871796e-05,
+      "loss": 1.0625,
+      "step": 262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.685897435897436e-05,
+      "loss": 1.1895,
+      "step": 263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 1.168,
+      "step": 264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.698717948717949e-05,
+      "loss": 1.2148,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7051282051282053e-05,
+      "loss": 1.2412,
+      "step": 266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7115384615384617e-05,
+      "loss": 1.2422,
+      "step": 267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.717948717948718e-05,
+      "loss": 1.2617,
+      "step": 268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7243589743589745e-05,
+      "loss": 1.0908,
+      "step": 269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 1.334,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7371794871794873e-05,
+      "loss": 1.2891,
+      "step": 271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7435897435897438e-05,
+      "loss": 1.124,
+      "step": 272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 1.2793,
+      "step": 273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7564102564102566e-05,
+      "loss": 1.2539,
+      "step": 274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.762820512820513e-05,
+      "loss": 1.2637,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 1.1895,
+      "step": 276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.775641025641026e-05,
+      "loss": 1.2832,
+      "step": 277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7820512820512823e-05,
+      "loss": 1.1562,
+      "step": 278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7884615384615387e-05,
+      "loss": 1.3672,
+      "step": 279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 1.2734,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8012820512820515e-05,
+      "loss": 1.3574,
+      "step": 281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.807692307692308e-05,
+      "loss": 1.2598,
+      "step": 282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8141025641025644e-05,
+      "loss": 1.3203,
+      "step": 283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8205128205128208e-05,
+      "loss": 1.002,
+      "step": 284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.826923076923077e-05,
+      "loss": 1.4648,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.9785,
+      "step": 286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8397435897435897e-05,
+      "loss": 1.2227,
+      "step": 287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.2715,
+      "step": 288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.852564102564103e-05,
+      "loss": 1.1084,
+      "step": 289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8589743589743593e-05,
+      "loss": 1.1045,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8653846153846157e-05,
+      "loss": 1.1221,
+      "step": 291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8717948717948718e-05,
+      "loss": 1.2754,
+      "step": 292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8782051282051282e-05,
+      "loss": 1.041,
+      "step": 293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 1.3574,
+      "step": 294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.891025641025641e-05,
+      "loss": 1.2227,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 1.207,
+      "step": 296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.903846153846154e-05,
+      "loss": 1.1328,
+      "step": 297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9102564102564106e-05,
+      "loss": 1.2129,
+      "step": 298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.916666666666667e-05,
+      "loss": 1.2451,
+      "step": 299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 1.0127,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9294871794871796e-05,
+      "loss": 1.1602,
+      "step": 301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.935897435897436e-05,
+      "loss": 1.3535,
+      "step": 302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9423076923076924e-05,
+      "loss": 1.2559,
+      "step": 303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 1.0859,
+      "step": 304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9551282051282052e-05,
+      "loss": 1.0879,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 1.3301,
+      "step": 306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.967948717948718e-05,
+      "loss": 1.0479,
+      "step": 307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9743589743589745e-05,
+      "loss": 1.085,
+      "step": 308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.980769230769231e-05,
+      "loss": 1.2041,
+      "step": 309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9871794871794873e-05,
+      "loss": 1.1484,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9935897435897437e-05,
+      "loss": 1.1133,
+      "step": 311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 1.166,
+      "step": 312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0064102564102566e-05,
+      "loss": 1.1084,
+      "step": 313
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.012820512820513e-05,
+      "loss": 1.2109,
+      "step": 314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0192307692307694e-05,
+      "loss": 1.0601,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.025641025641026e-05,
+      "loss": 1.2227,
+      "step": 316
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0320512820512823e-05,
+      "loss": 1.1367,
+      "step": 317
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0384615384615383e-05,
+      "loss": 1.1514,
+      "step": 318
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0448717948717948e-05,
+      "loss": 1.168,
+      "step": 319
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0512820512820512e-05,
+      "loss": 1.2852,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0576923076923076e-05,
+      "loss": 1.3535,
+      "step": 321
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0641025641025643e-05,
+      "loss": 1.2715,
+      "step": 322
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0705128205128208e-05,
+      "loss": 1.1328,
+      "step": 323
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0769230769230772e-05,
+      "loss": 1.2842,
+      "step": 324
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 1.2617,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.08974358974359e-05,
+      "loss": 1.1738,
+      "step": 326
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0961538461538464e-05,
+      "loss": 1.2773,
+      "step": 327
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.102564102564103e-05,
+      "loss": 1.127,
+      "step": 328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1089743589743593e-05,
+      "loss": 0.9961,
+      "step": 329
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1153846153846157e-05,
+      "loss": 1.0557,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.121794871794872e-05,
+      "loss": 1.1895,
+      "step": 331
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1282051282051285e-05,
+      "loss": 1.3672,
+      "step": 332
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1346153846153846e-05,
+      "loss": 1.2852,
+      "step": 333
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.141025641025641e-05,
+      "loss": 1.334,
+      "step": 334
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1474358974358974e-05,
+      "loss": 1.1621,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.153846153846154e-05,
+      "loss": 1.2031,
+      "step": 336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1602564102564103e-05,
+      "loss": 1.1738,
+      "step": 337
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1666666666666667e-05,
+      "loss": 1.1367,
+      "step": 338
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.173076923076923e-05,
+      "loss": 1.2793,
+      "step": 339
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1794871794871795e-05,
+      "loss": 1.3086,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.185897435897436e-05,
+      "loss": 1.375,
+      "step": 341
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1923076923076927e-05,
+      "loss": 1.2832,
+      "step": 342
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.198717948717949e-05,
+      "loss": 1.2676,
+      "step": 343
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2051282051282056e-05,
+      "loss": 1.3281,
+      "step": 344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.211538461538462e-05,
+      "loss": 1.0869,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2179487179487184e-05,
+      "loss": 1.2812,
+      "step": 346
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2243589743589745e-05,
+      "loss": 1.3047,
+      "step": 347
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.230769230769231e-05,
+      "loss": 1.2129,
+      "step": 348
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2371794871794873e-05,
+      "loss": 1.2246,
+      "step": 349
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2435897435897437e-05,
+      "loss": 1.2363,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.25e-05,
+      "loss": 1.2168,
+      "step": 351
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2564102564102566e-05,
+      "loss": 1.1777,
+      "step": 352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.262820512820513e-05,
+      "loss": 1.1191,
+      "step": 353
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2692307692307694e-05,
+      "loss": 1.2109,
+      "step": 354
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2756410256410258e-05,
+      "loss": 1.0928,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2820512820512822e-05,
+      "loss": 1.0215,
+      "step": 356
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2884615384615387e-05,
+      "loss": 1.1436,
+      "step": 357
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2948717948717947e-05,
+      "loss": 1.2871,
+      "step": 358
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.301282051282051e-05,
+      "loss": 1.1865,
+      "step": 359
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3076923076923076e-05,
+      "loss": 1.2246,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3141025641025647e-05,
+      "loss": 1.0957,
+      "step": 361
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3205128205128207e-05,
+      "loss": 1.1279,
+      "step": 362
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.326923076923077e-05,
+      "loss": 1.1816,
+      "step": 363
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 1.1797,
+      "step": 364
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.33974358974359e-05,
+      "loss": 0.9668,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3461538461538464e-05,
+      "loss": 1.3047,
+      "step": 366
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.352564102564103e-05,
+      "loss": 1.1348,
+      "step": 367
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3589743589743593e-05,
+      "loss": 1.2754,
+      "step": 368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3653846153846157e-05,
+      "loss": 1.1934,
+      "step": 369
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.371794871794872e-05,
+      "loss": 1.2295,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3782051282051285e-05,
+      "loss": 1.1777,
+      "step": 371
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.384615384615385e-05,
+      "loss": 1.2891,
+      "step": 372
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.391025641025641e-05,
+      "loss": 1.0703,
+      "step": 373
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3974358974358974e-05,
+      "loss": 1.1035,
+      "step": 374
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.403846153846154e-05,
+      "loss": 1.1719,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4102564102564103e-05,
+      "loss": 1.1523,
+      "step": 376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4166666666666667e-05,
+      "loss": 1.2188,
+      "step": 377
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.423076923076923e-05,
+      "loss": 1.1191,
+      "step": 378
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4294871794871795e-05,
+      "loss": 1.2441,
+      "step": 379
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.435897435897436e-05,
+      "loss": 1.2012,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4423076923076927e-05,
+      "loss": 1.248,
+      "step": 381
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.448717948717949e-05,
+      "loss": 1.1719,
+      "step": 382
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4551282051282055e-05,
+      "loss": 1.165,
+      "step": 383
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.461538461538462e-05,
+      "loss": 1.207,
+      "step": 384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4679487179487184e-05,
+      "loss": 1.0977,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4743589743589748e-05,
+      "loss": 1.0098,
+      "step": 386
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4807692307692312e-05,
+      "loss": 1.2324,
+      "step": 387
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4871794871794873e-05,
+      "loss": 0.8877,
+      "step": 388
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4935897435897437e-05,
+      "loss": 1.25,
+      "step": 389
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5e-05,
+      "loss": 1.1279,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5064102564102565e-05,
+      "loss": 1.2305,
+      "step": 391
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.512820512820513e-05,
+      "loss": 0.9189,
+      "step": 392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5192307692307694e-05,
+      "loss": 1.0996,
+      "step": 393
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5256410256410258e-05,
+      "loss": 1.1914,
+      "step": 394
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5320512820512822e-05,
+      "loss": 1.1914,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5384615384615386e-05,
+      "loss": 1.1348,
+      "step": 396
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.544871794871795e-05,
+      "loss": 1.168,
+      "step": 397
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5512820512820515e-05,
+      "loss": 1.3027,
+      "step": 398
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5576923076923075e-05,
+      "loss": 1.1748,
+      "step": 399
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5641025641025646e-05,
+      "loss": 1.1035,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.570512820512821e-05,
+      "loss": 1.0459,
+      "step": 401
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.576923076923077e-05,
+      "loss": 1.2227,
+      "step": 402
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5833333333333336e-05,
+      "loss": 1.0557,
+      "step": 403
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.58974358974359e-05,
+      "loss": 1.043,
+      "step": 404
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5961538461538464e-05,
+      "loss": 0.9912,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6025641025641028e-05,
+      "loss": 1.2031,
+      "step": 406
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6089743589743592e-05,
+      "loss": 0.9951,
+      "step": 407
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6153846153846157e-05,
+      "loss": 1.0869,
+      "step": 408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.621794871794872e-05,
+      "loss": 1.0879,
+      "step": 409
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6282051282051285e-05,
+      "loss": 1.2324,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.634615384615385e-05,
+      "loss": 1.1895,
+      "step": 411
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6410256410256413e-05,
+      "loss": 1.2422,
+      "step": 412
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6474358974358974e-05,
+      "loss": 1.1299,
+      "step": 413
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6538461538461538e-05,
+      "loss": 1.1777,
+      "step": 414
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6602564102564102e-05,
+      "loss": 1.2344,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 1.085,
+      "step": 416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.673076923076923e-05,
+      "loss": 1.2617,
+      "step": 417
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6794871794871795e-05,
+      "loss": 1.0605,
+      "step": 418
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.685897435897436e-05,
+      "loss": 1.1396,
+      "step": 419
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6923076923076927e-05,
+      "loss": 1.1943,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.698717948717949e-05,
+      "loss": 1.1895,
+      "step": 421
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7051282051282055e-05,
+      "loss": 1.2305,
+      "step": 422
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.711538461538462e-05,
+      "loss": 1.1143,
+      "step": 423
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7179487179487183e-05,
+      "loss": 1.1094,
+      "step": 424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7243589743589748e-05,
+      "loss": 1.2148,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7307692307692312e-05,
+      "loss": 1.0293,
+      "step": 426
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7371794871794876e-05,
+      "loss": 0.9648,
+      "step": 427
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7435897435897437e-05,
+      "loss": 1.2676,
+      "step": 428
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.75e-05,
+      "loss": 1.1689,
+      "step": 429
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7564102564102565e-05,
+      "loss": 1.1699,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.762820512820513e-05,
+      "loss": 1.0479,
+      "step": 431
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7692307692307694e-05,
+      "loss": 1.1338,
+      "step": 432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7756410256410258e-05,
+      "loss": 1.2041,
+      "step": 433
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7820512820512822e-05,
+      "loss": 1.0254,
+      "step": 434
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7884615384615386e-05,
+      "loss": 1.1211,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.794871794871795e-05,
+      "loss": 1.3223,
+      "step": 436
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8012820512820514e-05,
+      "loss": 1.2305,
+      "step": 437
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.807692307692308e-05,
+      "loss": 1.291,
+      "step": 438
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8141025641025646e-05,
+      "loss": 1.3633,
+      "step": 439
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.820512820512821e-05,
+      "loss": 1.168,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8269230769230775e-05,
+      "loss": 1.1777,
+      "step": 441
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.833333333333334e-05,
+      "loss": 1.1367,
+      "step": 442
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.83974358974359e-05,
+      "loss": 1.126,
+      "step": 443
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8461538461538464e-05,
+      "loss": 1.0,
+      "step": 444
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8525641025641028e-05,
+      "loss": 1.2109,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8589743589743592e-05,
+      "loss": 1.0332,
+      "step": 446
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8653846153846156e-05,
+      "loss": 1.3066,
+      "step": 447
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.871794871794872e-05,
+      "loss": 1.1289,
+      "step": 448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8782051282051285e-05,
+      "loss": 1.1289,
+      "step": 449
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.884615384615385e-05,
+      "loss": 1.0322,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8910256410256413e-05,
+      "loss": 1.1797,
+      "step": 451
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8974358974358977e-05,
+      "loss": 1.3047,
+      "step": 452
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.903846153846154e-05,
+      "loss": 1.1934,
+      "step": 453
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9102564102564102e-05,
+      "loss": 1.165,
+      "step": 454
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9166666666666666e-05,
+      "loss": 1.0264,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.923076923076923e-05,
+      "loss": 1.0947,
+      "step": 456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9294871794871795e-05,
+      "loss": 1.04,
+      "step": 457
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.935897435897436e-05,
+      "loss": 1.1865,
+      "step": 458
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9423076923076926e-05,
+      "loss": 1.2539,
+      "step": 459
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.948717948717949e-05,
+      "loss": 1.2305,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9551282051282055e-05,
+      "loss": 0.9727,
+      "step": 461
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.961538461538462e-05,
+      "loss": 1.2354,
+      "step": 462
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9679487179487183e-05,
+      "loss": 0.999,
+      "step": 463
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9743589743589747e-05,
+      "loss": 0.8911,
+      "step": 464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.980769230769231e-05,
+      "loss": 1.0264,
+      "step": 465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9871794871794876e-05,
+      "loss": 1.1396,
+      "step": 466
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.993589743589744e-05,
+      "loss": 0.9443,
+      "step": 467
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 1.1465,
+      "step": 468
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0064102564102565e-05,
+      "loss": 1.125,
+      "step": 469
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.012820512820513e-05,
+      "loss": 1.3438,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0192307692307693e-05,
+      "loss": 1.0537,
+      "step": 471
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0256410256410257e-05,
+      "loss": 1.3691,
+      "step": 472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.032051282051282e-05,
+      "loss": 1.3359,
+      "step": 473
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0384615384615386e-05,
+      "loss": 0.9971,
+      "step": 474
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.044871794871795e-05,
+      "loss": 1.252,
+      "step": 475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0512820512820514e-05,
+      "loss": 1.3574,
+      "step": 476
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.057692307692308e-05,
+      "loss": 1.1416,
+      "step": 477
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0641025641025646e-05,
+      "loss": 1.2793,
+      "step": 478
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.070512820512821e-05,
+      "loss": 1.0996,
+      "step": 479
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 1.2773,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0833333333333335e-05,
+      "loss": 1.1709,
+      "step": 481
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.08974358974359e-05,
+      "loss": 1.0586,
+      "step": 482
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0961538461538464e-05,
+      "loss": 1.3594,
+      "step": 483
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.102564102564103e-05,
+      "loss": 1.1191,
+      "step": 484
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.108974358974359e-05,
+      "loss": 1.0205,
+      "step": 485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.115384615384616e-05,
+      "loss": 1.0762,
+      "step": 486
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.121794871794872e-05,
+      "loss": 1.3457,
+      "step": 487
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.128205128205129e-05,
+      "loss": 1.1348,
+      "step": 488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.134615384615385e-05,
+      "loss": 1.25,
+      "step": 489
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.141025641025641e-05,
+      "loss": 1.1074,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.147435897435898e-05,
+      "loss": 1.1289,
+      "step": 491
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.153846153846154e-05,
+      "loss": 1.1523,
+      "step": 492
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1602564102564105e-05,
+      "loss": 1.1553,
+      "step": 493
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1666666666666666e-05,
+      "loss": 1.0605,
+      "step": 494
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1730769230769234e-05,
+      "loss": 1.2598,
+      "step": 495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1794871794871795e-05,
+      "loss": 1.2617,
+      "step": 496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.185897435897436e-05,
+      "loss": 1.2969,
+      "step": 497
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.192307692307693e-05,
+      "loss": 1.1816,
+      "step": 498
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.198717948717949e-05,
+      "loss": 1.1836,
+      "step": 499
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.205128205128206e-05,
+      "loss": 1.2715,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.211538461538462e-05,
+      "loss": 1.1143,
+      "step": 501
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2179487179487186e-05,
+      "loss": 0.9658,
+      "step": 502
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.224358974358975e-05,
+      "loss": 1.3047,
+      "step": 503
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.230769230769231e-05,
+      "loss": 1.21,
+      "step": 504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2371794871794876e-05,
+      "loss": 1.0293,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2435897435897436e-05,
+      "loss": 0.9873,
+      "step": 506
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 1.2715,
+      "step": 507
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2564102564102565e-05,
+      "loss": 0.8652,
+      "step": 508
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.262820512820513e-05,
+      "loss": 0.9551,
+      "step": 509
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.269230769230769e-05,
+      "loss": 1.168,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.275641025641026e-05,
+      "loss": 1.1582,
+      "step": 511
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.282051282051282e-05,
+      "loss": 1.1025,
+      "step": 512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.288461538461539e-05,
+      "loss": 1.1445,
+      "step": 513
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.294871794871795e-05,
+      "loss": 1.0527,
+      "step": 514
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.301282051282051e-05,
+      "loss": 1.1621,
+      "step": 515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.307692307692308e-05,
+      "loss": 1.1143,
+      "step": 516
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3141025641025646e-05,
+      "loss": 1.0635,
+      "step": 517
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3205128205128207e-05,
+      "loss": 1.1367,
+      "step": 518
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3269230769230774e-05,
+      "loss": 1.1504,
+      "step": 519
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 1.1875,
+      "step": 520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.33974358974359e-05,
+      "loss": 1.0586,
+      "step": 521
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.346153846153846e-05,
+      "loss": 1.0547,
+      "step": 522
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.352564102564103e-05,
+      "loss": 1.1768,
+      "step": 523
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.358974358974359e-05,
+      "loss": 1.3242,
+      "step": 524
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.365384615384616e-05,
+      "loss": 1.0703,
+      "step": 525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.371794871794872e-05,
+      "loss": 0.9922,
+      "step": 526
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.378205128205129e-05,
+      "loss": 1.1973,
+      "step": 527
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.384615384615385e-05,
+      "loss": 1.3242,
+      "step": 528
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.391025641025641e-05,
+      "loss": 1.0801,
+      "step": 529
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.397435897435898e-05,
+      "loss": 1.0605,
+      "step": 530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.403846153846154e-05,
+      "loss": 1.1074,
+      "step": 531
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4102564102564105e-05,
+      "loss": 1.2168,
+      "step": 532
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4166666666666666e-05,
+      "loss": 1.2422,
+      "step": 533
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4230769230769234e-05,
+      "loss": 1.1426,
+      "step": 534
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4294871794871794e-05,
+      "loss": 1.1895,
+      "step": 535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.435897435897436e-05,
+      "loss": 1.0889,
+      "step": 536
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.442307692307693e-05,
+      "loss": 0.8662,
+      "step": 537
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.448717948717949e-05,
+      "loss": 1.0273,
+      "step": 538
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.455128205128206e-05,
+      "loss": 1.0771,
+      "step": 539
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.461538461538462e-05,
+      "loss": 1.3047,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4679487179487186e-05,
+      "loss": 1.209,
+      "step": 541
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.474358974358975e-05,
+      "loss": 1.0928,
+      "step": 542
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4807692307692315e-05,
+      "loss": 1.1914,
+      "step": 543
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4871794871794875e-05,
+      "loss": 1.1719,
+      "step": 544
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4935897435897436e-05,
+      "loss": 1.0977,
+      "step": 545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5000000000000004e-05,
+      "loss": 1.3066,
+      "step": 546
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5064102564102565e-05,
+      "loss": 1.1445,
+      "step": 547
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.512820512820513e-05,
+      "loss": 1.373,
+      "step": 548
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.519230769230769e-05,
+      "loss": 1.2695,
+      "step": 549
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.525641025641026e-05,
+      "loss": 1.0479,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.532051282051282e-05,
+      "loss": 1.0342,
+      "step": 551
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.538461538461539e-05,
+      "loss": 1.2109,
+      "step": 552
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.544871794871795e-05,
+      "loss": 1.0693,
+      "step": 553
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.551282051282052e-05,
+      "loss": 1.1797,
+      "step": 554
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.557692307692308e-05,
+      "loss": 1.2832,
+      "step": 555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5641025641025646e-05,
+      "loss": 1.2441,
+      "step": 556
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.570512820512821e-05,
+      "loss": 1.0791,
+      "step": 557
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5769230769230774e-05,
+      "loss": 1.2402,
+      "step": 558
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5833333333333335e-05,
+      "loss": 1.1064,
+      "step": 559
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.58974358974359e-05,
+      "loss": 1.0664,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.596153846153846e-05,
+      "loss": 1.04,
+      "step": 561
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.602564102564103e-05,
+      "loss": 1.2461,
+      "step": 562
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.608974358974359e-05,
+      "loss": 1.2227,
+      "step": 563
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.615384615384616e-05,
+      "loss": 1.2266,
+      "step": 564
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.621794871794872e-05,
+      "loss": 1.1602,
+      "step": 565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.628205128205129e-05,
+      "loss": 0.833,
+      "step": 566
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.634615384615385e-05,
+      "loss": 1.125,
+      "step": 567
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6410256410256416e-05,
+      "loss": 1.1133,
+      "step": 568
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6474358974358977e-05,
+      "loss": 0.8848,
+      "step": 569
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.653846153846154e-05,
+      "loss": 1.1992,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6602564102564105e-05,
+      "loss": 1.1973,
+      "step": 571
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 1.1699,
+      "step": 572
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.673076923076923e-05,
+      "loss": 1.1523,
+      "step": 573
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6794871794871794e-05,
+      "loss": 0.8877,
+      "step": 574
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.685897435897436e-05,
+      "loss": 0.9688,
+      "step": 575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.692307692307693e-05,
+      "loss": 1.1992,
+      "step": 576
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.698717948717949e-05,
+      "loss": 1.0635,
+      "step": 577
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.705128205128206e-05,
+      "loss": 1.1338,
+      "step": 578
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.711538461538462e-05,
+      "loss": 1.168,
+      "step": 579
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7179487179487186e-05,
+      "loss": 1.1484,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.724358974358975e-05,
+      "loss": 1.0195,
+      "step": 581
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7307692307692314e-05,
+      "loss": 1.1738,
+      "step": 582
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7371794871794875e-05,
+      "loss": 1.0732,
+      "step": 583
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7435897435897436e-05,
+      "loss": 1.1455,
+      "step": 584
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 1.1289,
+      "step": 585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7564102564102564e-05,
+      "loss": 1.1211,
+      "step": 586
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.762820512820513e-05,
+      "loss": 1.1133,
+      "step": 587
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.769230769230769e-05,
+      "loss": 1.0645,
+      "step": 588
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.775641025641026e-05,
+      "loss": 1.1934,
+      "step": 589
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.782051282051282e-05,
+      "loss": 1.123,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.788461538461539e-05,
+      "loss": 1.2578,
+      "step": 591
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.794871794871795e-05,
+      "loss": 1.125,
+      "step": 592
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.801282051282052e-05,
+      "loss": 1.0156,
+      "step": 593
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.807692307692308e-05,
+      "loss": 1.1816,
+      "step": 594
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8141025641025645e-05,
+      "loss": 1.3301,
+      "step": 595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.820512820512821e-05,
+      "loss": 1.3135,
+      "step": 596
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8269230769230774e-05,
+      "loss": 1.2754,
+      "step": 597
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.833333333333334e-05,
+      "loss": 1.2051,
+      "step": 598
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.83974358974359e-05,
+      "loss": 1.2422,
+      "step": 599
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 1.1406,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.852564102564103e-05,
+      "loss": 1.2188,
+      "step": 601
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.858974358974359e-05,
+      "loss": 1.1699,
+      "step": 602
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.865384615384616e-05,
+      "loss": 1.1602,
+      "step": 603
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.871794871794872e-05,
+      "loss": 1.1445,
+      "step": 604
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.878205128205129e-05,
+      "loss": 0.9463,
+      "step": 605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.884615384615385e-05,
+      "loss": 1.1299,
+      "step": 606
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8910256410256416e-05,
+      "loss": 1.2695,
+      "step": 607
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8974358974358976e-05,
+      "loss": 1.3066,
+      "step": 608
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9038461538461544e-05,
+      "loss": 1.0957,
+      "step": 609
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9102564102564105e-05,
+      "loss": 1.1602,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9166666666666665e-05,
+      "loss": 1.3047,
+      "step": 611
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.923076923076923e-05,
+      "loss": 0.9863,
+      "step": 612
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9294871794871794e-05,
+      "loss": 1.1738,
+      "step": 613
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.935897435897436e-05,
+      "loss": 1.1953,
+      "step": 614
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.942307692307693e-05,
+      "loss": 1.0439,
+      "step": 615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.948717948717949e-05,
+      "loss": 1.0264,
+      "step": 616
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.955128205128206e-05,
+      "loss": 0.8604,
+      "step": 617
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.961538461538462e-05,
+      "loss": 1.084,
+      "step": 618
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9679487179487186e-05,
+      "loss": 1.1504,
+      "step": 619
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9743589743589747e-05,
+      "loss": 1.0557,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9807692307692314e-05,
+      "loss": 1.209,
+      "step": 621
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9871794871794875e-05,
+      "loss": 1.2227,
+      "step": 622
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.993589743589744e-05,
+      "loss": 1.1045,
+      "step": 623
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4e-05,
+      "loss": 1.2344,
+      "step": 624
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999999757329414e-05,
+      "loss": 1.0312,
+      "step": 625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999999902931764e-05,
+      "loss": 1.3281,
+      "step": 626
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999999781596472e-05,
+      "loss": 1.0801,
+      "step": 627
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999999611727067e-05,
+      "loss": 1.1953,
+      "step": 628
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999999393323552e-05,
+      "loss": 1.2188,
+      "step": 629
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999999126385934e-05,
+      "loss": 1.2578,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999998810914219e-05,
+      "loss": 1.1357,
+      "step": 631
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999984469084155e-05,
+      "loss": 1.0781,
+      "step": 632
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999998034368531e-05,
+      "loss": 1.3301,
+      "step": 633
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999975732945756e-05,
+      "loss": 1.2773,
+      "step": 634
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999997063686561e-05,
+      "loss": 1.0068,
+      "step": 635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999965055444995e-05,
+      "loss": 1.1006,
+      "step": 636
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999958988684044e-05,
+      "loss": 1.1816,
+      "step": 637
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999995243658291e-05,
+      "loss": 1.043,
+      "step": 638
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999994539914175e-05,
+      "loss": 1.252,
+      "step": 639
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999993787636073e-05,
+      "loss": 1.0449,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999992986824003e-05,
+      "loss": 1.1621,
+      "step": 641
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999992137477986e-05,
+      "loss": 1.1797,
+      "step": 642
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999991239598041e-05,
+      "loss": 1.0635,
+      "step": 643
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999902931841906e-05,
+      "loss": 1.166,
+      "step": 644
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999989298236457e-05,
+      "loss": 1.1299,
+      "step": 645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999882547548654e-05,
+      "loss": 1.2715,
+      "step": 646
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999987162739441e-05,
+      "loss": 1.3066,
+      "step": 647
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999986022190209e-05,
+      "loss": 1.0303,
+      "step": 648
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999984833107198e-05,
+      "loss": 1.1328,
+      "step": 649
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999835954904374e-05,
+      "loss": 1.1455,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999982309339957e-05,
+      "loss": 1.168,
+      "step": 651
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999809746557866e-05,
+      "loss": 1.252,
+      "step": 652
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999795914379604e-05,
+      "loss": 1.0674,
+      "step": 653
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999978159686511e-05,
+      "loss": 1.207,
+      "step": 654
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999976679401473e-05,
+      "loss": 0.8848,
+      "step": 655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999975150582884e-05,
+      "loss": 1.248,
+      "step": 656
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999973573230779e-05,
+      "loss": 1.1992,
+      "step": 657
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999971947345197e-05,
+      "loss": 1.2402,
+      "step": 658
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999970272926177e-05,
+      "loss": 0.916,
+      "step": 659
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999685499737616e-05,
+      "loss": 1.1055,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999667784879905e-05,
+      "loss": 0.9805,
+      "step": 661
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999964958468907e-05,
+      "loss": 0.9551,
+      "step": 662
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999963089916556e-05,
+      "loss": 1.2559,
+      "step": 663
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999961172830983e-05,
+      "loss": 1.0098,
+      "step": 664
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999959207212233e-05,
+      "loss": 1.2324,
+      "step": 665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999957193060355e-05,
+      "loss": 1.2676,
+      "step": 666
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999955130375398e-05,
+      "loss": 1.1377,
+      "step": 667
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999953019157411e-05,
+      "loss": 1.2324,
+      "step": 668
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999950859406446e-05,
+      "loss": 1.0381,
+      "step": 669
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999948651122556e-05,
+      "loss": 1.2314,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999946394305793e-05,
+      "loss": 1.0703,
+      "step": 671
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999944088956214e-05,
+      "loss": 1.1611,
+      "step": 672
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999941735073872e-05,
+      "loss": 1.0908,
+      "step": 673
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999939332658827e-05,
+      "loss": 1.042,
+      "step": 674
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999936881711135e-05,
+      "loss": 1.1025,
+      "step": 675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999934382230857e-05,
+      "loss": 1.0859,
+      "step": 676
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999318342180525e-05,
+      "loss": 1.082,
+      "step": 677
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999929237672785e-05,
+      "loss": 1.1045,
+      "step": 678
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999265925951155e-05,
+      "loss": 1.1338,
+      "step": 679
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.99992389898511e-05,
+      "loss": 1.1523,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999921156842832e-05,
+      "loss": 1.1387,
+      "step": 681
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.99991836616835e-05,
+      "loss": 1.1016,
+      "step": 682
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.99991552696173e-05,
+      "loss": 0.9736,
+      "step": 683
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999912639223042e-05,
+      "loss": 1.1387,
+      "step": 684
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999909702952357e-05,
+      "loss": 1.1133,
+      "step": 685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999906718149744e-05,
+      "loss": 1.0576,
+      "step": 686
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999903684815276e-05,
+      "loss": 1.1904,
+      "step": 687
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999006029490276e-05,
+      "loss": 1.2168,
+      "step": 688
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999897472551073e-05,
+      "loss": 1.0605,
+      "step": 689
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999894293621489e-05,
+      "loss": 1.0303,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9998910661603505e-05,
+      "loss": 1.1396,
+      "step": 691
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999887790167738e-05,
+      "loss": 1.248,
+      "step": 692
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.99988446564373e-05,
+      "loss": 1.1504,
+      "step": 693
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999881092588408e-05,
+      "loss": 1.1348,
+      "step": 694
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999877671001853e-05,
+      "loss": 1.0762,
+      "step": 695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999874200884149e-05,
+      "loss": 1.2109,
+      "step": 696
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9998706822353785e-05,
+      "loss": 1.1221,
+      "step": 697
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9998671150556286e-05,
+      "loss": 0.9004,
+      "step": 698
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999863499344985e-05,
+      "loss": 1.3223,
+      "step": 699
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999859835103536e-05,
+      "loss": 1.1406,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.99985612233137e-05,
+      "loss": 1.041,
+      "step": 701
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999852361028577e-05,
+      "loss": 0.9951,
+      "step": 702
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999848551195249e-05,
+      "loss": 1.0215,
+      "step": 703
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999844692831478e-05,
+      "loss": 1.2363,
+      "step": 704
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999840785937357e-05,
+      "loss": 1.1689,
+      "step": 705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9998368305129815e-05,
+      "loss": 1.2021,
+      "step": 706
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999832826558448e-05,
+      "loss": 1.1436,
+      "step": 707
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999828774073853e-05,
+      "loss": 1.043,
+      "step": 708
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9998246730592946e-05,
+      "loss": 1.1689,
+      "step": 709
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999820523514873e-05,
+      "loss": 1.2012,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999816325440688e-05,
+      "loss": 1.1016,
+      "step": 711
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999812078836843e-05,
+      "loss": 1.1309,
+      "step": 712
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.99980778370344e-05,
+      "loss": 1.2129,
+      "step": 713
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999803440040582e-05,
+      "loss": 1.1094,
+      "step": 714
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999799047848377e-05,
+      "loss": 0.9404,
+      "step": 715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9997946071269296e-05,
+      "loss": 1.0107,
+      "step": 716
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9997901178763486e-05,
+      "loss": 1.0596,
+      "step": 717
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999785580096743e-05,
+      "loss": 1.0781,
+      "step": 718
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9997809937882215e-05,
+      "loss": 0.9971,
+      "step": 719
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999776358950897e-05,
+      "loss": 1.2051,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9997716755848816e-05,
+      "loss": 1.1865,
+      "step": 721
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999766943690289e-05,
+      "loss": 0.9473,
+      "step": 722
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999762163267232e-05,
+      "loss": 1.1475,
+      "step": 723
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.99975733431583e-05,
+      "loss": 0.9707,
+      "step": 724
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999752456836199e-05,
+      "loss": 1.1895,
+      "step": 725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999747530828456e-05,
+      "loss": 1.0293,
+      "step": 726
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999742556292722e-05,
+      "loss": 1.0156,
+      "step": 727
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999737533229117e-05,
+      "loss": 1.1455,
+      "step": 728
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9997324616377625e-05,
+      "loss": 1.1641,
+      "step": 729
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9997273415187826e-05,
+      "loss": 1.0957,
+      "step": 730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999722172872301e-05,
+      "loss": 1.0439,
+      "step": 731
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999716955698444e-05,
+      "loss": 1.0791,
+      "step": 732
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999711689997337e-05,
+      "loss": 1.3281,
+      "step": 733
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999706375769107e-05,
+      "loss": 1.1836,
+      "step": 734
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999701013013886e-05,
+      "loss": 1.084,
+      "step": 735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999695601731802e-05,
+      "loss": 1.2715,
+      "step": 736
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9996901419229856e-05,
+      "loss": 0.9736,
+      "step": 737
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9996846335875705e-05,
+      "loss": 0.9854,
+      "step": 738
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999679076725691e-05,
+      "loss": 1.1846,
+      "step": 739
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999673471337481e-05,
+      "loss": 1.1416,
+      "step": 740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9996678174230756e-05,
+      "loss": 1.0332,
+      "step": 741
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999662114982615e-05,
+      "loss": 1.001,
+      "step": 742
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999656364016234e-05,
+      "loss": 1.2598,
+      "step": 743
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999650564524075e-05,
+      "loss": 1.042,
+      "step": 744
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999644716506277e-05,
+      "loss": 1.1953,
+      "step": 745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9996388199629825e-05,
+      "loss": 1.0156,
+      "step": 746
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9996328748943346e-05,
+      "loss": 1.1387,
+      "step": 747
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999626881300477e-05,
+      "loss": 1.0674,
+      "step": 748
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999620839181556e-05,
+      "loss": 1.1162,
+      "step": 749
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9996147485377184e-05,
+      "loss": 0.9521,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999608609369111e-05,
+      "loss": 0.8926,
+      "step": 751
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9996024216758835e-05,
+      "loss": 1.1562,
+      "step": 752
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9995961854581855e-05,
+      "loss": 1.1895,
+      "step": 753
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999589900716169e-05,
+      "loss": 1.1523,
+      "step": 754
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999583567449986e-05,
+      "loss": 1.1338,
+      "step": 755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99957718565979e-05,
+      "loss": 1.0459,
+      "step": 756
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999570755345738e-05,
+      "loss": 0.9961,
+      "step": 757
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999564276507983e-05,
+      "loss": 1.1787,
+      "step": 758
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999557749146683e-05,
+      "loss": 1.0645,
+      "step": 759
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9995511732619976e-05,
+      "loss": 1.2832,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999544548854086e-05,
+      "loss": 1.1475,
+      "step": 761
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999537875923108e-05,
+      "loss": 1.1318,
+      "step": 762
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9995311544692265e-05,
+      "loss": 1.0312,
+      "step": 763
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999524384492604e-05,
+      "loss": 0.916,
+      "step": 764
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999517565993405e-05,
+      "loss": 0.9453,
+      "step": 765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999510698971796e-05,
+      "loss": 1.1445,
+      "step": 766
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999503783427942e-05,
+      "loss": 1.0449,
+      "step": 767
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999496819362011e-05,
+      "loss": 1.2041,
+      "step": 768
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999489806774174e-05,
+      "loss": 1.2285,
+      "step": 769
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9994827456645985e-05,
+      "loss": 1.0684,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999475636033457e-05,
+      "loss": 1.2695,
+      "step": 771
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999468477880922e-05,
+      "loss": 1.0859,
+      "step": 772
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9994612712071675e-05,
+      "loss": 1.0605,
+      "step": 773
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9994540160123685e-05,
+      "loss": 1.1465,
+      "step": 774
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9994467122967006e-05,
+      "loss": 1.2461,
+      "step": 775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999439360060341e-05,
+      "loss": 1.1367,
+      "step": 776
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9994319593034685e-05,
+      "loss": 1.1621,
+      "step": 777
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999424510026262e-05,
+      "loss": 1.2188,
+      "step": 778
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999417012228903e-05,
+      "loss": 1.2617,
+      "step": 779
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9994094659115734e-05,
+      "loss": 1.1279,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999401871074455e-05,
+      "loss": 1.2148,
+      "step": 781
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999394227717735e-05,
+      "loss": 1.1895,
+      "step": 782
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9993865358415955e-05,
+      "loss": 1.2344,
+      "step": 783
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999378795446226e-05,
+      "loss": 0.9404,
+      "step": 784
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999371006531813e-05,
+      "loss": 1.1484,
+      "step": 785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9993631690985444e-05,
+      "loss": 1.207,
+      "step": 786
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9993552831466126e-05,
+      "loss": 1.209,
+      "step": 787
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9993473486762074e-05,
+      "loss": 1.2471,
+      "step": 788
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9993393656875226e-05,
+      "loss": 0.9746,
+      "step": 789
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9993313341807506e-05,
+      "loss": 0.894,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999323254156088e-05,
+      "loss": 1.1592,
+      "step": 791
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999315125613729e-05,
+      "loss": 1.1768,
+      "step": 792
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999306948553871e-05,
+      "loss": 1.1504,
+      "step": 793
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999298722976714e-05,
+      "loss": 1.0977,
+      "step": 794
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999290448882457e-05,
+      "loss": 1.1992,
+      "step": 795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999282126271299e-05,
+      "loss": 1.0693,
+      "step": 796
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9992737551434456e-05,
+      "loss": 1.2578,
+      "step": 797
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9992653354990965e-05,
+      "loss": 1.1211,
+      "step": 798
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999256867338457e-05,
+      "loss": 1.25,
+      "step": 799
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9992483506617345e-05,
+      "loss": 1.0127,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999239785469133e-05,
+      "loss": 1.2148,
+      "step": 801
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9992311717608613e-05,
+      "loss": 1.168,
+      "step": 802
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999222509537129e-05,
+      "loss": 0.9971,
+      "step": 803
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999213798798145e-05,
+      "loss": 1.2832,
+      "step": 804
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999205039544123e-05,
+      "loss": 1.1846,
+      "step": 805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999196231775274e-05,
+      "loss": 1.1104,
+      "step": 806
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9991873754918106e-05,
+      "loss": 1.2656,
+      "step": 807
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99917847069395e-05,
+      "loss": 1.0146,
+      "step": 808
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9991695173819074e-05,
+      "loss": 0.9873,
+      "step": 809
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9991605155558995e-05,
+      "loss": 1.124,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9991514652161454e-05,
+      "loss": 1.041,
+      "step": 811
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999142366362864e-05,
+      "loss": 1.0615,
+      "step": 812
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999133218996277e-05,
+      "loss": 1.168,
+      "step": 813
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9991240231166066e-05,
+      "loss": 0.9893,
+      "step": 814
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999114778724074e-05,
+      "loss": 1.0166,
+      "step": 815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999105485818907e-05,
+      "loss": 0.9678,
+      "step": 816
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999096144401327e-05,
+      "loss": 1.1641,
+      "step": 817
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999086754471564e-05,
+      "loss": 0.9883,
+      "step": 818
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999077316029844e-05,
+      "loss": 1.1143,
+      "step": 819
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9990678290763964e-05,
+      "loss": 1.0166,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999058293611452e-05,
+      "loss": 1.1982,
+      "step": 821
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999048709635242e-05,
+      "loss": 1.0518,
+      "step": 822
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999039077147998e-05,
+      "loss": 1.2012,
+      "step": 823
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999029396149955e-05,
+      "loss": 1.1826,
+      "step": 824
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999019666641348e-05,
+      "loss": 0.9502,
+      "step": 825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999009888622412e-05,
+      "loss": 1.1582,
+      "step": 826
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9990000620933855e-05,
+      "loss": 1.1885,
+      "step": 827
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998990187054506e-05,
+      "loss": 0.9678,
+      "step": 828
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998980263506013e-05,
+      "loss": 1.1875,
+      "step": 829
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998970291448148e-05,
+      "loss": 0.9697,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9989602708811536e-05,
+      "loss": 1.1865,
+      "step": 831
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998950201805271e-05,
+      "loss": 1.1094,
+      "step": 832
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998940084220747e-05,
+      "loss": 1.1426,
+      "step": 833
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998929918127825e-05,
+      "loss": 0.9229,
+      "step": 834
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998919703526752e-05,
+      "loss": 1.2158,
+      "step": 835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998909440417777e-05,
+      "loss": 1.2129,
+      "step": 836
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998899128801149e-05,
+      "loss": 1.1719,
+      "step": 837
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9988887686771166e-05,
+      "loss": 1.0791,
+      "step": 838
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998878360045933e-05,
+      "loss": 1.1553,
+      "step": 839
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99886790290785e-05,
+      "loss": 0.9961,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998857397263122e-05,
+      "loss": 1.1465,
+      "step": 841
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998846843112003e-05,
+      "loss": 1.1123,
+      "step": 842
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998836240454749e-05,
+      "loss": 1.0723,
+      "step": 843
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998825589291618e-05,
+      "loss": 1.2148,
+      "step": 844
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9988148896228684e-05,
+      "loss": 0.9844,
+      "step": 845
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99880414144876e-05,
+      "loss": 1.1973,
+      "step": 846
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998793344769553e-05,
+      "loss": 1.1777,
+      "step": 847
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99878249958551e-05,
+      "loss": 0.9717,
+      "step": 848
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998771605896894e-05,
+      "loss": 1.0254,
+      "step": 849
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998760663703969e-05,
+      "loss": 1.1475,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998749673007001e-05,
+      "loss": 1.1436,
+      "step": 851
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998738633806256e-05,
+      "loss": 1.082,
+      "step": 852
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998727546102004e-05,
+      "loss": 1.1816,
+      "step": 853
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998716409894511e-05,
+      "loss": 1.0713,
+      "step": 854
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998705225184049e-05,
+      "loss": 0.7969,
+      "step": 855
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998693991970889e-05,
+      "loss": 0.9424,
+      "step": 856
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9986827102553046e-05,
+      "loss": 1.1836,
+      "step": 857
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9986713800375686e-05,
+      "loss": 0.9707,
+      "step": 858
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998660001317955e-05,
+      "loss": 1.1836,
+      "step": 859
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998648574096742e-05,
+      "loss": 1.0742,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998637098374206e-05,
+      "loss": 1.0703,
+      "step": 861
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9986255741506246e-05,
+      "loss": 1.0605,
+      "step": 862
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998614001426279e-05,
+      "loss": 1.1543,
+      "step": 863
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998602380201449e-05,
+      "loss": 1.1445,
+      "step": 864
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998590710476417e-05,
+      "loss": 1.3125,
+      "step": 865
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9985789922514665e-05,
+      "loss": 1.0967,
+      "step": 866
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9985672255268804e-05,
+      "loss": 1.3271,
+      "step": 867
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998555410302946e-05,
+      "loss": 1.084,
+      "step": 868
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998543546579949e-05,
+      "loss": 1.1055,
+      "step": 869
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998531634358179e-05,
+      "loss": 1.1865,
+      "step": 870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998519673637922e-05,
+      "loss": 1.0635,
+      "step": 871
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9985076644194705e-05,
+      "loss": 1.0703,
+      "step": 872
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9984956067031156e-05,
+      "loss": 1.1494,
+      "step": 873
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998483500489149e-05,
+      "loss": 1.0547,
+      "step": 874
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9984713457778664e-05,
+      "loss": 1.1914,
+      "step": 875
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9984591425695603e-05,
+      "loss": 1.3242,
+      "step": 876
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9984468908645286e-05,
+      "loss": 1.0576,
+      "step": 877
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9984345906630686e-05,
+      "loss": 1.1436,
+      "step": 878
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998422241965477e-05,
+      "loss": 1.0117,
+      "step": 879
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998409844772056e-05,
+      "loss": 1.0586,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998397399083104e-05,
+      "loss": 1.0713,
+      "step": 881
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998384904898926e-05,
+      "loss": 1.1768,
+      "step": 882
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998372362219822e-05,
+      "loss": 1.0225,
+      "step": 883
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998359771046098e-05,
+      "loss": 0.999,
+      "step": 884
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9983471313780585e-05,
+      "loss": 1.1602,
+      "step": 885
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998334443216013e-05,
+      "loss": 1.2432,
+      "step": 886
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998321706560266e-05,
+      "loss": 1.0996,
+      "step": 887
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9983089214111284e-05,
+      "loss": 1.1914,
+      "step": 888
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99829608776891e-05,
+      "loss": 1.207,
+      "step": 889
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9982832056339225e-05,
+      "loss": 1.1377,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998270275006478e-05,
+      "loss": 1.1934,
+      "step": 891
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998257295886891e-05,
+      "loss": 1.0176,
+      "step": 892
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998244268275476e-05,
+      "loss": 1.0322,
+      "step": 893
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998231192172549e-05,
+      "loss": 1.1211,
+      "step": 894
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998218067578428e-05,
+      "loss": 1.248,
+      "step": 895
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998204894493431e-05,
+      "loss": 0.8857,
+      "step": 896
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9981916729178776e-05,
+      "loss": 1.0986,
+      "step": 897
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998178402852088e-05,
+      "loss": 1.1113,
+      "step": 898
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998165084296386e-05,
+      "loss": 0.9561,
+      "step": 899
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9981517172510935e-05,
+      "loss": 0.8789,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9981383017165356e-05,
+      "loss": 1.0713,
+      "step": 901
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998124837693037e-05,
+      "loss": 1.082,
+      "step": 902
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9981113251809246e-05,
+      "loss": 0.9731,
+      "step": 903
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9980977641805266e-05,
+      "loss": 1.1543,
+      "step": 904
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998084154692173e-05,
+      "loss": 1.1836,
+      "step": 905
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998070496716192e-05,
+      "loss": 1.2246,
+      "step": 906
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998056790252916e-05,
+      "loss": 0.9668,
+      "step": 907
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998043035302679e-05,
+      "loss": 1.1543,
+      "step": 908
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998029231865813e-05,
+      "loss": 1.0674,
+      "step": 909
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9980153799426536e-05,
+      "loss": 1.1543,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998001479533536e-05,
+      "loss": 0.999,
+      "step": 911
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997987530638799e-05,
+      "loss": 1.0674,
+      "step": 912
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997973533258781e-05,
+      "loss": 1.0107,
+      "step": 913
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9979594873938206e-05,
+      "loss": 0.9785,
+      "step": 914
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9979453930442596e-05,
+      "loss": 1.1318,
+      "step": 915
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99793125021044e-05,
+      "loss": 1.1045,
+      "step": 916
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9979170588927036e-05,
+      "loss": 1.0498,
+      "step": 917
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997902819091396e-05,
+      "loss": 1.1094,
+      "step": 918
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997888530806863e-05,
+      "loss": 1.0732,
+      "step": 919
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9978741940394514e-05,
+      "loss": 1.1816,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997859808789508e-05,
+      "loss": 1.1621,
+      "step": 921
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9978453750573826e-05,
+      "loss": 1.0635,
+      "step": 922
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9978308928434255e-05,
+      "loss": 1.2285,
+      "step": 923
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9978163621479886e-05,
+      "loss": 0.7295,
+      "step": 924
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997801782971423e-05,
+      "loss": 1.1396,
+      "step": 925
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997787155314084e-05,
+      "loss": 1.1348,
+      "step": 926
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997772479176326e-05,
+      "loss": 1.2305,
+      "step": 927
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9977577545585047e-05,
+      "loss": 1.0273,
+      "step": 928
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997742981460978e-05,
+      "loss": 0.9785,
+      "step": 929
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997728159884105e-05,
+      "loss": 1.1299,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997713289828245e-05,
+      "loss": 1.0693,
+      "step": 931
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997698371293757e-05,
+      "loss": 1.1621,
+      "step": 932
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9976834042810054e-05,
+      "loss": 1.209,
+      "step": 933
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997668388790352e-05,
+      "loss": 1.1748,
+      "step": 934
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9976533248221626e-05,
+      "loss": 0.7949,
+      "step": 935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997638212376801e-05,
+      "loss": 1.123,
+      "step": 936
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997623051454635e-05,
+      "loss": 1.2041,
+      "step": 937
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9976078420560334e-05,
+      "loss": 1.1768,
+      "step": 938
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997592584181363e-05,
+      "loss": 1.168,
+      "step": 939
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997577277830996e-05,
+      "loss": 0.9756,
+      "step": 940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9975619230053026e-05,
+      "loss": 1.0352,
+      "step": 941
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997546519704656e-05,
+      "loss": 0.8994,
+      "step": 942
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9975310679294305e-05,
+      "loss": 1.1318,
+      "step": 943
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99751556768e-05,
+      "loss": 1.0195,
+      "step": 944
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997500018956742e-05,
+      "loss": 1.0859,
+      "step": 945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9974844217600314e-05,
+      "loss": 1.2188,
+      "step": 946
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99746877609025e-05,
+      "loss": 0.8838,
+      "step": 947
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997453081947774e-05,
+      "loss": 1.1494,
+      "step": 948
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9974373393329874e-05,
+      "loss": 1.125,
+      "step": 949
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9974215482462705e-05,
+      "loss": 1.167,
+      "step": 950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997405708688007e-05,
+      "loss": 1.1094,
+      "step": 951
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997389820658581e-05,
+      "loss": 1.1436,
+      "step": 952
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997373884158378e-05,
+      "loss": 1.1426,
+      "step": 953
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9973578991877844e-05,
+      "loss": 1.2676,
+      "step": 954
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997341865747189e-05,
+      "loss": 1.1943,
+      "step": 955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997325783836981e-05,
+      "loss": 1.1348,
+      "step": 956
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997309653457549e-05,
+      "loss": 1.2148,
+      "step": 957
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997293474609286e-05,
+      "loss": 1.168,
+      "step": 958
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997277247292585e-05,
+      "loss": 1.2285,
+      "step": 959
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997260971507838e-05,
+      "loss": 1.293,
+      "step": 960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997244647255442e-05,
+      "loss": 1.1953,
+      "step": 961
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997228274535792e-05,
+      "loss": 0.9893,
+      "step": 962
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997211853349284e-05,
+      "loss": 1.0088,
+      "step": 963
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997195383696319e-05,
+      "loss": 0.96,
+      "step": 964
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997178865577295e-05,
+      "loss": 1.0117,
+      "step": 965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997162298992615e-05,
+      "loss": 1.1514,
+      "step": 966
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9971456839426776e-05,
+      "loss": 1.207,
+      "step": 967
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997129020427888e-05,
+      "loss": 1.0312,
+      "step": 968
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997112308448651e-05,
+      "loss": 0.9932,
+      "step": 969
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997095548005372e-05,
+      "loss": 0.9512,
+      "step": 970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9970787390984565e-05,
+      "loss": 1.0928,
+      "step": 971
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997061881728314e-05,
+      "loss": 1.0566,
+      "step": 972
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997044975895352e-05,
+      "loss": 0.9609,
+      "step": 973
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9970280215999816e-05,
+      "loss": 1.0098,
+      "step": 974
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9970110188426144e-05,
+      "loss": 1.125,
+      "step": 975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996993967623663e-05,
+      "loss": 1.0967,
+      "step": 976
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996976867943541e-05,
+      "loss": 1.0293,
+      "step": 977
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996959719802663e-05,
+      "loss": 1.0479,
+      "step": 978
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996942523201445e-05,
+      "loss": 0.9424,
+      "step": 979
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996925278140306e-05,
+      "loss": 1.1855,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996907984619662e-05,
+      "loss": 0.9688,
+      "step": 981
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9968906426399346e-05,
+      "loss": 1.2383,
+      "step": 982
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996873252201543e-05,
+      "loss": 0.9492,
+      "step": 983
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996855813304912e-05,
+      "loss": 1.0439,
+      "step": 984
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996838325950461e-05,
+      "loss": 1.0068,
+      "step": 985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996820790138618e-05,
+      "loss": 1.0684,
+      "step": 986
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996803205869806e-05,
+      "loss": 1.1465,
+      "step": 987
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9967855731444525e-05,
+      "loss": 0.9971,
+      "step": 988
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996767891962985e-05,
+      "loss": 1.2734,
+      "step": 989
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996750162325834e-05,
+      "loss": 1.2432,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996732384233428e-05,
+      "loss": 1.2285,
+      "step": 991
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9967145576862e-05,
+      "loss": 0.9961,
+      "step": 992
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996696682684581e-05,
+      "loss": 1.1689,
+      "step": 993
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996678759229006e-05,
+      "loss": 0.9717,
+      "step": 994
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9966607873199107e-05,
+      "loss": 1.085,
+      "step": 995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9966427669577286e-05,
+      "loss": 0.9971,
+      "step": 996
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9966246981428984e-05,
+      "loss": 1.0381,
+      "step": 997
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996606580875859e-05,
+      "loss": 1.0977,
+      "step": 998
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9965884151570494e-05,
+      "loss": 1.0967,
+      "step": 999
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9965702009869114e-05,
+      "loss": 1.1289,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996551938365886e-05,
+      "loss": 1.1074,
+      "step": 1001
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996533627294416e-05,
+      "loss": 1.1201,
+      "step": 1002
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9965152677729476e-05,
+      "loss": 1.1445,
+      "step": 1003
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996496859801925e-05,
+      "loss": 1.0938,
+      "step": 1004
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996478403381795e-05,
+      "loss": 1.0645,
+      "step": 1005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9964598985130056e-05,
+      "loss": 1.0332,
+      "step": 1006
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996441345196005e-05,
+      "loss": 0.9512,
+      "step": 1007
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996422743431245e-05,
+      "loss": 1.1074,
+      "step": 1008
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9964040932191764e-05,
+      "loss": 1.0322,
+      "step": 1009
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9963853945602514e-05,
+      "loss": 1.0664,
+      "step": 1010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996366647454924e-05,
+      "loss": 1.1074,
+      "step": 1011
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996347851903649e-05,
+      "loss": 1.1123,
+      "step": 1012
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9963290079068835e-05,
+      "loss": 1.1143,
+      "step": 1013
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9963101154650835e-05,
+      "loss": 1.1016,
+      "step": 1014
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996291174578708e-05,
+      "loss": 1.0996,
+      "step": 1015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996272185248216e-05,
+      "loss": 1.0498,
+      "step": 1016
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99625314747407e-05,
+      "loss": 0.998,
+      "step": 1017
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996234061256729e-05,
+      "loss": 1.2002,
+      "step": 1018
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99621492659666e-05,
+      "loss": 1.1641,
+      "step": 1019
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996195743494324e-05,
+      "loss": 1.1523,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9961765119501893e-05,
+      "loss": 1.0576,
+      "step": 1021
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99615723196472e-05,
+      "loss": 1.1562,
+      "step": 1022
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9961379035383855e-05,
+      "loss": 1.1133,
+      "step": 1023
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996118526671655e-05,
+      "loss": 1.083,
+      "step": 1024
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996099101364998e-05,
+      "loss": 1.3496,
+      "step": 1025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996079627618886e-05,
+      "loss": 0.9893,
+      "step": 1026
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996060105433792e-05,
+      "loss": 1.1162,
+      "step": 1027
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996040534810189e-05,
+      "loss": 1.0322,
+      "step": 1028
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9960209157485524e-05,
+      "loss": 1.1738,
+      "step": 1029
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996001248249358e-05,
+      "loss": 1.0215,
+      "step": 1030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9959815323130836e-05,
+      "loss": 1.0996,
+      "step": 1031
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9959617679402074e-05,
+      "loss": 1.1641,
+      "step": 1032
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9959419551312096e-05,
+      "loss": 0.9512,
+      "step": 1033
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9959220938865694e-05,
+      "loss": 1.0234,
+      "step": 1034
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99590218420677e-05,
+      "loss": 1.2324,
+      "step": 1035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995882226092295e-05,
+      "loss": 1.1953,
+      "step": 1036
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995862219543628e-05,
+      "loss": 1.125,
+      "step": 1037
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995842164561254e-05,
+      "loss": 1.1797,
+      "step": 1038
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99582206114566e-05,
+      "loss": 1.1758,
+      "step": 1039
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995801909297335e-05,
+      "loss": 0.9453,
+      "step": 1040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9957817090167666e-05,
+      "loss": 0.9941,
+      "step": 1041
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995761460304445e-05,
+      "loss": 1.1328,
+      "step": 1042
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9957411631608625e-05,
+      "loss": 1.1914,
+      "step": 1043
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9957208175865113e-05,
+      "loss": 1.0332,
+      "step": 1044
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995700423581886e-05,
+      "loss": 1.2383,
+      "step": 1045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995679981147479e-05,
+      "loss": 1.2266,
+      "step": 1046
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995659490283789e-05,
+      "loss": 1.002,
+      "step": 1047
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9956389509913115e-05,
+      "loss": 1.0371,
+      "step": 1048
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995618363270546e-05,
+      "loss": 0.9844,
+      "step": 1049
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9955977271219915e-05,
+      "loss": 1.209,
+      "step": 1050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995577042546149e-05,
+      "loss": 1.042,
+      "step": 1051
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995556309543521e-05,
+      "loss": 1.208,
+      "step": 1052
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9955355281146096e-05,
+      "loss": 1.0986,
+      "step": 1053
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9955146982599194e-05,
+      "loss": 1.1309,
+      "step": 1054
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995493819979956e-05,
+      "loss": 1.1328,
+      "step": 1055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995472893275227e-05,
+      "loss": 0.8359,
+      "step": 1056
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9954519181462386e-05,
+      "loss": 0.9014,
+      "step": 1057
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995430894593501e-05,
+      "loss": 1.127,
+      "step": 1058
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995409822617523e-05,
+      "loss": 0.876,
+      "step": 1059
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995388702218818e-05,
+      "loss": 1.1484,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995367533397897e-05,
+      "loss": 1.0674,
+      "step": 1061
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9953463161552746e-05,
+      "loss": 1.1562,
+      "step": 1062
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9953250504914645e-05,
+      "loss": 1.1055,
+      "step": 1063
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9953037364069834e-05,
+      "loss": 0.7993,
+      "step": 1064
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995282373902349e-05,
+      "loss": 1.1035,
+      "step": 1065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995260962978079e-05,
+      "loss": 1.2344,
+      "step": 1066
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995239503634694e-05,
+      "loss": 1.0859,
+      "step": 1067
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995217995872713e-05,
+      "loss": 1.1016,
+      "step": 1068
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99519643969266e-05,
+      "loss": 1.0527,
+      "step": 1069
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995174835095057e-05,
+      "loss": 1.1982,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995153182080428e-05,
+      "loss": 1.0664,
+      "step": 1071
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9951314806492985e-05,
+      "loss": 1.0771,
+      "step": 1072
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9951097308021955e-05,
+      "loss": 1.0078,
+      "step": 1073
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995087932539647e-05,
+      "loss": 1.0752,
+      "step": 1074
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9950660858621816e-05,
+      "loss": 1.0586,
+      "step": 1075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9950441907703297e-05,
+      "loss": 1.1816,
+      "step": 1076
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9950222472646224e-05,
+      "loss": 1.0186,
+      "step": 1077
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995000255345592e-05,
+      "loss": 1.1045,
+      "step": 1078
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994978215013773e-05,
+      "loss": 0.8184,
+      "step": 1079
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994956126269699e-05,
+      "loss": 1.1104,
+      "step": 1080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994933989113908e-05,
+      "loss": 1.0908,
+      "step": 1081
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994911803546935e-05,
+      "loss": 1.1533,
+      "step": 1082
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99488956956932e-05,
+      "loss": 1.0645,
+      "step": 1083
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994867287181601e-05,
+      "loss": 1.0518,
+      "step": 1084
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99484495638432e-05,
+      "loss": 0.9473,
+      "step": 1085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994822577178019e-05,
+      "loss": 1.1836,
+      "step": 1086
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9948001495632405e-05,
+      "loss": 1.1523,
+      "step": 1087
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9947776735405286e-05,
+      "loss": 1.2285,
+      "step": 1088
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994755149110429e-05,
+      "loss": 1.0098,
+      "step": 1089
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9947325762734885e-05,
+      "loss": 1.1484,
+      "step": 1090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994709955030255e-05,
+      "loss": 1.2119,
+      "step": 1091
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994687285381276e-05,
+      "loss": 0.9336,
+      "step": 1092
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9946645673271034e-05,
+      "loss": 1.3027,
+      "step": 1093
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9946418008682884e-05,
+      "loss": 1.1543,
+      "step": 1094
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994618986005382e-05,
+      "loss": 1.125,
+      "step": 1095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9945961227389396e-05,
+      "loss": 0.9131,
+      "step": 1096
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9945732110695145e-05,
+      "loss": 1.2656,
+      "step": 1097
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9945502509976634e-05,
+      "loss": 1.0879,
+      "step": 1098
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994527242523944e-05,
+      "loss": 1.0469,
+      "step": 1099
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994504185648914e-05,
+      "loss": 0.9873,
+      "step": 1100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994481080373133e-05,
+      "loss": 1.2383,
+      "step": 1101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994457926697162e-05,
+      "loss": 1.0996,
+      "step": 1102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994434724621562e-05,
+      "loss": 1.0703,
+      "step": 1103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994411474146897e-05,
+      "loss": 1.0664,
+      "step": 1104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9943881752737306e-05,
+      "loss": 0.7861,
+      "step": 1105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9943648280026287e-05,
+      "loss": 1.0146,
+      "step": 1106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994341432334157e-05,
+      "loss": 1.0938,
+      "step": 1107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994317988268884e-05,
+      "loss": 1.1758,
+      "step": 1108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99429449580738e-05,
+      "loss": 1.2051,
+      "step": 1109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9942709549502126e-05,
+      "loss": 1.0625,
+      "step": 1110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9942473656979535e-05,
+      "loss": 1.2051,
+      "step": 1111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9942237280511765e-05,
+      "loss": 1.3223,
+      "step": 1112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994200042010454e-05,
+      "loss": 0.8115,
+      "step": 1113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994176307576362e-05,
+      "loss": 1.1377,
+      "step": 1114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9941525247494745e-05,
+      "loss": 1.1562,
+      "step": 1115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9941286935303706e-05,
+      "loss": 1.0459,
+      "step": 1116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994104813919627e-05,
+      "loss": 0.9395,
+      "step": 1117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994080885917825e-05,
+      "loss": 1.2812,
+      "step": 1118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994056909525544e-05,
+      "loss": 0.9365,
+      "step": 1119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994032884743366e-05,
+      "loss": 0.9834,
+      "step": 1120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994008811571874e-05,
+      "loss": 1.0557,
+      "step": 1121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993984690011653e-05,
+      "loss": 1.0801,
+      "step": 1122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9939605200632866e-05,
+      "loss": 1.0264,
+      "step": 1123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993936301727363e-05,
+      "loss": 0.9727,
+      "step": 1124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9939120350044687e-05,
+      "loss": 0.999,
+      "step": 1125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993887719895194e-05,
+      "loss": 1.1836,
+      "step": 1126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993863356400128e-05,
+      "loss": 0.9873,
+      "step": 1127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993838944519862e-05,
+      "loss": 1.0117,
+      "step": 1128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993814484254988e-05,
+      "loss": 0.9756,
+      "step": 1129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9937899756061005e-05,
+      "loss": 1.1758,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993765418573794e-05,
+      "loss": 0.9189,
+      "step": 1131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993740813158664e-05,
+      "loss": 1.0859,
+      "step": 1132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9937161593613084e-05,
+      "loss": 1.2188,
+      "step": 1133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993691457182325e-05,
+      "loss": 0.8975,
+      "step": 1134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993666706622312e-05,
+      "loss": 1.168,
+      "step": 1135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993641907681872e-05,
+      "loss": 0.918,
+      "step": 1136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993617060361605e-05,
+      "loss": 1.0459,
+      "step": 1137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993592164662116e-05,
+      "loss": 1.1484,
+      "step": 1138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993567220584009e-05,
+      "loss": 1.1514,
+      "step": 1139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993542228127886e-05,
+      "loss": 1.123,
+      "step": 1140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993517187294358e-05,
+      "loss": 1.0322,
+      "step": 1141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99349209808403e-05,
+      "loss": 1.0869,
+      "step": 1142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9934669604975106e-05,
+      "loss": 0.9043,
+      "step": 1143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993441774535412e-05,
+      "loss": 1.1016,
+      "step": 1144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9934165401983425e-05,
+      "loss": 1.1104,
+      "step": 1145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993391257486917e-05,
+      "loss": 0.9385,
+      "step": 1146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993365926401748e-05,
+      "loss": 1.1875,
+      "step": 1147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9933405469434496e-05,
+      "loss": 1.0615,
+      "step": 1148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993315119112639e-05,
+      "loss": 1.1348,
+      "step": 1149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993289642909932e-05,
+      "loss": 1.1523,
+      "step": 1150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993264118335948e-05,
+      "loss": 0.915,
+      "step": 1151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993238545391306e-05,
+      "loss": 1.0996,
+      "step": 1152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993212924076626e-05,
+      "loss": 1.1445,
+      "step": 1153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9931872543925304e-05,
+      "loss": 0.9697,
+      "step": 1154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9931615363396416e-05,
+      "loss": 1.1406,
+      "step": 1155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993135769918584e-05,
+      "loss": 1.1006,
+      "step": 1156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993109955129983e-05,
+      "loss": 1.1006,
+      "step": 1157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993084091974465e-05,
+      "loss": 1.2129,
+      "step": 1158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993058180452657e-05,
+      "loss": 1.0928,
+      "step": 1159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993032220565189e-05,
+      "loss": 1.1055,
+      "step": 1160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.99300621231269e-05,
+      "loss": 1.1367,
+      "step": 1161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9929801556957917e-05,
+      "loss": 1.1807,
+      "step": 1162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992954050715126e-05,
+      "loss": 0.9463,
+      "step": 1163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992927897371327e-05,
+      "loss": 1.082,
+      "step": 1164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992901695665028e-05,
+      "loss": 1.0703,
+      "step": 1165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992875445596866e-05,
+      "loss": 1.0166,
+      "step": 1166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992849147167478e-05,
+      "loss": 1.168,
+      "step": 1167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992822800377502e-05,
+      "loss": 1.041,
+      "step": 1168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992796405227577e-05,
+      "loss": 1.1562,
+      "step": 1169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9927699617183447e-05,
+      "loss": 1.1133,
+      "step": 1170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992743469850445e-05,
+      "loss": 1.2451,
+      "step": 1171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992716929624522e-05,
+      "loss": 1.0908,
+      "step": 1172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.99269034104122e-05,
+      "loss": 1.1357,
+      "step": 1173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992663704101183e-05,
+      "loss": 1.1758,
+      "step": 1174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992637018805059e-05,
+      "loss": 1.2051,
+      "step": 1175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992610285153494e-05,
+      "loss": 1.0791,
+      "step": 1176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992583503147137e-05,
+      "loss": 1.082,
+      "step": 1177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9925566727866385e-05,
+      "loss": 1.0049,
+      "step": 1178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.99252979407265e-05,
+      "loss": 1.2109,
+      "step": 1179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992502867005823e-05,
+      "loss": 1.1533,
+      "step": 1180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9924758915868116e-05,
+      "loss": 1.166,
+      "step": 1181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992448867816269e-05,
+      "loss": 0.9854,
+      "step": 1182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992421795694853e-05,
+      "loss": 1.0889,
+      "step": 1183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992394675223219e-05,
+      "loss": 1.0889,
+      "step": 1184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992367506402026e-05,
+      "loss": 1.1895,
+      "step": 1185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992340289231932e-05,
+      "loss": 1.0381,
+      "step": 1186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9923130237135994e-05,
+      "loss": 1.0342,
+      "step": 1187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992285709847689e-05,
+      "loss": 0.8555,
+      "step": 1188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992258347634863e-05,
+      "loss": 1.1211,
+      "step": 1189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992230937075786e-05,
+      "loss": 0.9932,
+      "step": 1190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992203478171123e-05,
+      "loss": 0.8789,
+      "step": 1191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9921759709215404e-05,
+      "loss": 1.0674,
+      "step": 1192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992148415327706e-05,
+      "loss": 1.084,
+      "step": 1193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992120811390289e-05,
+      "loss": 1.1221,
+      "step": 1194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9920931591099575e-05,
+      "loss": 1.2031,
+      "step": 1195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992065458487384e-05,
+      "loss": 1.1562,
+      "step": 1196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.99203770952324e-05,
+      "loss": 1.2344,
+      "step": 1197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992009912218199e-05,
+      "loss": 1.2246,
+      "step": 1198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991982066572936e-05,
+      "loss": 1.0479,
+      "step": 1199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9919541725881264e-05,
+      "loss": 1.0879,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991926230264447e-05,
+      "loss": 0.8428,
+      "step": 1201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991898239602576e-05,
+      "loss": 1.25,
+      "step": 1202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991870200603194e-05,
+      "loss": 1.0703,
+      "step": 1203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9918421132669786e-05,
+      "loss": 1.0762,
+      "step": 1204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9918139775946134e-05,
+      "loss": 1.1191,
+      "step": 1205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9917857935867806e-05,
+      "loss": 1.209,
+      "step": 1206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991757561244164e-05,
+      "loss": 1.1025,
+      "step": 1207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991729280567449e-05,
+      "loss": 0.9385,
+      "step": 1208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991700951557322e-05,
+      "loss": 0.7646,
+      "step": 1209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991672574214471e-05,
+      "loss": 1.1953,
+      "step": 1210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991644148539583e-05,
+      "loss": 0.9785,
+      "step": 1211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991615674533348e-05,
+      "loss": 1.0967,
+      "step": 1212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9915871521964595e-05,
+      "loss": 0.8647,
+      "step": 1213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9915585815296066e-05,
+      "loss": 1.2676,
+      "step": 1214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9915299625334844e-05,
+      "loss": 1.1055,
+      "step": 1215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991501295208787e-05,
+      "loss": 1.1133,
+      "step": 1216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.99147257955621e-05,
+      "loss": 0.9668,
+      "step": 1217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991443815576449e-05,
+      "loss": 0.8682,
+      "step": 1218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9914150032702046e-05,
+      "loss": 0.9033,
+      "step": 1219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991386142638174e-05,
+      "loss": 1.1064,
+      "step": 1220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991357233681058e-05,
+      "loss": 1.0459,
+      "step": 1221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991328276399559e-05,
+      "loss": 1.0576,
+      "step": 1222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991299270794379e-05,
+      "loss": 1.0479,
+      "step": 1223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991270216866222e-05,
+      "loss": 1.0488,
+      "step": 1224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991241114615792e-05,
+      "loss": 0.9463,
+      "step": 1225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9912119640437965e-05,
+      "loss": 0.9785,
+      "step": 1226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9911827651509425e-05,
+      "loss": 1.0469,
+      "step": 1227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9911535179379394e-05,
+      "loss": 1.2344,
+      "step": 1228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991124222405495e-05,
+      "loss": 1.1289,
+      "step": 1229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991094878554322e-05,
+      "loss": 1.1143,
+      "step": 1230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9910654863851327e-05,
+      "loss": 1.0371,
+      "step": 1231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9910360458986385e-05,
+      "loss": 1.0791,
+      "step": 1232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9910065570955555e-05,
+      "loss": 1.2832,
+      "step": 1233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9909770199765985e-05,
+      "loss": 0.9355,
+      "step": 1234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9909474345424844e-05,
+      "loss": 1.167,
+      "step": 1235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9909178007939315e-05,
+      "loss": 1.2422,
+      "step": 1236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990888118731659e-05,
+      "loss": 1.1133,
+      "step": 1237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990858388356386e-05,
+      "loss": 1.0137,
+      "step": 1238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990828609668835e-05,
+      "loss": 1.0771,
+      "step": 1239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990798782669728e-05,
+      "loss": 1.0576,
+      "step": 1240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.99076890735979e-05,
+      "loss": 1.1602,
+      "step": 1241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990738983739745e-05,
+      "loss": 0.8926,
+      "step": 1242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.99070901181032e-05,
+      "loss": 1.2266,
+      "step": 1243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990678991572241e-05,
+      "loss": 1.0586,
+      "step": 1244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9906489230262374e-05,
+      "loss": 0.9199,
+      "step": 1245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9906188061730394e-05,
+      "loss": 1.1055,
+      "step": 1246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990588641013376e-05,
+      "loss": 1.1787,
+      "step": 1247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9905584275479816e-05,
+      "loss": 1.0762,
+      "step": 1248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990528165777588e-05,
+      "loss": 0.9648,
+      "step": 1249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990497855702929e-05,
+      "loss": 1.0576,
+      "step": 1250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990467497324741e-05,
+      "loss": 1.1758,
+      "step": 1251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9904370906437606e-05,
+      "loss": 1.001,
+      "step": 1252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9904066356607264e-05,
+      "loss": 1.2266,
+      "step": 1253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990376132376376e-05,
+      "loss": 1.0527,
+      "step": 1254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9903455807914505e-05,
+      "loss": 1.1064,
+      "step": 1255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990314980906691e-05,
+      "loss": 1.0771,
+      "step": 1256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990284332722841e-05,
+      "loss": 1.2344,
+      "step": 1257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9902536362406426e-05,
+      "loss": 1.1162,
+      "step": 1258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990222891460842e-05,
+      "loss": 1.1221,
+      "step": 1259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9901920983841846e-05,
+      "loss": 1.168,
+      "step": 1260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9901612570114184e-05,
+      "loss": 1.1621,
+      "step": 1261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9901303673432906e-05,
+      "loss": 0.9824,
+      "step": 1262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9900994293805514e-05,
+      "loss": 0.9619,
+      "step": 1263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990068443123953e-05,
+      "loss": 0.9463,
+      "step": 1264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990037408574245e-05,
+      "loss": 1.1484,
+      "step": 1265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990006325732182e-05,
+      "loss": 1.0996,
+      "step": 1266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989975194598518e-05,
+      "loss": 1.002,
+      "step": 1267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989944015174008e-05,
+      "loss": 0.9756,
+      "step": 1268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989912787459409e-05,
+      "loss": 1.125,
+      "step": 1269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989881511455479e-05,
+      "loss": 0.9443,
+      "step": 1270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989850187162977e-05,
+      "loss": 1.2207,
+      "step": 1271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989818814582662e-05,
+      "loss": 0.9561,
+      "step": 1272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9897873937152966e-05,
+      "loss": 0.9932,
+      "step": 1273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989755924561643e-05,
+      "loss": 1.1084,
+      "step": 1274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989724407122465e-05,
+      "loss": 1.0391,
+      "step": 1275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989692841398527e-05,
+      "loss": 1.1084,
+      "step": 1276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989661227390595e-05,
+      "loss": 1.0635,
+      "step": 1277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989629565099437e-05,
+      "loss": 1.0488,
+      "step": 1278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.98959785452582e-05,
+      "loss": 1.126,
+      "step": 1279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9895660956705156e-05,
+      "loss": 1.1006,
+      "step": 1280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989534288534292e-05,
+      "loss": 1.0293,
+      "step": 1281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989502433117922e-05,
+      "loss": 1.2246,
+      "step": 1282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989470529422179e-05,
+      "loss": 1.0352,
+      "step": 1283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9894385774478376e-05,
+      "loss": 1.1279,
+      "step": 1284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989406577195673e-05,
+      "loss": 1.042,
+      "step": 1285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.98937452866646e-05,
+      "loss": 1.2002,
+      "step": 1286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989342431860979e-05,
+      "loss": 1.1914,
+      "step": 1287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9893102867800065e-05,
+      "loss": 0.9688,
+      "step": 1288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9892780934243253e-05,
+      "loss": 1.0947,
+      "step": 1289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989245851794714e-05,
+      "loss": 1.1387,
+      "step": 1290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989213561891956e-05,
+      "loss": 0.9307,
+      "step": 1291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989181223716835e-05,
+      "loss": 1.1201,
+      "step": 1292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989148837270136e-05,
+      "loss": 0.9521,
+      "step": 1293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989116402552644e-05,
+      "loss": 1.0068,
+      "step": 1294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989083919565147e-05,
+      "loss": 1.293,
+      "step": 1295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9890513883084326e-05,
+      "loss": 0.9834,
+      "step": 1296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989018808783291e-05,
+      "loss": 0.9385,
+      "step": 1297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988986180990513e-05,
+      "loss": 0.8477,
+      "step": 1298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988953504930889e-05,
+      "loss": 1.1289,
+      "step": 1299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988920780605213e-05,
+      "loss": 1.0498,
+      "step": 1300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9888880080142783e-05,
+      "loss": 1.123,
+      "step": 1301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9888551871588814e-05,
+      "loss": 1.1016,
+      "step": 1302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9888223180398175e-05,
+      "loss": 0.9619,
+      "step": 1303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988789400657886e-05,
+      "loss": 0.9248,
+      "step": 1304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988756435013884e-05,
+      "loss": 1.1738,
+      "step": 1305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988723421108612e-05,
+      "loss": 0.9141,
+      "step": 1306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988690358942871e-05,
+      "loss": 1.042,
+      "step": 1307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988657248517464e-05,
+      "loss": 0.8906,
+      "step": 1308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9886240898331926e-05,
+      "loss": 1.0732,
+      "step": 1309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9885908828908643e-05,
+      "loss": 0.9121,
+      "step": 1310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9885576276912835e-05,
+      "loss": 1.0557,
+      "step": 1311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9885243242352565e-05,
+      "loss": 0.9932,
+      "step": 1312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9884909725235924e-05,
+      "loss": 0.9639,
+      "step": 1313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9884575725571004e-05,
+      "loss": 1.123,
+      "step": 1314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9884241243365914e-05,
+      "loss": 1.127,
+      "step": 1315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988390627862876e-05,
+      "loss": 1.0381,
+      "step": 1316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988357083136768e-05,
+      "loss": 1.0215,
+      "step": 1317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988323490159081e-05,
+      "loss": 1.1123,
+      "step": 1318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988289848930631e-05,
+      "loss": 0.9199,
+      "step": 1319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9882561594522326e-05,
+      "loss": 1.0918,
+      "step": 1320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9882224217247046e-05,
+      "loss": 1.1289,
+      "step": 1321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9881886357488667e-05,
+      "loss": 0.9648,
+      "step": 1322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988154801525537e-05,
+      "loss": 1.0508,
+      "step": 1323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9881209190555374e-05,
+      "loss": 1.0459,
+      "step": 1324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988086988339689e-05,
+      "loss": 1.0293,
+      "step": 1325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988053009378817e-05,
+      "loss": 1.0713,
+      "step": 1326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9880189821737455e-05,
+      "loss": 1.2246,
+      "step": 1327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9879849067252984e-05,
+      "loss": 1.0938,
+      "step": 1328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987950783034306e-05,
+      "loss": 1.1484,
+      "step": 1329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987916611101593e-05,
+      "loss": 0.9507,
+      "step": 1330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9878823909279906e-05,
+      "loss": 1.1074,
+      "step": 1331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9878481225143284e-05,
+      "loss": 1.1641,
+      "step": 1332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9878138058614386e-05,
+      "loss": 1.0303,
+      "step": 1333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987779440970154e-05,
+      "loss": 1.1465,
+      "step": 1334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987745027841308e-05,
+      "loss": 0.9795,
+      "step": 1335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987710566475736e-05,
+      "loss": 0.9404,
+      "step": 1336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987676056874274e-05,
+      "loss": 1.2148,
+      "step": 1337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9876414990377595e-05,
+      "loss": 1.0879,
+      "step": 1338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987606892967031e-05,
+      "loss": 1.0771,
+      "step": 1339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9875722386629295e-05,
+      "loss": 1.0908,
+      "step": 1340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9875375361262946e-05,
+      "loss": 1.0859,
+      "step": 1341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9875027853579686e-05,
+      "loss": 1.1602,
+      "step": 1342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987467986358795e-05,
+      "loss": 1.0693,
+      "step": 1343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9874331391296184e-05,
+      "loss": 1.0977,
+      "step": 1344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987398243671284e-05,
+      "loss": 1.0762,
+      "step": 1345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9873632999846396e-05,
+      "loss": 1.0996,
+      "step": 1346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987328308070533e-05,
+      "loss": 1.0762,
+      "step": 1347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987293267929812e-05,
+      "loss": 1.1279,
+      "step": 1348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987258179563328e-05,
+      "loss": 1.1484,
+      "step": 1349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9872230429719326e-05,
+      "loss": 0.9902,
+      "step": 1350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987187858156477e-05,
+      "loss": 1.1484,
+      "step": 1351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.987152625117817e-05,
+      "loss": 1.0498,
+      "step": 1352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.987117343856807e-05,
+      "loss": 1.2109,
+      "step": 1353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.987082014374303e-05,
+      "loss": 1.0283,
+      "step": 1354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.987046636671162e-05,
+      "loss": 1.1445,
+      "step": 1355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.987011210748243e-05,
+      "loss": 1.0,
+      "step": 1356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9869757366064056e-05,
+      "loss": 1.1543,
+      "step": 1357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98694021424651e-05,
+      "loss": 0.7451,
+      "step": 1358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986904643669419e-05,
+      "loss": 0.999,
+      "step": 1359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986869024875996e-05,
+      "loss": 0.9717,
+      "step": 1360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986833357867105e-05,
+      "loss": 0.9639,
+      "step": 1361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986797642643611e-05,
+      "loss": 1.1377,
+      "step": 1362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9867618792063816e-05,
+      "loss": 0.9395,
+      "step": 1363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986726067556284e-05,
+      "loss": 1.0771,
+      "step": 1364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986690207694187e-05,
+      "loss": 1.0664,
+      "step": 1365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9866542996209615e-05,
+      "loss": 1.042,
+      "step": 1366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986618343337479e-05,
+      "loss": 1.0771,
+      "step": 1367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986582338844613e-05,
+      "loss": 1.1855,
+      "step": 1368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986546286143234e-05,
+      "loss": 1.0479,
+      "step": 1369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986510185234219e-05,
+      "loss": 1.1836,
+      "step": 1370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986474036118445e-05,
+      "loss": 1.1846,
+      "step": 1371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9864378387967874e-05,
+      "loss": 1.1836,
+      "step": 1372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9864015932701254e-05,
+      "loss": 0.9727,
+      "step": 1373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986365299539339e-05,
+      "loss": 1.0811,
+      "step": 1374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986328957605308e-05,
+      "loss": 0.9932,
+      "step": 1375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9862925674689155e-05,
+      "loss": 1.0723,
+      "step": 1376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986256129131043e-05,
+      "loss": 1.2969,
+      "step": 1377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986219642592576e-05,
+      "loss": 1.0068,
+      "step": 1378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986183107854399e-05,
+      "loss": 1.043,
+      "step": 1379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9861465249174e-05,
+      "loss": 1.0361,
+      "step": 1380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986109893782465e-05,
+      "loss": 1.0352,
+      "step": 1381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986073214450485e-05,
+      "loss": 1.1387,
+      "step": 1382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986036486922347e-05,
+      "loss": 1.0107,
+      "step": 1383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9859997111989465e-05,
+      "loss": 1.1943,
+      "step": 1384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985962887281172e-05,
+      "loss": 1.1797,
+      "step": 1385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985926015169919e-05,
+      "loss": 1.1201,
+      "step": 1386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9858890948660826e-05,
+      "loss": 0.9746,
+      "step": 1387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9858521263705584e-05,
+      "loss": 1.1465,
+      "step": 1388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985815109684242e-05,
+      "loss": 1.2773,
+      "step": 1389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9857780448080336e-05,
+      "loss": 1.085,
+      "step": 1390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985740931742832e-05,
+      "loss": 1.0918,
+      "step": 1391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985703770489538e-05,
+      "loss": 1.0908,
+      "step": 1392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985666561049053e-05,
+      "loss": 0.9199,
+      "step": 1393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985629303422281e-05,
+      "loss": 1.0898,
+      "step": 1394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9855919976101244e-05,
+      "loss": 1.085,
+      "step": 1395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98555464361349e-05,
+      "loss": 1.083,
+      "step": 1396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9855172414332835e-05,
+      "loss": 1.1211,
+      "step": 1397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9854797910704124e-05,
+      "loss": 1.0723,
+      "step": 1398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985442292525786e-05,
+      "loss": 1.124,
+      "step": 1399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985404745800314e-05,
+      "loss": 1.0752,
+      "step": 1400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9853671508949086e-05,
+      "loss": 1.1992,
+      "step": 1401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9853295078104806e-05,
+      "loss": 1.0879,
+      "step": 1402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9852918165479435e-05,
+      "loss": 1.1191,
+      "step": 1403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985254077108213e-05,
+      "loss": 1.0645,
+      "step": 1404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985216289492205e-05,
+      "loss": 1.166,
+      "step": 1405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985178453700835e-05,
+      "loss": 1.1641,
+      "step": 1406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9851405697350235e-05,
+      "loss": 1.0811,
+      "step": 1407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9851026375956874e-05,
+      "loss": 1.002,
+      "step": 1408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985064657283748e-05,
+      "loss": 1.04,
+      "step": 1409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9850266288001286e-05,
+      "loss": 1.0029,
+      "step": 1410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9849885521457496e-05,
+      "loss": 0.7949,
+      "step": 1411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984950427321536e-05,
+      "loss": 1.042,
+      "step": 1412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984912254328413e-05,
+      "loss": 1.1504,
+      "step": 1413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984874033167308e-05,
+      "loss": 1.1504,
+      "step": 1414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9848357638391465e-05,
+      "loss": 1.0762,
+      "step": 1415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984797446344858e-05,
+      "loss": 1.0166,
+      "step": 1416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984759080685373e-05,
+      "loss": 1.0713,
+      "step": 1417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984720666861623e-05,
+      "loss": 1.1289,
+      "step": 1418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984682204874538e-05,
+      "loss": 1.0869,
+      "step": 1419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984643694725053e-05,
+      "loss": 1.0449,
+      "step": 1420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984605136414102e-05,
+      "loss": 1.2129,
+      "step": 1421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9845665299426216e-05,
+      "loss": 0.9775,
+      "step": 1422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984527875311547e-05,
+      "loss": 0.9922,
+      "step": 1423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9844891725218175e-05,
+      "loss": 1.0518,
+      "step": 1424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984450421574372e-05,
+      "loss": 1.0713,
+      "step": 1425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984411622470151e-05,
+      "loss": 0.915,
+      "step": 1426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984372775210096e-05,
+      "loss": 0.9814,
+      "step": 1427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984333879795149e-05,
+      "loss": 1.082,
+      "step": 1428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9842949362262546e-05,
+      "loss": 1.0645,
+      "step": 1429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984255944504358e-05,
+      "loss": 1.0234,
+      "step": 1430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984216904630404e-05,
+      "loss": 1.2344,
+      "step": 1431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984177816605343e-05,
+      "loss": 1.1221,
+      "step": 1432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984138680430121e-05,
+      "loss": 0.7764,
+      "step": 1433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984099496105688e-05,
+      "loss": 1.0879,
+      "step": 1434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9840602636329956e-05,
+      "loss": 1.209,
+      "step": 1435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984020983012995e-05,
+      "loss": 1.2266,
+      "step": 1436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9839816542466396e-05,
+      "loss": 1.1338,
+      "step": 1437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983942277334885e-05,
+      "loss": 0.9316,
+      "step": 1438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983902852278686e-05,
+      "loss": 1.1523,
+      "step": 1439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9838633790789985e-05,
+      "loss": 0.917,
+      "step": 1440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9838238577367816e-05,
+      "loss": 1.0039,
+      "step": 1441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983784288252993e-05,
+      "loss": 1.1436,
+      "step": 1442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9837446706285946e-05,
+      "loss": 0.9395,
+      "step": 1443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9837050048645477e-05,
+      "loss": 1.0225,
+      "step": 1444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983665290961813e-05,
+      "loss": 0.9121,
+      "step": 1445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983625528921356e-05,
+      "loss": 1.2656,
+      "step": 1446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983585718744141e-05,
+      "loss": 1.209,
+      "step": 1447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983545860431134e-05,
+      "loss": 1.0557,
+      "step": 1448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9835059539833025e-05,
+      "loss": 1.0732,
+      "step": 1449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983465999401615e-05,
+      "loss": 1.1035,
+      "step": 1450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98342599668704e-05,
+      "loss": 1.1299,
+      "step": 1451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98338594584055e-05,
+      "loss": 1.04,
+      "step": 1452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983345846863116e-05,
+      "loss": 0.918,
+      "step": 1453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983305699755711e-05,
+      "loss": 1.0303,
+      "step": 1454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983265504519309e-05,
+      "loss": 1.0977,
+      "step": 1455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9832252611548854e-05,
+      "loss": 0.959,
+      "step": 1456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983184969663418e-05,
+      "loss": 1.0791,
+      "step": 1457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9831446300458835e-05,
+      "loss": 0.8828,
+      "step": 1458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9831042423032615e-05,
+      "loss": 1.1914,
+      "step": 1459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983063806436531e-05,
+      "loss": 1.0771,
+      "step": 1460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9830233224466733e-05,
+      "loss": 0.9883,
+      "step": 1461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982982790334672e-05,
+      "loss": 1.1465,
+      "step": 1462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982942210101511e-05,
+      "loss": 1.126,
+      "step": 1463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982901581748174e-05,
+      "loss": 1.2773,
+      "step": 1464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982860905275646e-05,
+      "loss": 0.8301,
+      "step": 1465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982820180684916e-05,
+      "loss": 1.2461,
+      "step": 1466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982779407976971e-05,
+      "loss": 1.0791,
+      "step": 1467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982738587152801e-05,
+      "loss": 1.1953,
+      "step": 1468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982697718213397e-05,
+      "loss": 0.9834,
+      "step": 1469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98265680115975e-05,
+      "loss": 1.0498,
+      "step": 1470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982615835992853e-05,
+      "loss": 1.0898,
+      "step": 1471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9825748227137006e-05,
+      "loss": 0.9551,
+      "step": 1472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9825337613232874e-05,
+      "loss": 0.9814,
+      "step": 1473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982492651822611e-05,
+      "loss": 1.0059,
+      "step": 1474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9824514942126676e-05,
+      "loss": 1.1855,
+      "step": 1475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9824102884944566e-05,
+      "loss": 0.9404,
+      "step": 1476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982369034668979e-05,
+      "loss": 1.0547,
+      "step": 1477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982327732737234e-05,
+      "loss": 1.1357,
+      "step": 1478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982286382700224e-05,
+      "loss": 1.0693,
+      "step": 1479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982244984558954e-05,
+      "loss": 1.0938,
+      "step": 1480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982203538314428e-05,
+      "loss": 1.166,
+      "step": 1481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9821620439676514e-05,
+      "loss": 1.1406,
+      "step": 1482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982120501519632e-05,
+      "loss": 1.0957,
+      "step": 1483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982078910971376e-05,
+      "loss": 0.9902,
+      "step": 1484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9820372723238946e-05,
+      "loss": 1.0078,
+      "step": 1485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9819955855781976e-05,
+      "loss": 1.0703,
+      "step": 1486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981953850735296e-05,
+      "loss": 1.0508,
+      "step": 1487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981912067796203e-05,
+      "loss": 1.1172,
+      "step": 1488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981870236761934e-05,
+      "loss": 1.0215,
+      "step": 1489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981828357633501e-05,
+      "loss": 0.8311,
+      "step": 1490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9817864304119236e-05,
+      "loss": 1.0,
+      "step": 1491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981744455098217e-05,
+      "loss": 0.9473,
+      "step": 1492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9817024316934014e-05,
+      "loss": 1.0146,
+      "step": 1493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981660360198495e-05,
+      "loss": 1.1914,
+      "step": 1494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98161824061452e-05,
+      "loss": 1.082,
+      "step": 1495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981576072942498e-05,
+      "loss": 0.8564,
+      "step": 1496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981533857183452e-05,
+      "loss": 0.8623,
+      "step": 1497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981491593338407e-05,
+      "loss": 1.209,
+      "step": 1498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981449281408388e-05,
+      "loss": 1.0078,
+      "step": 1499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981406921394422e-05,
+      "loss": 1.0049,
+      "step": 1500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981364513297538e-05,
+      "loss": 1.0049,
+      "step": 1501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981322057118764e-05,
+      "loss": 0.9287,
+      "step": 1502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98127955285913e-05,
+      "loss": 1.1621,
+      "step": 1503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9812370005196686e-05,
+      "loss": 1.1699,
+      "step": 1504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981194400101411e-05,
+      "loss": 1.1543,
+      "step": 1505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981151751605393e-05,
+      "loss": 0.918,
+      "step": 1506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981109055032647e-05,
+      "loss": 1.2031,
+      "step": 1507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981066310384211e-05,
+      "loss": 1.0752,
+      "step": 1508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9810235176611224e-05,
+      "loss": 1.1895,
+      "step": 1509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980980676864419e-05,
+      "loss": 1.0674,
+      "step": 1510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9809377879951394e-05,
+      "loss": 0.9785,
+      "step": 1511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980894851054326e-05,
+      "loss": 1.042,
+      "step": 1512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98085186604302e-05,
+      "loss": 1.0303,
+      "step": 1513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980808832962265e-05,
+      "loss": 1.0537,
+      "step": 1514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980765751813105e-05,
+      "loss": 0.8965,
+      "step": 1515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980722622596585e-05,
+      "loss": 1.1934,
+      "step": 1516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980679445313752e-05,
+      "loss": 1.0645,
+      "step": 1517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980636219965654e-05,
+      "loss": 0.9082,
+      "step": 1518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9805929465533405e-05,
+      "loss": 1.001,
+      "step": 1519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98054962507786e-05,
+      "loss": 1.0361,
+      "step": 1520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9805062555402647e-05,
+      "loss": 1.1426,
+      "step": 1521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9804628379416074e-05,
+      "loss": 1.0205,
+      "step": 1522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980419372282941e-05,
+      "loss": 0.958,
+      "step": 1523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980375858565321e-05,
+      "loss": 1.0107,
+      "step": 1524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980332296789803e-05,
+      "loss": 1.0137,
+      "step": 1525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980288686957444e-05,
+      "loss": 1.0928,
+      "step": 1526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980245029069302e-05,
+      "loss": 0.9395,
+      "step": 1527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9802013231264374e-05,
+      "loss": 1.125,
+      "step": 1528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9801575691299096e-05,
+      "loss": 1.0547,
+      "step": 1529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980113767080782e-05,
+      "loss": 1.127,
+      "step": 1530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980069916980116e-05,
+      "loss": 1.0049,
+      "step": 1531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980026018828976e-05,
+      "loss": 1.2031,
+      "step": 1532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979982072628428e-05,
+      "loss": 1.0869,
+      "step": 1533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979938078379539e-05,
+      "loss": 1.0518,
+      "step": 1534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979894036083374e-05,
+      "loss": 1.1611,
+      "step": 1535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9798499457410035e-05,
+      "loss": 0.9316,
+      "step": 1536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979805807353498e-05,
+      "loss": 1.0234,
+      "step": 1537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9797616209219276e-05,
+      "loss": 0.9824,
+      "step": 1538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979717386447365e-05,
+      "loss": 0.958,
+      "step": 1539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979673103930884e-05,
+      "loss": 0.9268,
+      "step": 1540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9796287733735584e-05,
+      "loss": 0.9697,
+      "step": 1541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979584394776464e-05,
+      "loss": 1.0322,
+      "step": 1542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979539968140679e-05,
+      "loss": 0.9688,
+      "step": 1543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.97949549346728e-05,
+      "loss": 0.9385,
+      "step": 1544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979450970757347e-05,
+      "loss": 0.957,
+      "step": 1545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.97940640001196e-05,
+      "loss": 1.0615,
+      "step": 1546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979361781232201e-05,
+      "loss": 0.9229,
+      "step": 1547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979317114419153e-05,
+      "loss": 1.0928,
+      "step": 1548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979272399573899e-05,
+      "loss": 1.0664,
+      "step": 1549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979227636697525e-05,
+      "loss": 1.0645,
+      "step": 1550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9791828257911176e-05,
+      "loss": 0.9912,
+      "step": 1551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9791379668557626e-05,
+      "loss": 0.9717,
+      "step": 1552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.97909305989255e-05,
+      "loss": 1.1123,
+      "step": 1553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979048104902569e-05,
+      "loss": 1.0361,
+      "step": 1554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979003101886911e-05,
+      "loss": 0.8027,
+      "step": 1555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9789580508466676e-05,
+      "loss": 1.082,
+      "step": 1556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9789129517829324e-05,
+      "loss": 1.0752,
+      "step": 1557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9788678046968e-05,
+      "loss": 0.9336,
+      "step": 1558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.978822609589364e-05,
+      "loss": 1.1113,
+      "step": 1559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978777366461724e-05,
+      "loss": 1.1172,
+      "step": 1560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9787320753149766e-05,
+      "loss": 0.875,
+      "step": 1561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978686736150221e-05,
+      "loss": 1.0078,
+      "step": 1562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9786413489685574e-05,
+      "loss": 1.0869,
+      "step": 1563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978595913771087e-05,
+      "loss": 1.1016,
+      "step": 1564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978550430558913e-05,
+      "loss": 0.918,
+      "step": 1565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9785048993331384e-05,
+      "loss": 1.0576,
+      "step": 1566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978459320094868e-05,
+      "loss": 1.0713,
+      "step": 1567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978413692845209e-05,
+      "loss": 1.0977,
+      "step": 1568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978368017585268e-05,
+      "loss": 1.084,
+      "step": 1569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978322294316153e-05,
+      "loss": 1.1035,
+      "step": 1570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978276523038974e-05,
+      "loss": 1.1074,
+      "step": 1571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978230703754842e-05,
+      "loss": 0.9004,
+      "step": 1572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978184836464869e-05,
+      "loss": 0.9971,
+      "step": 1573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978138921170166e-05,
+      "loss": 1.1152,
+      "step": 1574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978092957871851e-05,
+      "loss": 1.0332,
+      "step": 1575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978046946571036e-05,
+      "loss": 1.0049,
+      "step": 1576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978000887268839e-05,
+      "loss": 1.2041,
+      "step": 1577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977954779966378e-05,
+      "loss": 0.9756,
+      "step": 1578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977908624664771e-05,
+      "loss": 1.0273,
+      "step": 1579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977862421365139e-05,
+      "loss": 1.084,
+      "step": 1580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977816170068602e-05,
+      "loss": 1.1348,
+      "step": 1581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977769870776284e-05,
+      "loss": 1.1084,
+      "step": 1582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977723523489306e-05,
+      "loss": 1.1738,
+      "step": 1583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977677128208796e-05,
+      "loss": 1.0039,
+      "step": 1584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9776306849358775e-05,
+      "loss": 0.873,
+      "step": 1585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977584193671679e-05,
+      "loss": 1.0957,
+      "step": 1586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9775376544173274e-05,
+      "loss": 1.0332,
+      "step": 1587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9774910671739525e-05,
+      "loss": 1.0005,
+      "step": 1588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9774444319426855e-05,
+      "loss": 1.0312,
+      "step": 1589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977397748724657e-05,
+      "loss": 0.8584,
+      "step": 1590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977351017521001e-05,
+      "loss": 1.0332,
+      "step": 1591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977304238332851e-05,
+      "loss": 1.1123,
+      "step": 1592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977257411161342e-05,
+      "loss": 1.2656,
+      "step": 1593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9772105360076096e-05,
+      "loss": 1.0615,
+      "step": 1594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9771636128727936e-05,
+      "loss": 1.0752,
+      "step": 1595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977116641758031e-05,
+      "loss": 1.1133,
+      "step": 1596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977069622664462e-05,
+      "loss": 1.0439,
+      "step": 1597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977022555593227e-05,
+      "loss": 1.002,
+      "step": 1598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976975440545469e-05,
+      "loss": 1.0332,
+      "step": 1599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976928277522332e-05,
+      "loss": 1.0693,
+      "step": 1600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9768810665249585e-05,
+      "loss": 1.1621,
+      "step": 1601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976833807554496e-05,
+      "loss": 1.1172,
+      "step": 1602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9767865006120904e-05,
+      "loss": 0.9902,
+      "step": 1603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97673914569889e-05,
+      "loss": 1.0439,
+      "step": 1604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9766917428160444e-05,
+      "loss": 1.0264,
+      "step": 1605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976644291964703e-05,
+      "loss": 1.125,
+      "step": 1606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976596793146017e-05,
+      "loss": 1.0303,
+      "step": 1607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976549246361141e-05,
+      "loss": 0.9541,
+      "step": 1608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976501651611227e-05,
+      "loss": 1.1934,
+      "step": 1609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976454008897431e-05,
+      "loss": 1.0156,
+      "step": 1610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9764063182209086e-05,
+      "loss": 1.0615,
+      "step": 1611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976358579582817e-05,
+      "loss": 1.0225,
+      "step": 1612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976310792984315e-05,
+      "loss": 0.9082,
+      "step": 1613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9762629584265624e-05,
+      "loss": 1.084,
+      "step": 1614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9762150759107194e-05,
+      "loss": 0.9102,
+      "step": 1615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9761671454379484e-05,
+      "loss": 1.0547,
+      "step": 1616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9761191670094124e-05,
+      "loss": 0.916,
+      "step": 1617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976071140626276e-05,
+      "loss": 0.9053,
+      "step": 1618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976023066289705e-05,
+      "loss": 1.1016,
+      "step": 1619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975974944000865e-05,
+      "loss": 1.2266,
+      "step": 1620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975926773760925e-05,
+      "loss": 1.002,
+      "step": 1621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9758785555710516e-05,
+      "loss": 1.1592,
+      "step": 1622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975830289432417e-05,
+      "loss": 0.96,
+      "step": 1623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975781975346192e-05,
+      "loss": 1.1523,
+      "step": 1624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97573361331355e-05,
+      "loss": 1.1182,
+      "step": 1625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9756852033356626e-05,
+      "loss": 0.9756,
+      "step": 1626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975636745413706e-05,
+      "loss": 1.1338,
+      "step": 1627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9755882395488554e-05,
+      "loss": 1.1045,
+      "step": 1628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975539685742289e-05,
+      "loss": 1.2061,
+      "step": 1629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9754910839951835e-05,
+      "loss": 0.9756,
+      "step": 1630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975442434308719e-05,
+      "loss": 0.9521,
+      "step": 1631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975393736684077e-05,
+      "loss": 0.9766,
+      "step": 1632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9753449911224376e-05,
+      "loss": 1.1562,
+      "step": 1633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9752961976249846e-05,
+      "loss": 1.0732,
+      "step": 1634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9752473561929025e-05,
+      "loss": 1.1709,
+      "step": 1635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9751984668273757e-05,
+      "loss": 1.0322,
+      "step": 1636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975149529529591e-05,
+      "loss": 0.9502,
+      "step": 1637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9751005443007366e-05,
+      "loss": 1.2637,
+      "step": 1638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975051511142e-05,
+      "loss": 1.1992,
+      "step": 1639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975002430054572e-05,
+      "loss": 0.9912,
+      "step": 1640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974953301039643e-05,
+      "loss": 1.2129,
+      "step": 1641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974904124098405e-05,
+      "loss": 0.8574,
+      "step": 1642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974854899232052e-05,
+      "loss": 0.7346,
+      "step": 1643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9748056264417793e-05,
+      "loss": 1.0977,
+      "step": 1644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974756305728781e-05,
+      "loss": 1.082,
+      "step": 1645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9747069370942554e-05,
+      "loss": 1.0,
+      "step": 1646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974657520539399e-05,
+      "loss": 1.0889,
+      "step": 1647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9746080560654124e-05,
+      "loss": 1.207,
+      "step": 1648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974558543673495e-05,
+      "loss": 1.0986,
+      "step": 1649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974508983364849e-05,
+      "loss": 1.1387,
+      "step": 1650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974459375140676e-05,
+      "loss": 0.9541,
+      "step": 1651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974409719002182e-05,
+      "loss": 1.0781,
+      "step": 1652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9743600149505696e-05,
+      "loss": 1.1914,
+      "step": 1653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9743102629870465e-05,
+      "loss": 1.0107,
+      "step": 1654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97426046311282e-05,
+      "loss": 1.1875,
+      "step": 1655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974210615329098e-05,
+      "loss": 1.1562,
+      "step": 1656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974160719637089e-05,
+      "loss": 1.0742,
+      "step": 1657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9741107760380064e-05,
+      "loss": 0.9971,
+      "step": 1658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974060784533061e-05,
+      "loss": 1.209,
+      "step": 1659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974010745123465e-05,
+      "loss": 1.2012,
+      "step": 1660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973960657810434e-05,
+      "loss": 1.1211,
+      "step": 1661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973910522595183e-05,
+      "loss": 1.1094,
+      "step": 1662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973860339478929e-05,
+      "loss": 0.8896,
+      "step": 1663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973810108462889e-05,
+      "loss": 1.0771,
+      "step": 1664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9737598295482834e-05,
+      "loss": 0.8594,
+      "step": 1665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97370950273633e-05,
+      "loss": 1.0439,
+      "step": 1666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973659128028252e-05,
+      "loss": 1.043,
+      "step": 1667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9736087054252716e-05,
+      "loss": 1.0928,
+      "step": 1668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9735582349286115e-05,
+      "loss": 0.8516,
+      "step": 1669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973507716539498e-05,
+      "loss": 0.9346,
+      "step": 1670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973457150259156e-05,
+      "loss": 1.0605,
+      "step": 1671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973406536088812e-05,
+      "loss": 1.1016,
+      "step": 1672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973355874029695e-05,
+      "loss": 1.1357,
+      "step": 1673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9733051640830345e-05,
+      "loss": 1.2402,
+      "step": 1674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97325440625006e-05,
+      "loss": 1.0869,
+      "step": 1675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9732036005320055e-05,
+      "loss": 0.9775,
+      "step": 1676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973152746930103e-05,
+      "loss": 0.9326,
+      "step": 1677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973101845445585e-05,
+      "loss": 1.0586,
+      "step": 1678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973050896079688e-05,
+      "loss": 0.9268,
+      "step": 1679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972999898833649e-05,
+      "loss": 1.0205,
+      "step": 1680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9729488537087044e-05,
+      "loss": 1.0088,
+      "step": 1681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972897760706093e-05,
+      "loss": 1.0352,
+      "step": 1682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9728466198270556e-05,
+      "loss": 1.1904,
+      "step": 1683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972795431072833e-05,
+      "loss": 1.209,
+      "step": 1684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972744194444666e-05,
+      "loss": 1.0625,
+      "step": 1685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9726929099438e-05,
+      "loss": 1.0645,
+      "step": 1686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972641577571478e-05,
+      "loss": 0.9375,
+      "step": 1687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9725901973289464e-05,
+      "loss": 1.1553,
+      "step": 1688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9725387692174517e-05,
+      "loss": 0.8916,
+      "step": 1689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972487293238242e-05,
+      "loss": 1.0527,
+      "step": 1690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9724357693925675e-05,
+      "loss": 0.8428,
+      "step": 1691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9723841976816765e-05,
+      "loss": 1.2344,
+      "step": 1692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972332578106822e-05,
+      "loss": 1.0811,
+      "step": 1693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9722809106692565e-05,
+      "loss": 0.9141,
+      "step": 1694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972229195370233e-05,
+      "loss": 1.1084,
+      "step": 1695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972177432211008e-05,
+      "loss": 1.0879,
+      "step": 1696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972125621192836e-05,
+      "loss": 0.9619,
+      "step": 1697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972073762316975e-05,
+      "loss": 1.0713,
+      "step": 1698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9720218555846834e-05,
+      "loss": 1.0254,
+      "step": 1699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971969900997221e-05,
+      "loss": 1.1387,
+      "step": 1700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971917898555849e-05,
+      "loss": 1.1777,
+      "step": 1701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971865848261828e-05,
+      "loss": 1.0869,
+      "step": 1702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9718137501164224e-05,
+      "loss": 1.0703,
+      "step": 1703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971761604120896e-05,
+      "loss": 1.0781,
+      "step": 1704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971709410276513e-05,
+      "loss": 1.0762,
+      "step": 1705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971657168584543e-05,
+      "loss": 1.168,
+      "step": 1706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971604879046251e-05,
+      "loss": 0.9395,
+      "step": 1707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971552541662907e-05,
+      "loss": 0.9609,
+      "step": 1708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971500156435781e-05,
+      "loss": 0.9883,
+      "step": 1709
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9714477233661436e-05,
+      "loss": 1.1514,
+      "step": 1710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971395242455269e-05,
+      "loss": 1.0215,
+      "step": 1711
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9713427137044285e-05,
+      "loss": 1.0137,
+      "step": 1712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971290137114898e-05,
+      "loss": 0.8789,
+      "step": 1713
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971237512687954e-05,
+      "loss": 0.9424,
+      "step": 1714
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9711848404248716e-05,
+      "loss": 1.1055,
+      "step": 1715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971132120326931e-05,
+      "loss": 0.9258,
+      "step": 1716
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9710793523954104e-05,
+      "loss": 1.0537,
+      "step": 1717
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97102653663159e-05,
+      "loss": 0.9229,
+      "step": 1718
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970973673036754e-05,
+      "loss": 1.1582,
+      "step": 1719
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970920761612181e-05,
+      "loss": 1.1172,
+      "step": 1720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970867802359159e-05,
+      "loss": 1.1035,
+      "step": 1721
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9708147952789716e-05,
+      "loss": 1.1309,
+      "step": 1722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9707617403729044e-05,
+      "loss": 1.1221,
+      "step": 1723
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970708637642246e-05,
+      "loss": 1.1006,
+      "step": 1724
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9706554870882845e-05,
+      "loss": 1.043,
+      "step": 1725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97060228871231e-05,
+      "loss": 0.9316,
+      "step": 1726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970549042515613e-05,
+      "loss": 1.0059,
+      "step": 1727
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970495748499485e-05,
+      "loss": 1.084,
+      "step": 1728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970442406665222e-05,
+      "loss": 1.0645,
+      "step": 1729
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970389017014116e-05,
+      "loss": 0.9873,
+      "step": 1730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970335579547463e-05,
+      "loss": 1.0127,
+      "step": 1731
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97028209426656e-05,
+      "loss": 0.7783,
+      "step": 1732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9702285611727045e-05,
+      "loss": 1.0459,
+      "step": 1733
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970174980267197e-05,
+      "loss": 1.0537,
+      "step": 1734
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970121351551337e-05,
+      "loss": 1.0693,
+      "step": 1735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970067675026425e-05,
+      "loss": 0.958,
+      "step": 1736
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970013950693764e-05,
+      "loss": 0.9521,
+      "step": 1737
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9699601785546586e-05,
+      "loss": 0.9697,
+      "step": 1738
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969906358610413e-05,
+      "loss": 1.2285,
+      "step": 1739
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969852490862333e-05,
+      "loss": 0.9844,
+      "step": 1740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9697985753117265e-05,
+      "loss": 0.9619,
+      "step": 1741
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9697446119599015e-05,
+      "loss": 1.0459,
+      "step": 1742
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9696906008081676e-05,
+      "loss": 1.001,
+      "step": 1743
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969636541857836e-05,
+      "loss": 1.0049,
+      "step": 1744
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969582435110217e-05,
+      "loss": 1.0215,
+      "step": 1745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9695282805666254e-05,
+      "loss": 1.0557,
+      "step": 1746
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9694740782283744e-05,
+      "loss": 0.9727,
+      "step": 1747
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969419828096779e-05,
+      "loss": 1.1396,
+      "step": 1748
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969365530173157e-05,
+      "loss": 0.998,
+      "step": 1749
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9693111844588245e-05,
+      "loss": 1.1572,
+      "step": 1750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969256790955102e-05,
+      "loss": 1.0498,
+      "step": 1751
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969202349663308e-05,
+      "loss": 1.1719,
+      "step": 1752
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9691478605847645e-05,
+      "loss": 1.1367,
+      "step": 1753
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969093323720794e-05,
+      "loss": 1.0293,
+      "step": 1754
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9690387390727184e-05,
+      "loss": 1.1436,
+      "step": 1755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.968984106641864e-05,
+      "loss": 1.0986,
+      "step": 1756
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.968929426429555e-05,
+      "loss": 1.0938,
+      "step": 1757
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.96887469843712e-05,
+      "loss": 1.2021,
+      "step": 1758
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.968819922665886e-05,
+      "loss": 0.8057,
+      "step": 1759
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9687650991171825e-05,
+      "loss": 0.8604,
+      "step": 1760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.96871022779234e-05,
+      "loss": 0.9795,
+      "step": 1761
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9686553086926904e-05,
+      "loss": 0.9404,
+      "step": 1762
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9686003418195664e-05,
+      "loss": 1.0498,
+      "step": 1763
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9685453271743006e-05,
+      "loss": 0.9941,
+      "step": 1764
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.968490264758229e-05,
+      "loss": 0.9365,
+      "step": 1765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9684351545726884e-05,
+      "loss": 1.1035,
+      "step": 1766
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.968379996619016e-05,
+      "loss": 1.0303,
+      "step": 1767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.968324790898549e-05,
+      "loss": 1.0293,
+      "step": 1768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.968269537412628e-05,
+      "loss": 1.1221,
+      "step": 1769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9682142361625944e-05,
+      "loss": 1.0205,
+      "step": 1770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.968158887149789e-05,
+      "loss": 0.9473,
+      "step": 1771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.968103490375556e-05,
+      "loss": 1.0654,
+      "step": 1772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.968048045841239e-05,
+      "loss": 0.9854,
+      "step": 1773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9679925535481837e-05,
+      "loss": 1.0488,
+      "step": 1774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9679370134977366e-05,
+      "loss": 0.9336,
+      "step": 1775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967881425691246e-05,
+      "loss": 1.1006,
+      "step": 1776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9678257901300605e-05,
+      "loss": 1.0742,
+      "step": 1777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9677701068155304e-05,
+      "loss": 1.082,
+      "step": 1778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967714375749006e-05,
+      "loss": 1.2812,
+      "step": 1779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9676585969318414e-05,
+      "loss": 1.085,
+      "step": 1780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9676027703653896e-05,
+      "loss": 1.0273,
+      "step": 1781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967546896051004e-05,
+      "loss": 0.8721,
+      "step": 1782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967490973990042e-05,
+      "loss": 1.127,
+      "step": 1783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967435004183861e-05,
+      "loss": 0.916,
+      "step": 1784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967378986633818e-05,
+      "loss": 1.0439,
+      "step": 1785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967322921341272e-05,
+      "loss": 0.9307,
+      "step": 1786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9672668083075856e-05,
+      "loss": 0.9023,
+      "step": 1787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967210647534119e-05,
+      "loss": 1.042,
+      "step": 1788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967154439022235e-05,
+      "loss": 0.96,
+      "step": 1789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9670981827732985e-05,
+      "loss": 1.0303,
+      "step": 1790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967041878788673e-05,
+      "loss": 1.0547,
+      "step": 1791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966985527069727e-05,
+      "loss": 1.0088,
+      "step": 1792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966929127617826e-05,
+      "loss": 0.9551,
+      "step": 1793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966872680434341e-05,
+      "loss": 1.0742,
+      "step": 1794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9668161855206396e-05,
+      "loss": 1.0762,
+      "step": 1795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9667596428780934e-05,
+      "loss": 1.0537,
+      "step": 1796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966703052508075e-05,
+      "loss": 0.8799,
+      "step": 1797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966646414411957e-05,
+      "loss": 1.04,
+      "step": 1798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966589728591115e-05,
+      "loss": 1.0078,
+      "step": 1799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966532995046924e-05,
+      "loss": 1.0654,
+      "step": 1800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9664762137807595e-05,
+      "loss": 1.0381,
+      "step": 1801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9664193847940004e-05,
+      "loss": 0.9385,
+      "step": 1802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966362508088026e-05,
+      "loss": 1.0498,
+      "step": 1803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966305583664218e-05,
+      "loss": 0.9912,
+      "step": 1804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966248611523955e-05,
+      "loss": 0.9873,
+      "step": 1805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96619159166862e-05,
+      "loss": 1.0967,
+      "step": 1806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966134524099598e-05,
+      "loss": 0.9541,
+      "step": 1807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966077408818274e-05,
+      "loss": 1.1318,
+      "step": 1808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9660202458260324e-05,
+      "loss": 1.0391,
+      "step": 1809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965963035124261e-05,
+      "loss": 0.9932,
+      "step": 1810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9659057767143493e-05,
+      "loss": 0.9629,
+      "step": 1811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965848470597686e-05,
+      "loss": 1.1777,
+      "step": 1812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965791116775661e-05,
+      "loss": 1.0361,
+      "step": 1813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965733715249667e-05,
+      "loss": 1.0469,
+      "step": 1814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965676266021097e-05,
+      "loss": 1.1387,
+      "step": 1815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965618769091345e-05,
+      "loss": 0.999,
+      "step": 1816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965561224461806e-05,
+      "loss": 1.2041,
+      "step": 1817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9655036321338765e-05,
+      "loss": 0.9873,
+      "step": 1818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9654459921089545e-05,
+      "loss": 1.1484,
+      "step": 1819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9653883043884374e-05,
+      "loss": 1.0186,
+      "step": 1820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965330568973727e-05,
+      "loss": 1.1289,
+      "step": 1821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9652727858662237e-05,
+      "loss": 1.1875,
+      "step": 1822
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965214955067329e-05,
+      "loss": 1.0166,
+      "step": 1823
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965157076578447e-05,
+      "loss": 0.9766,
+      "step": 1824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9650991504009824e-05,
+      "loss": 0.9746,
+      "step": 1825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96504117653634e-05,
+      "loss": 1.0049,
+      "step": 1826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964983154985928e-05,
+      "loss": 1.0469,
+      "step": 1827
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964925085751152e-05,
+      "loss": 1.0566,
+      "step": 1828
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9648669688334244e-05,
+      "loss": 1.0195,
+      "step": 1829
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964808804234154e-05,
+      "loss": 1.1191,
+      "step": 1830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9647505919547516e-05,
+      "loss": 0.9434,
+      "step": 1831
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964692331996631e-05,
+      "loss": 1.0039,
+      "step": 1832
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9646340243612045e-05,
+      "loss": 1.1934,
+      "step": 1833
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9645756690498886e-05,
+      "loss": 1.0078,
+      "step": 1834
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9645172660640984e-05,
+      "loss": 1.1016,
+      "step": 1835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964458815405252e-05,
+      "loss": 0.9648,
+      "step": 1836
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964400317074767e-05,
+      "loss": 1.0645,
+      "step": 1837
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9643417710740644e-05,
+      "loss": 0.9736,
+      "step": 1838
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964283177404563e-05,
+      "loss": 1.1494,
+      "step": 1839
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9642245360676855e-05,
+      "loss": 1.1621,
+      "step": 1840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964165847064856e-05,
+      "loss": 1.1494,
+      "step": 1841
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964107110397497e-05,
+      "loss": 0.9668,
+      "step": 1842
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964048326067035e-05,
+      "loss": 1.1973,
+      "step": 1843
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963989494074897e-05,
+      "loss": 1.1777,
+      "step": 1844
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963930614422508e-05,
+      "loss": 1.1436,
+      "step": 1845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9638716871113e-05,
+      "loss": 0.9199,
+      "step": 1846
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9638127121427014e-05,
+      "loss": 1.1777,
+      "step": 1847
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963753689518144e-05,
+      "loss": 0.9922,
+      "step": 1848
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963694619239059e-05,
+      "loss": 1.0605,
+      "step": 1849
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963635501306881e-05,
+      "loss": 1.1777,
+      "step": 1850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963576335723044e-05,
+      "loss": 1.04,
+      "step": 1851
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9635171224889846e-05,
+      "loss": 1.1133,
+      "step": 1852
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963457861606139e-05,
+      "loss": 1.0439,
+      "step": 1853
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963398553075945e-05,
+      "loss": 1.0059,
+      "step": 1854
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963339196899842e-05,
+      "loss": 0.9795,
+      "step": 1855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963279793079272e-05,
+      "loss": 0.8613,
+      "step": 1856
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963220341615674e-05,
+      "loss": 1.0576,
+      "step": 1857
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963160842510492e-05,
+      "loss": 0.9189,
+      "step": 1858
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96310129576517e-05,
+      "loss": 0.751,
+      "step": 1859
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963041701381153e-05,
+      "loss": 1.0098,
+      "step": 1860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962982059359887e-05,
+      "loss": 0.9736,
+      "step": 1861
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962922369702819e-05,
+      "loss": 1.0322,
+      "step": 1862
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9628626324113986e-05,
+      "loss": 1.1797,
+      "step": 1863
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962802847487074e-05,
+      "loss": 1.0391,
+      "step": 1864
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9627430149312964e-05,
+      "loss": 0.9131,
+      "step": 1865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962683134745519e-05,
+      "loss": 1.1758,
+      "step": 1866
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962623206931193e-05,
+      "loss": 1.0859,
+      "step": 1867
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962563231489775e-05,
+      "loss": 1.0674,
+      "step": 1868
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9625032084227176e-05,
+      "loss": 0.9512,
+      "step": 1869
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9624431377314795e-05,
+      "loss": 1.0947,
+      "step": 1870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9623830194175175e-05,
+      "loss": 1.0293,
+      "step": 1871
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9623228534822905e-05,
+      "loss": 1.0098,
+      "step": 1872
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96226263992726e-05,
+      "loss": 1.1348,
+      "step": 1873
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962202378753885e-05,
+      "loss": 1.1377,
+      "step": 1874
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962142069963629e-05,
+      "loss": 1.1631,
+      "step": 1875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9620817135579555e-05,
+      "loss": 1.0,
+      "step": 1876
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9620213095383295e-05,
+      "loss": 1.1309,
+      "step": 1877
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961960857906216e-05,
+      "loss": 1.0801,
+      "step": 1878
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9619003586630824e-05,
+      "loss": 1.0508,
+      "step": 1879
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9618398118103973e-05,
+      "loss": 1.1055,
+      "step": 1880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961779217349629e-05,
+      "loss": 0.8447,
+      "step": 1881
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961718575282249e-05,
+      "loss": 1.1348,
+      "step": 1882
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961657885609728e-05,
+      "loss": 1.0762,
+      "step": 1883
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961597148333539e-05,
+      "loss": 1.2363,
+      "step": 1884
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961536363455157e-05,
+      "loss": 1.1719,
+      "step": 1885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961475530976055e-05,
+      "loss": 1.1113,
+      "step": 1886
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961414650897711e-05,
+      "loss": 0.8662,
+      "step": 1887
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961353723221601e-05,
+      "loss": 1.0205,
+      "step": 1888
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961292747949205e-05,
+      "loss": 1.1689,
+      "step": 1889
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9612317250820015e-05,
+      "loss": 1.0664,
+      "step": 1890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961170654621472e-05,
+      "loss": 1.1221,
+      "step": 1891
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961109536569099e-05,
+      "loss": 1.1602,
+      "step": 1892
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961048370926364e-05,
+      "loss": 1.1924,
+      "step": 1893
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960987157694753e-05,
+      "loss": 1.0439,
+      "step": 1894
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96092589687575e-05,
+      "loss": 1.085,
+      "step": 1895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960864588470843e-05,
+      "loss": 0.9121,
+      "step": 1896
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9608032324815185e-05,
+      "loss": 1.0283,
+      "step": 1897
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960741828909266e-05,
+      "loss": 1.126,
+      "step": 1898
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9606803777555764e-05,
+      "loss": 1.209,
+      "step": 1899
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96061887902194e-05,
+      "loss": 1.0996,
+      "step": 1900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960557332709849e-05,
+      "loss": 0.9219,
+      "step": 1901
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960495738820797e-05,
+      "loss": 1.1289,
+      "step": 1902
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96043409735628e-05,
+      "loss": 0.9902,
+      "step": 1903
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960372408317793e-05,
+      "loss": 1.1074,
+      "step": 1904
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9603106717068315e-05,
+      "loss": 0.9883,
+      "step": 1905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960248887524896e-05,
+      "loss": 1.041,
+      "step": 1906
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960187055773485e-05,
+      "loss": 1.0283,
+      "step": 1907
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960125176454099e-05,
+      "loss": 1.1572,
+      "step": 1908
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960063249568239e-05,
+      "loss": 1.0977,
+      "step": 1909
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9600012751174086e-05,
+      "loss": 1.291,
+      "step": 1910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959939253103111e-05,
+      "loss": 0.9268,
+      "step": 1911
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959877183526851e-05,
+      "loss": 0.917,
+      "step": 1912
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9598150663901374e-05,
+      "loss": 0.9561,
+      "step": 1913
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9597529016944746e-05,
+      "loss": 1.0605,
+      "step": 1914
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9596906894413724e-05,
+      "loss": 1.1992,
+      "step": 1915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9596284296323406e-05,
+      "loss": 1.0195,
+      "step": 1916
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9595661222688896e-05,
+      "loss": 1.0801,
+      "step": 1917
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959503767352532e-05,
+      "loss": 1.126,
+      "step": 1918
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959441364884781e-05,
+      "loss": 0.9336,
+      "step": 1919
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.95937891486715e-05,
+      "loss": 0.8789,
+      "step": 1920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959316417301155e-05,
+      "loss": 0.9775,
+      "step": 1921
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9592538721883126e-05,
+      "loss": 1.1309,
+      "step": 1922
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9591912795301415e-05,
+      "loss": 1.0801,
+      "step": 1923
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959128639328159e-05,
+      "loss": 0.7739,
+      "step": 1924
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9590659515838865e-05,
+      "loss": 1.1006,
+      "step": 1925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9590032162988444e-05,
+      "loss": 1.1484,
+      "step": 1926
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958940433474555e-05,
+      "loss": 1.2578,
+      "step": 1927
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9588776031125436e-05,
+      "loss": 0.9082,
+      "step": 1928
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958814725214333e-05,
+      "loss": 1.0977,
+      "step": 1929
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9587517997814495e-05,
+      "loss": 1.0,
+      "step": 1930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9586888268154205e-05,
+      "loss": 0.8633,
+      "step": 1931
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958625806317774e-05,
+      "loss": 0.96,
+      "step": 1932
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9585627382900386e-05,
+      "loss": 0.9888,
+      "step": 1933
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958499622733747e-05,
+      "loss": 1.0547,
+      "step": 1934
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958436459650428e-05,
+      "loss": 0.8828,
+      "step": 1935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958373249041616e-05,
+      "loss": 1.0713,
+      "step": 1936
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958309990908845e-05,
+      "loss": 1.168,
+      "step": 1937
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958246685253649e-05,
+      "loss": 1.1914,
+      "step": 1938
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958183332077565e-05,
+      "loss": 1.1211,
+      "step": 1939
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9581199313821306e-05,
+      "loss": 1.0605,
+      "step": 1940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958056483168885e-05,
+      "loss": 0.9746,
+      "step": 1941
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957992987439366e-05,
+      "loss": 0.9609,
+      "step": 1942
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9579294441951156e-05,
+      "loss": 1.0039,
+      "step": 1943
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9578658534376755e-05,
+      "loss": 1.0195,
+      "step": 1944
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9578022151685894e-05,
+      "loss": 0.9297,
+      "step": 1945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9577385293894e-05,
+      "loss": 0.9424,
+      "step": 1946
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957674796101656e-05,
+      "loss": 1.0801,
+      "step": 1947
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957611015306901e-05,
+      "loss": 0.9414,
+      "step": 1948
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957547187006683e-05,
+      "loss": 1.0264,
+      "step": 1949
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957483311202553e-05,
+      "loss": 1.1875,
+      "step": 1950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957419387896059e-05,
+      "loss": 1.1709,
+      "step": 1951
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957355417088753e-05,
+      "loss": 1.1836,
+      "step": 1952
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9572913987821875e-05,
+      "loss": 1.0059,
+      "step": 1953
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9572273329779166e-05,
+      "loss": 0.9326,
+      "step": 1954
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9571632196774935e-05,
+      "loss": 1.0889,
+      "step": 1955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9570990588824756e-05,
+      "loss": 1.1777,
+      "step": 1956
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9570348505944185e-05,
+      "loss": 1.0596,
+      "step": 1957
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9569705948148815e-05,
+      "loss": 0.9414,
+      "step": 1958
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956906291545423e-05,
+      "loss": 1.1172,
+      "step": 1959
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956841940787604e-05,
+      "loss": 1.083,
+      "step": 1960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956777542542986e-05,
+      "loss": 1.0645,
+      "step": 1961
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9567130968131315e-05,
+      "loss": 1.0361,
+      "step": 1962
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956648603599606e-05,
+      "loss": 1.0762,
+      "step": 1963
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956584062903972e-05,
+      "loss": 1.1748,
+      "step": 1964
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956519474727796e-05,
+      "loss": 1.0938,
+      "step": 1965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9564548390726475e-05,
+      "loss": 1.0176,
+      "step": 1966
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956390155940094e-05,
+      "loss": 1.0625,
+      "step": 1967
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956325425331704e-05,
+      "loss": 1.0078,
+      "step": 1968
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.95626064724905e-05,
+      "loss": 0.8809,
+      "step": 1969
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956195821693703e-05,
+      "loss": 0.9766,
+      "step": 1970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956130948667236e-05,
+      "loss": 0.7861,
+      "step": 1971
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956066028171224e-05,
+      "loss": 1.0898,
+      "step": 1972
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9560010602072416e-05,
+      "loss": 0.9854,
+      "step": 1973
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.955936044776867e-05,
+      "loss": 1.0127,
+      "step": 1974
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.955870981881676e-05,
+      "loss": 0.9385,
+      "step": 1975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955805871523248e-05,
+      "loss": 1.1641,
+      "step": 1976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955740713703164e-05,
+      "loss": 1.1934,
+      "step": 1977
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9556755084230045e-05,
+      "loss": 1.1172,
+      "step": 1978
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9556102556843516e-05,
+      "loss": 1.0303,
+      "step": 1979
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955544955488789e-05,
+      "loss": 1.0078,
+      "step": 1980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955479607837901e-05,
+      "loss": 1.0068,
+      "step": 1981
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9554142127332745e-05,
+      "loss": 1.1162,
+      "step": 1982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955348770176495e-05,
+      "loss": 0.9834,
+      "step": 1983
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955283280169152e-05,
+      "loss": 1.0527,
+      "step": 1984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955217742712834e-05,
+      "loss": 1.1094,
+      "step": 1985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955152157809131e-05,
+      "loss": 1.0186,
+      "step": 1986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955086525459635e-05,
+      "loss": 1.2012,
+      "step": 1987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.95502084566594e-05,
+      "loss": 1.0322,
+      "step": 1988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954955118429637e-05,
+      "loss": 1.1309,
+      "step": 1989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954889343752323e-05,
+      "loss": 1.1211,
+      "step": 1990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9548235216355936e-05,
+      "loss": 1.042,
+      "step": 1991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9547576520810474e-05,
+      "loss": 0.9639,
+      "step": 1992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954691735090281e-05,
+      "loss": 1.0898,
+      "step": 1993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954625770664894e-05,
+      "loss": 0.8252,
+      "step": 1994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954559758806489e-05,
+      "loss": 1.0703,
+      "step": 1995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954493699516666e-05,
+      "loss": 1.085,
+      "step": 1996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.95442759279703e-05,
+      "loss": 1.1133,
+      "step": 1997
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954361438649183e-05,
+      "loss": 1.0439,
+      "step": 1998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9542952370747317e-05,
+      "loss": 1.0117,
+      "step": 1999
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9542289880752827e-05,
+      "loss": 0.8105,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9541626916524423e-05,
+      "loss": 1.1055,
+      "step": 2001
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954096347807822e-05,
+      "loss": 0.9941,
+      "step": 2002
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954029956543029e-05,
+      "loss": 0.7393,
+      "step": 2003
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9539635178596754e-05,
+      "loss": 1.1094,
+      "step": 2004
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953897031759374e-05,
+      "loss": 0.9932,
+      "step": 2005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953830498243738e-05,
+      "loss": 1.0381,
+      "step": 2006
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953763917314381e-05,
+      "loss": 0.9912,
+      "step": 2007
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953697288972921e-05,
+      "loss": 1.125,
+      "step": 2008
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9536306132209726e-05,
+      "loss": 0.9043,
+      "step": 2009
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9535638900601546e-05,
+      "loss": 1.2061,
+      "step": 2010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953497119492086e-05,
+      "loss": 0.9497,
+      "step": 2011
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9534303015183886e-05,
+      "loss": 1.084,
+      "step": 2012
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953363436140682e-05,
+      "loss": 1.0312,
+      "step": 2013
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9532965233605886e-05,
+      "loss": 1.0059,
+      "step": 2014
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9532295631797336e-05,
+      "loss": 0.9746,
+      "step": 2015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953162555599741e-05,
+      "loss": 1.0859,
+      "step": 2016
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9530955006222374e-05,
+      "loss": 1.2988,
+      "step": 2017
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.95302839824885e-05,
+      "loss": 1.1504,
+      "step": 2018
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952961248481207e-05,
+      "loss": 1.1152,
+      "step": 2019
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952894051320938e-05,
+      "loss": 1.1123,
+      "step": 2020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9528268067696735e-05,
+      "loss": 1.1104,
+      "step": 2021
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9527595148290455e-05,
+      "loss": 1.1152,
+      "step": 2022
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952692175500687e-05,
+      "loss": 1.041,
+      "step": 2023
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952624788786232e-05,
+      "loss": 1.0283,
+      "step": 2024
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952557354687316e-05,
+      "loss": 1.0234,
+      "step": 2025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9524898732055746e-05,
+      "loss": 1.2461,
+      "step": 2026
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9524223443426464e-05,
+      "loss": 1.1289,
+      "step": 2027
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.95235476810017e-05,
+      "loss": 1.0703,
+      "step": 2028
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952287144479785e-05,
+      "loss": 1.0986,
+      "step": 2029
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952219473483133e-05,
+      "loss": 0.9268,
+      "step": 2030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952151755111855e-05,
+      "loss": 1.001,
+      "step": 2031
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9520839893675944e-05,
+      "loss": 1.0762,
+      "step": 2032
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952016176251997e-05,
+      "loss": 1.1514,
+      "step": 2033
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951948315766708e-05,
+      "loss": 0.8682,
+      "step": 2034
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951880407913373e-05,
+      "loss": 0.8457,
+      "step": 2035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9518124526936414e-05,
+      "loss": 1.207,
+      "step": 2036
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9517444501091616e-05,
+      "loss": 1.042,
+      "step": 2037
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951676400161583e-05,
+      "loss": 0.7373,
+      "step": 2038
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951608302852559e-05,
+      "loss": 1.0332,
+      "step": 2039
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951540158183741e-05,
+      "loss": 1.0488,
+      "step": 2040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9514719661567814e-05,
+      "loss": 1.0869,
+      "step": 2041
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9514037267733375e-05,
+      "loss": 1.1025,
+      "step": 2042
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951335440035063e-05,
+      "loss": 0.8057,
+      "step": 2043
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951267105943617e-05,
+      "loss": 0.9082,
+      "step": 2044
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9511987245006565e-05,
+      "loss": 1.0342,
+      "step": 2045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9511302957078405e-05,
+      "loss": 1.0127,
+      "step": 2046
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951061819566832e-05,
+      "loss": 1.1367,
+      "step": 2047
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9509932960792894e-05,
+      "loss": 0.8623,
+      "step": 2048
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9509247252468774e-05,
+      "loss": 1.0801,
+      "step": 2049
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9508561070712604e-05,
+      "loss": 1.248,
+      "step": 2050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9507874415541025e-05,
+      "loss": 0.9805,
+      "step": 2051
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.95071872869707e-05,
+      "loss": 1.0508,
+      "step": 2052
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.950649968501832e-05,
+      "loss": 0.8926,
+      "step": 2053
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.950581160970055e-05,
+      "loss": 1.0918,
+      "step": 2054
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.95051230610341e-05,
+      "loss": 1.0762,
+      "step": 2055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9504434039035676e-05,
+      "loss": 1.0498,
+      "step": 2056
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9503744543721995e-05,
+      "loss": 1.0264,
+      "step": 2057
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9503054575109796e-05,
+      "loss": 0.9258,
+      "step": 2058
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.950236413321582e-05,
+      "loss": 1.1143,
+      "step": 2059
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.950167321805682e-05,
+      "loss": 0.79,
+      "step": 2060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.950098182964956e-05,
+      "loss": 1.1602,
+      "step": 2061
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.950028996801082e-05,
+      "loss": 1.0244,
+      "step": 2062
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.94995976331574e-05,
+      "loss": 1.0557,
+      "step": 2063
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949890482510608e-05,
+      "loss": 0.9697,
+      "step": 2064
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.94982115438737e-05,
+      "loss": 0.8232,
+      "step": 2065
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949751778947706e-05,
+      "loss": 1.0439,
+      "step": 2066
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9496823561932995e-05,
+      "loss": 1.0186,
+      "step": 2067
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949612886125837e-05,
+      "loss": 0.9551,
+      "step": 2068
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949543368747003e-05,
+      "loss": 1.1758,
+      "step": 2069
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949473804058485e-05,
+      "loss": 1.1172,
+      "step": 2070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9494041920619704e-05,
+      "loss": 1.0615,
+      "step": 2071
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9493345327591497e-05,
+      "loss": 1.1875,
+      "step": 2072
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9492648261517126e-05,
+      "loss": 1.0186,
+      "step": 2073
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949195072241351e-05,
+      "loss": 1.2031,
+      "step": 2074
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9491252710297565e-05,
+      "loss": 0.9282,
+      "step": 2075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949055422518624e-05,
+      "loss": 0.9521,
+      "step": 2076
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948985526709649e-05,
+      "loss": 0.8945,
+      "step": 2077
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9489155836045275e-05,
+      "loss": 0.9229,
+      "step": 2078
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948845593204955e-05,
+      "loss": 1.0361,
+      "step": 2079
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948775555512632e-05,
+      "loss": 1.0303,
+      "step": 2080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948705470529257e-05,
+      "loss": 1.0205,
+      "step": 2081
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948635338256532e-05,
+      "loss": 1.0273,
+      "step": 2082
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948565158696157e-05,
+      "loss": 1.1191,
+      "step": 2083
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948494931849837e-05,
+      "loss": 1.0791,
+      "step": 2084
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948424657719275e-05,
+      "loss": 1.0039,
+      "step": 2085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948354336306176e-05,
+      "loss": 1.2227,
+      "step": 2086
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948283967612248e-05,
+      "loss": 1.1084,
+      "step": 2087
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948213551639198e-05,
+      "loss": 1.2227,
+      "step": 2088
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948143088388734e-05,
+      "loss": 0.9902,
+      "step": 2089
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948072577862567e-05,
+      "loss": 1.1309,
+      "step": 2090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948002020062407e-05,
+      "loss": 0.9814,
+      "step": 2091
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947931414989967e-05,
+      "loss": 1.1758,
+      "step": 2092
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9478607626469604e-05,
+      "loss": 0.9277,
+      "step": 2093
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947790063035102e-05,
+      "loss": 1.124,
+      "step": 2094
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947719316156106e-05,
+      "loss": 1.1426,
+      "step": 2095
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947648522011691e-05,
+      "loss": 1.126,
+      "step": 2096
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947577680603574e-05,
+      "loss": 0.9512,
+      "step": 2097
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9475067919334745e-05,
+      "loss": 0.9541,
+      "step": 2098
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947435856003112e-05,
+      "loss": 1.1016,
+      "step": 2099
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9473648728142094e-05,
+      "loss": 0.9814,
+      "step": 2100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9472938423684874e-05,
+      "loss": 1.1201,
+      "step": 2101
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9472227646676714e-05,
+      "loss": 0.8496,
+      "step": 2102
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9471516397134853e-05,
+      "loss": 0.8013,
+      "step": 2103
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947080467507655e-05,
+      "loss": 1.0264,
+      "step": 2104
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947009248051908e-05,
+      "loss": 1.0234,
+      "step": 2105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9469379813479724e-05,
+      "loss": 1.0566,
+      "step": 2106
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.946866667397577e-05,
+      "loss": 0.9521,
+      "step": 2107
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9467953062024544e-05,
+      "loss": 1.334,
+      "step": 2108
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9467238977643345e-05,
+      "loss": 1.0508,
+      "step": 2109
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9466524420849504e-05,
+      "loss": 0.9141,
+      "step": 2110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.946580939166037e-05,
+      "loss": 0.8203,
+      "step": 2111
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.946509389009329e-05,
+      "loss": 1.1953,
+      "step": 2112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9464377916165627e-05,
+      "loss": 1.0762,
+      "step": 2113
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9463661469894746e-05,
+      "loss": 1.0166,
+      "step": 2114
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9462944551298046e-05,
+      "loss": 1.1494,
+      "step": 2115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9462227160392916e-05,
+      "loss": 0.9355,
+      "step": 2116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9461509297196776e-05,
+      "loss": 1.1035,
+      "step": 2117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.946079096172704e-05,
+      "loss": 1.0859,
+      "step": 2118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9460072154001136e-05,
+      "loss": 1.0596,
+      "step": 2119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945935287403651e-05,
+      "loss": 1.084,
+      "step": 2120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945863312185062e-05,
+      "loss": 0.9678,
+      "step": 2121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945791289746093e-05,
+      "loss": 0.9844,
+      "step": 2122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945719220088492e-05,
+      "loss": 0.9531,
+      "step": 2123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9456471032140074e-05,
+      "loss": 1.1064,
+      "step": 2124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945574939124389e-05,
+      "loss": 1.0957,
+      "step": 2125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9455027278213895e-05,
+      "loss": 0.8896,
+      "step": 2126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9454304693067596e-05,
+      "loss": 1.1699,
+      "step": 2127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945358163582254e-05,
+      "loss": 0.9443,
+      "step": 2128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945285810649626e-05,
+      "loss": 1.0566,
+      "step": 2129
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9452134105106334e-05,
+      "loss": 1.0137,
+      "step": 2130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945140963167031e-05,
+      "loss": 1.0156,
+      "step": 2131
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945068468620579e-05,
+      "loss": 0.9211,
+      "step": 2132
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944995926873035e-05,
+      "loss": 1.0645,
+      "step": 2133
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9449233379261596e-05,
+      "loss": 1.1162,
+      "step": 2134
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944850701781715e-05,
+      "loss": 1.2344,
+      "step": 2135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9447780184414635e-05,
+      "loss": 1.0635,
+      "step": 2136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944705287907168e-05,
+      "loss": 1.0049,
+      "step": 2137
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944632510180595e-05,
+      "loss": 1.0732,
+      "step": 2138
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9445596852635104e-05,
+      "loss": 1.0986,
+      "step": 2139
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9444868131576795e-05,
+      "loss": 1.0664,
+      "step": 2140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944413893864874e-05,
+      "loss": 1.0918,
+      "step": 2141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.94434092738686e-05,
+      "loss": 1.1289,
+      "step": 2142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944267913725411e-05,
+      "loss": 1.0996,
+      "step": 2143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944194852882297e-05,
+      "loss": 1.123,
+      "step": 2144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944121744859292e-05,
+      "loss": 1.0645,
+      "step": 2145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944048589658169e-05,
+      "loss": 1.124,
+      "step": 2146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943975387280705e-05,
+      "loss": 0.9951,
+      "step": 2147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943902137728675e-05,
+      "loss": 1.1348,
+      "step": 2148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943828841003856e-05,
+      "loss": 0.998,
+      "step": 2149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943755497108029e-05,
+      "loss": 1.2227,
+      "step": 2150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9436821060429715e-05,
+      "loss": 1.0859,
+      "step": 2151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943608667810466e-05,
+      "loss": 0.9756,
+      "step": 2152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943535182412294e-05,
+      "loss": 1.0596,
+      "step": 2153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943461649850238e-05,
+      "loss": 0.9707,
+      "step": 2154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943388070126085e-05,
+      "loss": 1.1836,
+      "step": 2155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943314443241617e-05,
+      "loss": 0.8506,
+      "step": 2156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9432407691986234e-05,
+      "loss": 1.1387,
+      "step": 2157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9431670479988914e-05,
+      "loss": 1.1357,
+      "step": 2158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.94309327964421e-05,
+      "loss": 1.1133,
+      "step": 2159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943019464136368e-05,
+      "loss": 1.0596,
+      "step": 2160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942945601477159e-05,
+      "loss": 0.9365,
+      "step": 2161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9428716916683736e-05,
+      "loss": 1.0713,
+      "step": 2162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942797734711806e-05,
+      "loss": 0.9453,
+      "step": 2163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9427237306092515e-05,
+      "loss": 0.8887,
+      "step": 2164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942649679362505e-05,
+      "loss": 1.0605,
+      "step": 2165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9425755809733636e-05,
+      "loss": 1.2441,
+      "step": 2166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942501435443626e-05,
+      "loss": 1.1445,
+      "step": 2167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942427242775092e-05,
+      "loss": 1.0244,
+      "step": 2168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942353002969561e-05,
+      "loss": 0.9619,
+      "step": 2169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9422787160288337e-05,
+      "loss": 1.2109,
+      "step": 2170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942204381954715e-05,
+      "loss": 1.0391,
+      "step": 2171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942130000749008e-05,
+      "loss": 0.8994,
+      "step": 2172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9420555724135165e-05,
+      "loss": 1.0459,
+      "step": 2173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9419810969500487e-05,
+      "loss": 1.168,
+      "step": 2174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.94190657436041e-05,
+      "loss": 1.1084,
+      "step": 2175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9418320046464104e-05,
+      "loss": 0.8828,
+      "step": 2176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9417573878098585e-05,
+      "loss": 1.1934,
+      "step": 2177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.941682723852565e-05,
+      "loss": 0.9834,
+      "step": 2178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.941608012776342e-05,
+      "loss": 1.1797,
+      "step": 2179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.941533254583003e-05,
+      "loss": 1.0391,
+      "step": 2180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.941458449274362e-05,
+      "loss": 0.8965,
+      "step": 2181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9413835968522335e-05,
+      "loss": 1.0234,
+      "step": 2182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.941308697318435e-05,
+      "loss": 0.8057,
+      "step": 2183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9412337506747825e-05,
+      "loss": 1.1143,
+      "step": 2184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9411587569230974e-05,
+      "loss": 1.1328,
+      "step": 2185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.941083716065197e-05,
+      "loss": 1.001,
+      "step": 2186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.941008628102904e-05,
+      "loss": 0.9053,
+      "step": 2187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940933493038039e-05,
+      "loss": 1.1338,
+      "step": 2188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940858310872426e-05,
+      "loss": 0.9961,
+      "step": 2189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940783081607891e-05,
+      "loss": 1.1055,
+      "step": 2190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9407078052462575e-05,
+      "loss": 1.0371,
+      "step": 2191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940632481789353e-05,
+      "loss": 0.9395,
+      "step": 2192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940557111239005e-05,
+      "loss": 1.1973,
+      "step": 2193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940481693597044e-05,
+      "loss": 1.1934,
+      "step": 2194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940406228865298e-05,
+      "loss": 1.0195,
+      "step": 2195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9403307170456e-05,
+      "loss": 1.0938,
+      "step": 2196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940255158139781e-05,
+      "loss": 1.2051,
+      "step": 2197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9401795521496765e-05,
+      "loss": 1.0186,
+      "step": 2198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9401038990771194e-05,
+      "loss": 1.1348,
+      "step": 2199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940028198923946e-05,
+      "loss": 1.2148,
+      "step": 2200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939952451691994e-05,
+      "loss": 1.0303,
+      "step": 2201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939876657383102e-05,
+      "loss": 1.0,
+      "step": 2202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939800815999108e-05,
+      "loss": 1.1182,
+      "step": 2203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9397249275418526e-05,
+      "loss": 1.043,
+      "step": 2204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939648992013178e-05,
+      "loss": 0.9492,
+      "step": 2205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9395730094149266e-05,
+      "loss": 1.0986,
+      "step": 2206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939496979748942e-05,
+      "loss": 1.0498,
+      "step": 2207
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93942090301707e-05,
+      "loss": 1.3418,
+      "step": 2208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9393447792211555e-05,
+      "loss": 0.96,
+      "step": 2209
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939268608363048e-05,
+      "loss": 1.1211,
+      "step": 2210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939192390444593e-05,
+      "loss": 1.1836,
+      "step": 2211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9391161254676424e-05,
+      "loss": 1.0879,
+      "step": 2212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939039813434047e-05,
+      "loss": 0.9082,
+      "step": 2213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938963454345657e-05,
+      "loss": 1.1572,
+      "step": 2214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938887048204326e-05,
+      "loss": 0.9033,
+      "step": 2215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938810595011909e-05,
+      "loss": 0.9092,
+      "step": 2216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93873409477026e-05,
+      "loss": 1.1016,
+      "step": 2217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938657547481237e-05,
+      "loss": 1.0508,
+      "step": 2218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938580953146696e-05,
+      "loss": 0.8643,
+      "step": 2219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938504311768497e-05,
+      "loss": 1.1113,
+      "step": 2220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938427623348499e-05,
+      "loss": 1.1387,
+      "step": 2221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938350887888563e-05,
+      "loss": 1.2383,
+      "step": 2222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938274105390551e-05,
+      "loss": 1.1104,
+      "step": 2223
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938197275856328e-05,
+      "loss": 1.0996,
+      "step": 2224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9381203992877563e-05,
+      "loss": 1.0352,
+      "step": 2225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938043475686703e-05,
+      "loss": 0.9316,
+      "step": 2226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937966505055033e-05,
+      "loss": 1.0117,
+      "step": 2227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9378894873946166e-05,
+      "loss": 1.0029,
+      "step": 2228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937812422707321e-05,
+      "loss": 1.1094,
+      "step": 2229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937735310995016e-05,
+      "loss": 1.0508,
+      "step": 2230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937658152259575e-05,
+      "loss": 1.127,
+      "step": 2231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937580946502868e-05,
+      "loss": 1.0996,
+      "step": 2232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93750369372677e-05,
+      "loss": 0.9092,
+      "step": 2233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937426393933156e-05,
+      "loss": 1.0693,
+      "step": 2234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9373490471239e-05,
+      "loss": 1.0625,
+      "step": 2235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937271653300881e-05,
+      "loss": 1.1582,
+      "step": 2236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937194212465976e-05,
+      "loss": 0.8701,
+      "step": 2237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937116724621064e-05,
+      "loss": 0.9922,
+      "step": 2238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937039189768027e-05,
+      "loss": 1.0469,
+      "step": 2239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.936961607908745e-05,
+      "loss": 0.8818,
+      "step": 2240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.936883979045101e-05,
+      "loss": 1.1328,
+      "step": 2241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9368063031789796e-05,
+      "loss": 0.9971,
+      "step": 2242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.936728580312265e-05,
+      "loss": 1.168,
+      "step": 2243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9366508104468434e-05,
+      "loss": 1.041,
+      "step": 2244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.936572993584601e-05,
+      "loss": 0.9355,
+      "step": 2245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.936495129727429e-05,
+      "loss": 1.3086,
+      "step": 2246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.936417218877214e-05,
+      "loss": 0.959,
+      "step": 2247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9363392610358486e-05,
+      "loss": 0.9219,
+      "step": 2248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9362612562052244e-05,
+      "loss": 0.8608,
+      "step": 2249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.936183204387232e-05,
+      "loss": 1.0107,
+      "step": 2250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.936105105583769e-05,
+      "loss": 1.0684,
+      "step": 2251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9360269597967285e-05,
+      "loss": 1.1904,
+      "step": 2252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9359487670280074e-05,
+      "loss": 1.126,
+      "step": 2253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935870527279504e-05,
+      "loss": 1.1953,
+      "step": 2254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9357922405531143e-05,
+      "loss": 0.9834,
+      "step": 2255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935713906850741e-05,
+      "loss": 1.0596,
+      "step": 2256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9356355261742835e-05,
+      "loss": 0.8457,
+      "step": 2257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935557098525644e-05,
+      "loss": 1.0693,
+      "step": 2258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935478623906726e-05,
+      "loss": 1.0859,
+      "step": 2259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935400102319435e-05,
+      "loss": 1.041,
+      "step": 2260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9353215337656744e-05,
+      "loss": 1.0234,
+      "step": 2261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935242918247352e-05,
+      "loss": 1.0811,
+      "step": 2262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935164255766375e-05,
+      "loss": 1.0625,
+      "step": 2263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9350855463246534e-05,
+      "loss": 1.1846,
+      "step": 2264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935006789924096e-05,
+      "loss": 1.0566,
+      "step": 2265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9349279865666136e-05,
+      "loss": 1.1094,
+      "step": 2266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934849136254121e-05,
+      "loss": 0.9785,
+      "step": 2267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934770238988529e-05,
+      "loss": 1.0996,
+      "step": 2268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934691294771753e-05,
+      "loss": 1.166,
+      "step": 2269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93461230360571e-05,
+      "loss": 0.8955,
+      "step": 2270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934533265492315e-05,
+      "loss": 1.0635,
+      "step": 2271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934454180433488e-05,
+      "loss": 0.9219,
+      "step": 2272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934375048431146e-05,
+      "loss": 0.9561,
+      "step": 2273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93429586948721e-05,
+      "loss": 0.7695,
+      "step": 2274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934216643603603e-05,
+      "loss": 1.0898,
+      "step": 2275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934137370782246e-05,
+      "loss": 0.9619,
+      "step": 2276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934058051025063e-05,
+      "loss": 1.1406,
+      "step": 2277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933978684333979e-05,
+      "loss": 1.0381,
+      "step": 2278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93389927071092e-05,
+      "loss": 1.0996,
+      "step": 2279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9338198101578126e-05,
+      "loss": 0.9238,
+      "step": 2280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933740302676586e-05,
+      "loss": 1.0029,
+      "step": 2281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933660748269169e-05,
+      "loss": 1.1152,
+      "step": 2282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9335811469374926e-05,
+      "loss": 0.917,
+      "step": 2283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933501498683488e-05,
+      "loss": 1.0215,
+      "step": 2284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9334218035090884e-05,
+      "loss": 1.0908,
+      "step": 2285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933342061416228e-05,
+      "loss": 0.9893,
+      "step": 2286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9332622724068406e-05,
+      "loss": 0.9473,
+      "step": 2287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933182436482864e-05,
+      "loss": 0.8906,
+      "step": 2288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933102553646235e-05,
+      "loss": 1.1885,
+      "step": 2289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933022623898892e-05,
+      "loss": 0.9902,
+      "step": 2290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932942647242774e-05,
+      "loss": 1.04,
+      "step": 2291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932862623679824e-05,
+      "loss": 1.0879,
+      "step": 2292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932782553211981e-05,
+      "loss": 1.1719,
+      "step": 2293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93270243584119e-05,
+      "loss": 1.0273,
+      "step": 2294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932622271569395e-05,
+      "loss": 1.0391,
+      "step": 2295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932542060398541e-05,
+      "loss": 1.0391,
+      "step": 2296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932461802330575e-05,
+      "loss": 1.0225,
+      "step": 2297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932381497367444e-05,
+      "loss": 1.1426,
+      "step": 2298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9323011455110965e-05,
+      "loss": 1.127,
+      "step": 2299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9322207467634835e-05,
+      "loss": 1.0303,
+      "step": 2300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9321403011265546e-05,
+      "loss": 0.9443,
+      "step": 2301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932059808602263e-05,
+      "loss": 1.2246,
+      "step": 2302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.931979269192563e-05,
+      "loss": 0.8447,
+      "step": 2303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.931898682899407e-05,
+      "loss": 1.0088,
+      "step": 2304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.931818049724752e-05,
+      "loss": 1.2363,
+      "step": 2305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9317373696705535e-05,
+      "loss": 0.958,
+      "step": 2306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93165664273877e-05,
+      "loss": 1.1455,
+      "step": 2307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.931575868931361e-05,
+      "loss": 0.9531,
+      "step": 2308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.931495048250286e-05,
+      "loss": 1.0635,
+      "step": 2309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.931414180697507e-05,
+      "loss": 1.1289,
+      "step": 2310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.931333266274985e-05,
+      "loss": 0.9404,
+      "step": 2311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9312523049846855e-05,
+      "loss": 1.0156,
+      "step": 2312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.931171296828572e-05,
+      "loss": 1.1641,
+      "step": 2313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93109024180861e-05,
+      "loss": 1.0781,
+      "step": 2314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.931009139926767e-05,
+      "loss": 1.0674,
+      "step": 2315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930927991185011e-05,
+      "loss": 1.0869,
+      "step": 2316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9308467955853114e-05,
+      "loss": 0.8799,
+      "step": 2317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930765553129639e-05,
+      "loss": 1.167,
+      "step": 2318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930684263819964e-05,
+      "loss": 0.8496,
+      "step": 2319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930602927658261e-05,
+      "loss": 1.0352,
+      "step": 2320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9305215446465014e-05,
+      "loss": 0.8545,
+      "step": 2321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930440114786662e-05,
+      "loss": 1.001,
+      "step": 2322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9303586380807185e-05,
+      "loss": 0.8799,
+      "step": 2323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930277114530647e-05,
+      "loss": 1.0,
+      "step": 2324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9301955441384276e-05,
+      "loss": 0.9443,
+      "step": 2325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930113926906038e-05,
+      "loss": 1.0508,
+      "step": 2326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93003226283546e-05,
+      "loss": 1.0449,
+      "step": 2327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9299505519286754e-05,
+      "loss": 0.9307,
+      "step": 2328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929868794187666e-05,
+      "loss": 0.9648,
+      "step": 2329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929786989614417e-05,
+      "loss": 1.1084,
+      "step": 2330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9297051382109125e-05,
+      "loss": 1.1855,
+      "step": 2331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9296232399791405e-05,
+      "loss": 1.291,
+      "step": 2332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929541294921086e-05,
+      "loss": 1.0586,
+      "step": 2333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.92945930303874e-05,
+      "loss": 1.1172,
+      "step": 2334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.92937726433409e-05,
+      "loss": 1.0996,
+      "step": 2335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929295178809128e-05,
+      "loss": 1.1582,
+      "step": 2336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929213046465846e-05,
+      "loss": 0.998,
+      "step": 2337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929130867306238e-05,
+      "loss": 0.9521,
+      "step": 2338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929048641332296e-05,
+      "loss": 0.9297,
+      "step": 2339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928966368546017e-05,
+      "loss": 1.001,
+      "step": 2340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928884048949397e-05,
+      "loss": 1.1328,
+      "step": 2341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928801682544434e-05,
+      "loss": 1.1621,
+      "step": 2342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928719269333127e-05,
+      "loss": 1.166,
+      "step": 2343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9286368093174745e-05,
+      "loss": 1.0449,
+      "step": 2344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928554302499479e-05,
+      "loss": 1.2021,
+      "step": 2345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928471748881142e-05,
+      "loss": 1.0664,
+      "step": 2346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9283891484644673e-05,
+      "loss": 1.1279,
+      "step": 2347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9283065012514594e-05,
+      "loss": 0.937,
+      "step": 2348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928223807244123e-05,
+      "loss": 1.1309,
+      "step": 2349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928141066444466e-05,
+      "loss": 1.0684,
+      "step": 2350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928058278854496e-05,
+      "loss": 0.9551,
+      "step": 2351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.927975444476221e-05,
+      "loss": 1.1943,
+      "step": 2352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.927892563311653e-05,
+      "loss": 1.1641,
+      "step": 2353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.927809635362802e-05,
+      "loss": 0.9746,
+      "step": 2354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.92772666063168e-05,
+      "loss": 0.9883,
+      "step": 2355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9276436391203014e-05,
+      "loss": 1.0195,
+      "step": 2356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.927560570830681e-05,
+      "loss": 1.0762,
+      "step": 2357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9274774557648336e-05,
+      "loss": 1.1699,
+      "step": 2358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9273942939247776e-05,
+      "loss": 0.9775,
+      "step": 2359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9273110853125305e-05,
+      "loss": 1.0615,
+      "step": 2360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.92722782993011e-05,
+      "loss": 0.8623,
+      "step": 2361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.927144527779539e-05,
+      "loss": 0.8984,
+      "step": 2362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.927061178862838e-05,
+      "loss": 1.2578,
+      "step": 2363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926977783182029e-05,
+      "loss": 1.0615,
+      "step": 2364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926894340739136e-05,
+      "loss": 0.8477,
+      "step": 2365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926810851536185e-05,
+      "loss": 0.8838,
+      "step": 2366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926727315575201e-05,
+      "loss": 0.9912,
+      "step": 2367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.92664373285821e-05,
+      "loss": 1.1875,
+      "step": 2368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926560103387243e-05,
+      "loss": 0.9023,
+      "step": 2369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9264764271643275e-05,
+      "loss": 0.7754,
+      "step": 2370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9263927041914954e-05,
+      "loss": 0.958,
+      "step": 2371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926308934470777e-05,
+      "loss": 0.9824,
+      "step": 2372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926225118004206e-05,
+      "loss": 0.9551,
+      "step": 2373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926141254793816e-05,
+      "loss": 1.0996,
+      "step": 2374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926057344841643e-05,
+      "loss": 1.0508,
+      "step": 2375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9259733881497224e-05,
+      "loss": 0.833,
+      "step": 2376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9258893847200916e-05,
+      "loss": 1.1641,
+      "step": 2377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925805334554789e-05,
+      "loss": 0.8721,
+      "step": 2378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9257212376558546e-05,
+      "loss": 1.208,
+      "step": 2379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925637094025329e-05,
+      "loss": 0.9756,
+      "step": 2380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925552903665255e-05,
+      "loss": 1.0146,
+      "step": 2381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925468666577674e-05,
+      "loss": 1.1309,
+      "step": 2382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925384382764631e-05,
+      "loss": 1.1797,
+      "step": 2383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925300052228172e-05,
+      "loss": 1.0879,
+      "step": 2384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925215674970343e-05,
+      "loss": 1.1445,
+      "step": 2385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925131250993191e-05,
+      "loss": 1.0098,
+      "step": 2386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925046780298764e-05,
+      "loss": 0.9775,
+      "step": 2387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.924962262889115e-05,
+      "loss": 1.1104,
+      "step": 2388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9248776987662924e-05,
+      "loss": 1.1816,
+      "step": 2389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.924793087932348e-05,
+      "loss": 0.9766,
+      "step": 2390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.924708430389337e-05,
+      "loss": 1.0078,
+      "step": 2391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.924623726139312e-05,
+      "loss": 1.0361,
+      "step": 2392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.92453897518433e-05,
+      "loss": 1.1914,
+      "step": 2393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.924454177526447e-05,
+      "loss": 0.9355,
+      "step": 2394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9243693331677197e-05,
+      "loss": 1.2559,
+      "step": 2395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9242844421102094e-05,
+      "loss": 0.9102,
+      "step": 2396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.924199504355974e-05,
+      "loss": 1.0264,
+      "step": 2397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9241145199070756e-05,
+      "loss": 1.0498,
+      "step": 2398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.924029488765576e-05,
+      "loss": 1.165,
+      "step": 2399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9239444109335395e-05,
+      "loss": 1.1523,
+      "step": 2400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.92385928641303e-05,
+      "loss": 0.9736,
+      "step": 2401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923774115206114e-05,
+      "loss": 1.0742,
+      "step": 2402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9236888973148575e-05,
+      "loss": 1.0039,
+      "step": 2403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923603632741329e-05,
+      "loss": 0.9824,
+      "step": 2404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9235183214875976e-05,
+      "loss": 1.1406,
+      "step": 2405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923432963555733e-05,
+      "loss": 1.0322,
+      "step": 2406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923347558947807e-05,
+      "loss": 1.083,
+      "step": 2407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9232621076658926e-05,
+      "loss": 1.1123,
+      "step": 2408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923176609712062e-05,
+      "loss": 0.9902,
+      "step": 2409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923091065088392e-05,
+      "loss": 0.8882,
+      "step": 2410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923005473796957e-05,
+      "loss": 1.1084,
+      "step": 2411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9229198358398344e-05,
+      "loss": 0.9658,
+      "step": 2412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.922834151219102e-05,
+      "loss": 1.168,
+      "step": 2413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.922748419936841e-05,
+      "loss": 0.8652,
+      "step": 2414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.922662641995129e-05,
+      "loss": 0.9951,
+      "step": 2415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.92257681739605e-05,
+      "loss": 1.002,
+      "step": 2416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.922490946141686e-05,
+      "loss": 1.0439,
+      "step": 2417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.92240502823412e-05,
+      "loss": 1.2168,
+      "step": 2418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.922319063675437e-05,
+      "loss": 1.0303,
+      "step": 2419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.922233052467724e-05,
+      "loss": 1.0391,
+      "step": 2420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9221469946130685e-05,
+      "loss": 1.0596,
+      "step": 2421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.922060890113558e-05,
+      "loss": 0.9756,
+      "step": 2422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921974738971282e-05,
+      "loss": 1.0029,
+      "step": 2423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921888541188332e-05,
+      "loss": 0.9541,
+      "step": 2424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9218022967667986e-05,
+      "loss": 1.0342,
+      "step": 2425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921716005708775e-05,
+      "loss": 1.0605,
+      "step": 2426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921629668016356e-05,
+      "loss": 1.1816,
+      "step": 2427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921543283691636e-05,
+      "loss": 1.0859,
+      "step": 2428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921456852736712e-05,
+      "loss": 1.1074,
+      "step": 2429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921370375153681e-05,
+      "loss": 1.0996,
+      "step": 2430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921283850944641e-05,
+      "loss": 1.0977,
+      "step": 2431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9211972801116924e-05,
+      "loss": 1.0156,
+      "step": 2432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9211106626569364e-05,
+      "loss": 1.2188,
+      "step": 2433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9210239985824736e-05,
+      "loss": 0.9941,
+      "step": 2434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.920937287890408e-05,
+      "loss": 0.9863,
+      "step": 2435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9208505305828445e-05,
+      "loss": 1.126,
+      "step": 2436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.920763726661887e-05,
+      "loss": 1.0664,
+      "step": 2437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9206768761296426e-05,
+      "loss": 1.0576,
+      "step": 2438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.920589978988219e-05,
+      "loss": 1.0244,
+      "step": 2439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9205030352397246e-05,
+      "loss": 0.9326,
+      "step": 2440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.92041604488627e-05,
+      "loss": 1.1289,
+      "step": 2441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9203290079299655e-05,
+      "loss": 1.0225,
+      "step": 2442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.920241924372923e-05,
+      "loss": 1.0781,
+      "step": 2443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.920154794217257e-05,
+      "loss": 1.168,
+      "step": 2444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9200676174650814e-05,
+      "loss": 1.1113,
+      "step": 2445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91998039411851e-05,
+      "loss": 1.0166,
+      "step": 2446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919893124179663e-05,
+      "loss": 1.1055,
+      "step": 2447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919805807650654e-05,
+      "loss": 1.0977,
+      "step": 2448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919718444533606e-05,
+      "loss": 1.0156,
+      "step": 2449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919631034830635e-05,
+      "loss": 1.0635,
+      "step": 2450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919543578543865e-05,
+      "loss": 1.0391,
+      "step": 2451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919456075675417e-05,
+      "loss": 0.9639,
+      "step": 2452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919368526227416e-05,
+      "loss": 1.0039,
+      "step": 2453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9192809302019856e-05,
+      "loss": 1.0508,
+      "step": 2454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919193287601251e-05,
+      "loss": 1.1445,
+      "step": 2455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9191055984273395e-05,
+      "loss": 0.7979,
+      "step": 2456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9190178626823784e-05,
+      "loss": 1.1914,
+      "step": 2457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.918930080368499e-05,
+      "loss": 0.998,
+      "step": 2458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.918842251487828e-05,
+      "loss": 1.0439,
+      "step": 2459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9187543760424995e-05,
+      "loss": 1.0752,
+      "step": 2460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.918666454034645e-05,
+      "loss": 1.2168,
+      "step": 2461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9185784854663985e-05,
+      "loss": 1.0078,
+      "step": 2462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.918490470339895e-05,
+      "loss": 1.0781,
+      "step": 2463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9184024086572685e-05,
+      "loss": 1.0547,
+      "step": 2464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.918314300420659e-05,
+      "loss": 1.1543,
+      "step": 2465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.918226145632202e-05,
+      "loss": 1.0234,
+      "step": 2466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9181379442940376e-05,
+      "loss": 1.1104,
+      "step": 2467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9180496964083066e-05,
+      "loss": 0.9775,
+      "step": 2468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91796140197715e-05,
+      "loss": 0.8188,
+      "step": 2469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.917873061002711e-05,
+      "loss": 1.1895,
+      "step": 2470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.917784673487133e-05,
+      "loss": 1.1211,
+      "step": 2471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.917696239432561e-05,
+      "loss": 1.0654,
+      "step": 2472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.917607758841141e-05,
+      "loss": 0.8379,
+      "step": 2473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9175192317150205e-05,
+      "loss": 1.1406,
+      "step": 2474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.917430658056347e-05,
+      "loss": 1.1543,
+      "step": 2475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91734203786727e-05,
+      "loss": 1.0469,
+      "step": 2476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.917253371149942e-05,
+      "loss": 0.8672,
+      "step": 2477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.917164657906511e-05,
+      "loss": 1.0127,
+      "step": 2478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.917075898139134e-05,
+      "loss": 0.8584,
+      "step": 2479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9169870918499614e-05,
+      "loss": 0.9795,
+      "step": 2480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91689823904115e-05,
+      "loss": 0.8984,
+      "step": 2481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9168093397148564e-05,
+      "loss": 1.1436,
+      "step": 2482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.916720393873236e-05,
+      "loss": 1.1201,
+      "step": 2483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91663140151845e-05,
+      "loss": 1.0703,
+      "step": 2484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.916542362652655e-05,
+      "loss": 0.9141,
+      "step": 2485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.916453277278015e-05,
+      "loss": 1.1758,
+      "step": 2486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9163641453966886e-05,
+      "loss": 0.957,
+      "step": 2487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.916274967010841e-05,
+      "loss": 1.0654,
+      "step": 2488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.916185742122635e-05,
+      "loss": 1.0811,
+      "step": 2489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.916096470734237e-05,
+      "loss": 1.042,
+      "step": 2490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.916007152847812e-05,
+      "loss": 1.0361,
+      "step": 2491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9159177884655284e-05,
+      "loss": 1.0225,
+      "step": 2492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9158283775895543e-05,
+      "loss": 1.1719,
+      "step": 2493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91573892022206e-05,
+      "loss": 1.1035,
+      "step": 2494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9156494163652167e-05,
+      "loss": 1.0352,
+      "step": 2495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9155598660211946e-05,
+      "loss": 1.0752,
+      "step": 2496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915470269192169e-05,
+      "loss": 0.9492,
+      "step": 2497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9153806258803125e-05,
+      "loss": 0.9355,
+      "step": 2498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915290936087802e-05,
+      "loss": 0.9805,
+      "step": 2499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915201199816812e-05,
+      "loss": 1.1445,
+      "step": 2500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915111417069522e-05,
+      "loss": 0.9971,
+      "step": 2501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9150215878481096e-05,
+      "loss": 1.2363,
+      "step": 2502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914931712154756e-05,
+      "loss": 0.8721,
+      "step": 2503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9148417899916405e-05,
+      "loss": 1.3223,
+      "step": 2504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9147518213609465e-05,
+      "loss": 0.8203,
+      "step": 2505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914661806264857e-05,
+      "loss": 1.1348,
+      "step": 2506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914571744705556e-05,
+      "loss": 1.0791,
+      "step": 2507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914481636685229e-05,
+      "loss": 1.0947,
+      "step": 2508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9143914822060635e-05,
+      "loss": 0.9814,
+      "step": 2509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9143012812702465e-05,
+      "loss": 1.0684,
+      "step": 2510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9142110338799674e-05,
+      "loss": 1.1074,
+      "step": 2511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914120740037416e-05,
+      "loss": 1.0342,
+      "step": 2512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9140303997447835e-05,
+      "loss": 0.9941,
+      "step": 2513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913940013004262e-05,
+      "loss": 1.0801,
+      "step": 2514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913849579818044e-05,
+      "loss": 0.9561,
+      "step": 2515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913759100188327e-05,
+      "loss": 1.0459,
+      "step": 2516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913668574117304e-05,
+      "loss": 0.9717,
+      "step": 2517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913578001607173e-05,
+      "loss": 1.0645,
+      "step": 2518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91348738266013e-05,
+      "loss": 1.1621,
+      "step": 2519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913396717278377e-05,
+      "loss": 0.96,
+      "step": 2520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913306005464112e-05,
+      "loss": 1.1172,
+      "step": 2521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913215247219539e-05,
+      "loss": 1.0176,
+      "step": 2522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913124442546856e-05,
+      "loss": 1.0117,
+      "step": 2523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9130335914482705e-05,
+      "loss": 1.0527,
+      "step": 2524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9129426939259856e-05,
+      "loss": 1.209,
+      "step": 2525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.912851749982207e-05,
+      "loss": 1.0107,
+      "step": 2526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.912760759619142e-05,
+      "loss": 0.999,
+      "step": 2527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9126697228389984e-05,
+      "loss": 1.0312,
+      "step": 2528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.912578639643986e-05,
+      "loss": 1.0469,
+      "step": 2529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9124875100363145e-05,
+      "loss": 1.0498,
+      "step": 2530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9123963340181956e-05,
+      "loss": 1.0,
+      "step": 2531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9123051115918415e-05,
+      "loss": 1.0488,
+      "step": 2532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9122138427594666e-05,
+      "loss": 1.1016,
+      "step": 2533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9121225275232856e-05,
+      "loss": 1.0605,
+      "step": 2534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9120311658855134e-05,
+      "loss": 0.9141,
+      "step": 2535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911939757848369e-05,
+      "loss": 0.9199,
+      "step": 2536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911848303414068e-05,
+      "loss": 1.0635,
+      "step": 2537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911756802584832e-05,
+      "loss": 1.1582,
+      "step": 2538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911665255362882e-05,
+      "loss": 1.0527,
+      "step": 2539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911573661750436e-05,
+      "loss": 1.25,
+      "step": 2540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91148202174972e-05,
+      "loss": 0.9248,
+      "step": 2541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9113903353629574e-05,
+      "loss": 0.9746,
+      "step": 2542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911298602592371e-05,
+      "loss": 1.0537,
+      "step": 2543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91120682344019e-05,
+      "loss": 1.0557,
+      "step": 2544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911114997908639e-05,
+      "loss": 1.0713,
+      "step": 2545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9110231259999476e-05,
+      "loss": 1.0449,
+      "step": 2546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9109312077163456e-05,
+      "loss": 0.9268,
+      "step": 2547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.910839243060063e-05,
+      "loss": 1.0479,
+      "step": 2548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.910747232033331e-05,
+      "loss": 1.0352,
+      "step": 2549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.910655174638383e-05,
+      "loss": 0.9941,
+      "step": 2550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.910563070877453e-05,
+      "loss": 0.9844,
+      "step": 2551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9104709207527764e-05,
+      "loss": 1.0684,
+      "step": 2552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.910378724266589e-05,
+      "loss": 1.1465,
+      "step": 2553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9102864814211284e-05,
+      "loss": 1.0273,
+      "step": 2554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.910194192218632e-05,
+      "loss": 1.0527,
+      "step": 2555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9101018566613406e-05,
+      "loss": 0.9961,
+      "step": 2556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.910009474751495e-05,
+      "loss": 1.1758,
+      "step": 2557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9099170464913364e-05,
+      "loss": 0.9785,
+      "step": 2558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9098245718831076e-05,
+      "loss": 1.0352,
+      "step": 2559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909732050929053e-05,
+      "loss": 0.7139,
+      "step": 2560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909639483631418e-05,
+      "loss": 1.1016,
+      "step": 2561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909546869992449e-05,
+      "loss": 1.1299,
+      "step": 2562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909454210014393e-05,
+      "loss": 0.9668,
+      "step": 2563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909361503699499e-05,
+      "loss": 1.1689,
+      "step": 2564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909268751050016e-05,
+      "loss": 1.0576,
+      "step": 2565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9091759520681966e-05,
+      "loss": 1.1348,
+      "step": 2566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9090831067562905e-05,
+      "loss": 1.0068,
+      "step": 2567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908990215116552e-05,
+      "loss": 0.9248,
+      "step": 2568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908897277151235e-05,
+      "loss": 1.0352,
+      "step": 2569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9088042928625955e-05,
+      "loss": 1.0596,
+      "step": 2570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908711262252889e-05,
+      "loss": 1.1299,
+      "step": 2571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908618185324374e-05,
+      "loss": 0.9502,
+      "step": 2572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9085250620793076e-05,
+      "loss": 1.0723,
+      "step": 2573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908431892519951e-05,
+      "loss": 0.8525,
+      "step": 2574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908338676648566e-05,
+      "loss": 0.835,
+      "step": 2575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908245414467412e-05,
+      "loss": 0.9482,
+      "step": 2576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908152105978754e-05,
+      "loss": 0.9785,
+      "step": 2577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908058751184857e-05,
+      "loss": 0.9033,
+      "step": 2578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.907965350087985e-05,
+      "loss": 1.1836,
+      "step": 2579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.907871902690405e-05,
+      "loss": 0.9551,
+      "step": 2580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.907778408994385e-05,
+      "loss": 1.0127,
+      "step": 2581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9076848690021936e-05,
+      "loss": 1.1816,
+      "step": 2582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.907591282716101e-05,
+      "loss": 1.1025,
+      "step": 2583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.907497650138377e-05,
+      "loss": 1.0293,
+      "step": 2584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9074039712712955e-05,
+      "loss": 0.9263,
+      "step": 2585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9073102461171285e-05,
+      "loss": 0.9707,
+      "step": 2586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.907216474678152e-05,
+      "loss": 1.082,
+      "step": 2587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.90712265695664e-05,
+      "loss": 1.0371,
+      "step": 2588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.90702879295487e-05,
+      "loss": 0.9141,
+      "step": 2589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.906934882675119e-05,
+      "loss": 1.0488,
+      "step": 2590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.906840926119667e-05,
+      "loss": 1.0684,
+      "step": 2591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.906746923290794e-05,
+      "loss": 1.0938,
+      "step": 2592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9066528741907796e-05,
+      "loss": 1.0088,
+      "step": 2593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.906558778821908e-05,
+      "loss": 0.769,
+      "step": 2594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.906464637186461e-05,
+      "loss": 1.209,
+      "step": 2595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.906370449286724e-05,
+      "loss": 1.1611,
+      "step": 2596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.906276215124984e-05,
+      "loss": 0.959,
+      "step": 2597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.906181934703526e-05,
+      "loss": 0.8682,
+      "step": 2598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9060876080246375e-05,
+      "loss": 1.0088,
+      "step": 2599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.905993235090608e-05,
+      "loss": 1.085,
+      "step": 2600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9058988159037294e-05,
+      "loss": 0.9248,
+      "step": 2601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.905804350466291e-05,
+      "loss": 1.041,
+      "step": 2602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.905709838780585e-05,
+      "loss": 0.9111,
+      "step": 2603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9056152808489075e-05,
+      "loss": 0.999,
+      "step": 2604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.90552067667355e-05,
+      "loss": 1.2363,
+      "step": 2605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.90542602625681e-05,
+      "loss": 0.8721,
+      "step": 2606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9053313296009836e-05,
+      "loss": 1.1289,
+      "step": 2607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.90523658670837e-05,
+      "loss": 0.8516,
+      "step": 2608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.905141797581267e-05,
+      "loss": 1.0615,
+      "step": 2609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.905046962221975e-05,
+      "loss": 0.9775,
+      "step": 2610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.904952080632796e-05,
+      "loss": 0.9102,
+      "step": 2611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.904857152816033e-05,
+      "loss": 1.0254,
+      "step": 2612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9047621787739875e-05,
+      "loss": 1.1357,
+      "step": 2613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.904667158508967e-05,
+      "loss": 0.9971,
+      "step": 2614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9045720920232755e-05,
+      "loss": 1.1504,
+      "step": 2615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9044769793192204e-05,
+      "loss": 1.0029,
+      "step": 2616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.90438182039911e-05,
+      "loss": 1.0625,
+      "step": 2617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9042866152652534e-05,
+      "loss": 0.9307,
+      "step": 2618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.904191363919961e-05,
+      "loss": 0.9912,
+      "step": 2619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9040960663655435e-05,
+      "loss": 1.0098,
+      "step": 2620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9040007226043154e-05,
+      "loss": 0.9736,
+      "step": 2621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.903905332638588e-05,
+      "loss": 1.0732,
+      "step": 2622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9038098964706785e-05,
+      "loss": 1.0967,
+      "step": 2623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9037144141029014e-05,
+      "loss": 0.9541,
+      "step": 2624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.903618885537573e-05,
+      "loss": 1.0996,
+      "step": 2625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.903523310777014e-05,
+      "loss": 1.0693,
+      "step": 2626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.903427689823541e-05,
+      "loss": 1.0234,
+      "step": 2627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.903332022679477e-05,
+      "loss": 0.9404,
+      "step": 2628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.903236309347142e-05,
+      "loss": 1.0381,
+      "step": 2629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9031405498288586e-05,
+      "loss": 1.1582,
+      "step": 2630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.903044744126952e-05,
+      "loss": 1.0059,
+      "step": 2631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.902948892243745e-05,
+      "loss": 0.9385,
+      "step": 2632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.902852994181565e-05,
+      "loss": 1.1172,
+      "step": 2633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9027570499427394e-05,
+      "loss": 1.1016,
+      "step": 2634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.902661059529595e-05,
+      "loss": 1.0723,
+      "step": 2635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.902565022944463e-05,
+      "loss": 1.0898,
+      "step": 2636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9024689401896726e-05,
+      "loss": 0.8447,
+      "step": 2637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.902372811267557e-05,
+      "loss": 1.1152,
+      "step": 2638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9022766361804474e-05,
+      "loss": 1.0596,
+      "step": 2639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.902180414930678e-05,
+      "loss": 1.1953,
+      "step": 2640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.902084147520584e-05,
+      "loss": 0.9072,
+      "step": 2641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.901987833952502e-05,
+      "loss": 0.8721,
+      "step": 2642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9018914742287686e-05,
+      "loss": 1.1035,
+      "step": 2643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.901795068351722e-05,
+      "loss": 1.0195,
+      "step": 2644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.901698616323703e-05,
+      "loss": 1.2344,
+      "step": 2645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9016021181470505e-05,
+      "loss": 1.1104,
+      "step": 2646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.901505573824107e-05,
+      "loss": 1.0537,
+      "step": 2647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.901408983357216e-05,
+      "loss": 1.1328,
+      "step": 2648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.90131234674872e-05,
+      "loss": 0.9795,
+      "step": 2649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.901215664000965e-05,
+      "loss": 1.248,
+      "step": 2650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.901118935116298e-05,
+      "loss": 1.0625,
+      "step": 2651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.901022160097065e-05,
+      "loss": 0.9326,
+      "step": 2652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.900925338945614e-05,
+      "loss": 1.0908,
+      "step": 2653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.900828471664296e-05,
+      "loss": 1.0449,
+      "step": 2654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.900731558255462e-05,
+      "loss": 1.127,
+      "step": 2655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.900634598721462e-05,
+      "loss": 0.9834,
+      "step": 2656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.90053759306465e-05,
+      "loss": 1.0127,
+      "step": 2657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.90044054128738e-05,
+      "loss": 0.9248,
+      "step": 2658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.900343443392007e-05,
+      "loss": 1.1064,
+      "step": 2659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.900246299380887e-05,
+      "loss": 1.0479,
+      "step": 2660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9001491092563785e-05,
+      "loss": 1.0752,
+      "step": 2661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9000518730208384e-05,
+      "loss": 1.0527,
+      "step": 2662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.899954590676628e-05,
+      "loss": 0.916,
+      "step": 2663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8998572622261065e-05,
+      "loss": 0.9062,
+      "step": 2664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.899759887671637e-05,
+      "loss": 0.9419,
+      "step": 2665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.899662467015582e-05,
+      "loss": 1.0928,
+      "step": 2666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.899565000260306e-05,
+      "loss": 1.2754,
+      "step": 2667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8994674874081735e-05,
+      "loss": 1.1396,
+      "step": 2668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.899369928461551e-05,
+      "loss": 0.9141,
+      "step": 2669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.899272323422806e-05,
+      "loss": 1.0303,
+      "step": 2670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8991746722943075e-05,
+      "loss": 1.0771,
+      "step": 2671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.899076975078425e-05,
+      "loss": 0.874,
+      "step": 2672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8989792317775295e-05,
+      "loss": 1.0586,
+      "step": 2673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.898881442393992e-05,
+      "loss": 0.9492,
+      "step": 2674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.898783606930187e-05,
+      "loss": 0.9814,
+      "step": 2675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8986857253884885e-05,
+      "loss": 1.1797,
+      "step": 2676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.898587797771271e-05,
+      "loss": 0.8789,
+      "step": 2677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.89848982408091e-05,
+      "loss": 1.0332,
+      "step": 2678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8983918043197854e-05,
+      "loss": 0.9941,
+      "step": 2679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8982937384902746e-05,
+      "loss": 1.1133,
+      "step": 2680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.898195626594757e-05,
+      "loss": 1.0234,
+      "step": 2681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8980974686356144e-05,
+      "loss": 1.0566,
+      "step": 2682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.897999264615228e-05,
+      "loss": 1.0586,
+      "step": 2683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.897901014535981e-05,
+      "loss": 1.1484,
+      "step": 2684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8978027184002586e-05,
+      "loss": 0.9551,
+      "step": 2685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.897704376210445e-05,
+      "loss": 1.0801,
+      "step": 2686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.897605987968927e-05,
+      "loss": 1.0449,
+      "step": 2687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.897507553678093e-05,
+      "loss": 0.9707,
+      "step": 2688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8974090733403304e-05,
+      "loss": 1.0586,
+      "step": 2689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.89731054695803e-05,
+      "loss": 1.0059,
+      "step": 2690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.897211974533582e-05,
+      "loss": 1.1211,
+      "step": 2691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.89711335606938e-05,
+      "loss": 1.0742,
+      "step": 2692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8970146915678145e-05,
+      "loss": 1.0576,
+      "step": 2693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.896915981031282e-05,
+      "loss": 0.8984,
+      "step": 2694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8968172244621774e-05,
+      "loss": 1.0029,
+      "step": 2695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.896718421862897e-05,
+      "loss": 1.1201,
+      "step": 2696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.896619573235839e-05,
+      "loss": 0.9141,
+      "step": 2697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.896520678583401e-05,
+      "loss": 1.0879,
+      "step": 2698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.896421737907984e-05,
+      "loss": 0.9023,
+      "step": 2699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.896322751211989e-05,
+      "loss": 1.1777,
+      "step": 2700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.896223718497817e-05,
+      "loss": 0.9736,
+      "step": 2701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8961246397678724e-05,
+      "loss": 0.9717,
+      "step": 2702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.896025515024559e-05,
+      "loss": 1.1377,
+      "step": 2703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8959263442702824e-05,
+      "loss": 1.0107,
+      "step": 2704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.895827127507448e-05,
+      "loss": 1.0879,
+      "step": 2705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.895727864738466e-05,
+      "loss": 0.9902,
+      "step": 2706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8956285559657434e-05,
+      "loss": 1.0498,
+      "step": 2707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.895529201191691e-05,
+      "loss": 1.0703,
+      "step": 2708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.895429800418719e-05,
+      "loss": 0.8872,
+      "step": 2709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.895330353649241e-05,
+      "loss": 1.1152,
+      "step": 2710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.895230860885668e-05,
+      "loss": 1.0039,
+      "step": 2711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.895131322130416e-05,
+      "loss": 0.9277,
+      "step": 2712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8950317373858996e-05,
+      "loss": 1.1133,
+      "step": 2713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.894932106654537e-05,
+      "loss": 0.9902,
+      "step": 2714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.894832429938745e-05,
+      "loss": 0.9443,
+      "step": 2715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.894732707240942e-05,
+      "loss": 0.999,
+      "step": 2716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.894632938563548e-05,
+      "loss": 1.1504,
+      "step": 2717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8945331239089855e-05,
+      "loss": 1.0166,
+      "step": 2718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.894433263279675e-05,
+      "loss": 1.0898,
+      "step": 2719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.894333356678041e-05,
+      "loss": 0.9102,
+      "step": 2720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8942334041065066e-05,
+      "loss": 1.0654,
+      "step": 2721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8941334055674994e-05,
+      "loss": 1.1074,
+      "step": 2722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8940333610634435e-05,
+      "loss": 1.0645,
+      "step": 2723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.893933270596769e-05,
+      "loss": 1.0957,
+      "step": 2724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.893833134169904e-05,
+      "loss": 1.0127,
+      "step": 2725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8937329517852786e-05,
+      "loss": 0.8887,
+      "step": 2726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.893632723445323e-05,
+      "loss": 0.9717,
+      "step": 2727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.893532449152471e-05,
+      "loss": 1.1816,
+      "step": 2728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8934321289091533e-05,
+      "loss": 1.1758,
+      "step": 2729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.893331762717808e-05,
+      "loss": 0.8086,
+      "step": 2730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.893231350580869e-05,
+      "loss": 1.2031,
+      "step": 2731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8931308925007724e-05,
+      "loss": 1.0654,
+      "step": 2732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.893030388479956e-05,
+      "loss": 1.1543,
+      "step": 2733
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.89292983852086e-05,
+      "loss": 1.0293,
+      "step": 2734
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.892829242625923e-05,
+      "loss": 1.0332,
+      "step": 2735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8927286007975877e-05,
+      "loss": 1.002,
+      "step": 2736
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.892627913038295e-05,
+      "loss": 1.04,
+      "step": 2737
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8925271793504895e-05,
+      "loss": 0.7988,
+      "step": 2738
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8924263997366145e-05,
+      "loss": 1.0576,
+      "step": 2739
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.892325574199117e-05,
+      "loss": 1.0059,
+      "step": 2740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8922247027404416e-05,
+      "loss": 0.9551,
+      "step": 2741
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.892123785363039e-05,
+      "loss": 1.0264,
+      "step": 2742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.892022822069355e-05,
+      "loss": 1.0625,
+      "step": 2743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8919218128618424e-05,
+      "loss": 0.9678,
+      "step": 2744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.891820757742951e-05,
+      "loss": 1.1074,
+      "step": 2745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.891719656715134e-05,
+      "loss": 0.9814,
+      "step": 2746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.891618509780844e-05,
+      "loss": 0.8848,
+      "step": 2747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8915173169425355e-05,
+      "loss": 1.0391,
+      "step": 2748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.891416078202665e-05,
+      "loss": 1.0332,
+      "step": 2749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.891314793563688e-05,
+      "loss": 1.1631,
+      "step": 2750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.891213463028064e-05,
+      "loss": 1.0107,
+      "step": 2751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8911120865982506e-05,
+      "loss": 1.0537,
+      "step": 2752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8910106642767085e-05,
+      "loss": 0.9707,
+      "step": 2753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.890909196065899e-05,
+      "loss": 0.9678,
+      "step": 2754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.890807681968284e-05,
+      "loss": 0.9658,
+      "step": 2755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.890706121986327e-05,
+      "loss": 0.8428,
+      "step": 2756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8906045161224935e-05,
+      "loss": 1.1738,
+      "step": 2757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8905028643792485e-05,
+      "loss": 1.1201,
+      "step": 2758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.890401166759058e-05,
+      "loss": 1.0469,
+      "step": 2759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.890299423264391e-05,
+      "loss": 1.0195,
+      "step": 2760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.890197633897715e-05,
+      "loss": 0.8457,
+      "step": 2761
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8900957986615034e-05,
+      "loss": 0.9834,
+      "step": 2762
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.889993917558224e-05,
+      "loss": 1.0547,
+      "step": 2763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.889891990590351e-05,
+      "loss": 0.9844,
+      "step": 2764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.889790017760358e-05,
+      "loss": 0.9932,
+      "step": 2765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.889687999070718e-05,
+      "loss": 1.0098,
+      "step": 2766
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.889585934523908e-05,
+      "loss": 1.1045,
+      "step": 2767
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8894838241224036e-05,
+      "loss": 1.084,
+      "step": 2768
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8893816678686844e-05,
+      "loss": 1.2441,
+      "step": 2769
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.889279465765229e-05,
+      "loss": 1.1035,
+      "step": 2770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.889177217814516e-05,
+      "loss": 1.0127,
+      "step": 2771
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8890749240190284e-05,
+      "loss": 0.9668,
+      "step": 2772
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.888972584381248e-05,
+      "loss": 1.0127,
+      "step": 2773
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.888870198903658e-05,
+      "loss": 1.0537,
+      "step": 2774
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8887677675887434e-05,
+      "loss": 1.0566,
+      "step": 2775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8886652904389895e-05,
+      "loss": 0.958,
+      "step": 2776
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.888562767456883e-05,
+      "loss": 1.04,
+      "step": 2777
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8884601986449136e-05,
+      "loss": 0.9766,
+      "step": 2778
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8883575840055686e-05,
+      "loss": 1.1875,
+      "step": 2779
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.888254923541338e-05,
+      "loss": 0.7295,
+      "step": 2780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.888152217254713e-05,
+      "loss": 0.6982,
+      "step": 2781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.888049465148188e-05,
+      "loss": 1.1016,
+      "step": 2782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8879466672242535e-05,
+      "loss": 1.1133,
+      "step": 2783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.887843823485407e-05,
+      "loss": 0.8457,
+      "step": 2784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.887740933934142e-05,
+      "loss": 1.1328,
+      "step": 2785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.887637998572956e-05,
+      "loss": 1.0078,
+      "step": 2786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.887535017404347e-05,
+      "loss": 1.3105,
+      "step": 2787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8874319904308157e-05,
+      "loss": 1.0986,
+      "step": 2788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.887328917654859e-05,
+      "loss": 0.9189,
+      "step": 2789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8872257990789806e-05,
+      "loss": 1.1084,
+      "step": 2790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.887122634705682e-05,
+      "loss": 1.0684,
+      "step": 2791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8870194245374673e-05,
+      "loss": 1.1201,
+      "step": 2792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.88691616857684e-05,
+      "loss": 1.1553,
+      "step": 2793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.886812866826308e-05,
+      "loss": 0.9395,
+      "step": 2794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.886709519288375e-05,
+      "loss": 1.1113,
+      "step": 2795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8866061259655514e-05,
+      "loss": 0.9893,
+      "step": 2796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8865026868603454e-05,
+      "loss": 0.9961,
+      "step": 2797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.886399201975266e-05,
+      "loss": 1.0039,
+      "step": 2798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.886295671312828e-05,
+      "loss": 1.0049,
+      "step": 2799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.886192094875539e-05,
+      "loss": 1.0352,
+      "step": 2800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8860884726659166e-05,
+      "loss": 0.9951,
+      "step": 2801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.885984804686473e-05,
+      "loss": 1.1592,
+      "step": 2802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.885881090939725e-05,
+      "loss": 1.1787,
+      "step": 2803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.885777331428189e-05,
+      "loss": 1.1074,
+      "step": 2804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.885673526154383e-05,
+      "loss": 0.9922,
+      "step": 2805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.885569675120826e-05,
+      "loss": 0.9766,
+      "step": 2806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.885465778330039e-05,
+      "loss": 1.1016,
+      "step": 2807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.885361835784541e-05,
+      "loss": 1.0625,
+      "step": 2808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.885257847486857e-05,
+      "loss": 1.0166,
+      "step": 2809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8851538134395095e-05,
+      "loss": 1.0352,
+      "step": 2810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.885049733645022e-05,
+      "loss": 1.0342,
+      "step": 2811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8849456081059224e-05,
+      "loss": 0.9766,
+      "step": 2812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8848414368247354e-05,
+      "loss": 1.0234,
+      "step": 2813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.88473721980399e-05,
+      "loss": 1.1816,
+      "step": 2814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8846329570462155e-05,
+      "loss": 0.9307,
+      "step": 2815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.884528648553942e-05,
+      "loss": 0.9072,
+      "step": 2816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8844242943297e-05,
+      "loss": 1.0234,
+      "step": 2817
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8843198943760216e-05,
+      "loss": 1.1318,
+      "step": 2818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8842154486954416e-05,
+      "loss": 0.9863,
+      "step": 2819
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.884110957290494e-05,
+      "loss": 0.9419,
+      "step": 2820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.884006420163714e-05,
+      "loss": 1.0205,
+      "step": 2821
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.883901837317639e-05,
+      "loss": 0.9961,
+      "step": 2822
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.883797208754807e-05,
+      "loss": 1.2422,
+      "step": 2823
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.883692534477757e-05,
+      "loss": 0.7734,
+      "step": 2824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.883587814489028e-05,
+      "loss": 1.0996,
+      "step": 2825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.883483048791162e-05,
+      "loss": 0.7852,
+      "step": 2826
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.883378237386703e-05,
+      "loss": 1.0254,
+      "step": 2827
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.883273380278192e-05,
+      "loss": 1.1152,
+      "step": 2828
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.883168477468175e-05,
+      "loss": 1.0244,
+      "step": 2829
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.883063528959197e-05,
+      "loss": 1.0303,
+      "step": 2830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8829585347538056e-05,
+      "loss": 1.0439,
+      "step": 2831
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.882853494854547e-05,
+      "loss": 0.9609,
+      "step": 2832
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8827484092639726e-05,
+      "loss": 0.957,
+      "step": 2833
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.88264327798463e-05,
+      "loss": 1.0312,
+      "step": 2834
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8825381010190725e-05,
+      "loss": 1.0596,
+      "step": 2835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.882432878369852e-05,
+      "loss": 0.7661,
+      "step": 2836
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.882327610039521e-05,
+      "loss": 1.042,
+      "step": 2837
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.882222296030635e-05,
+      "loss": 1.1738,
+      "step": 2838
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.882116936345749e-05,
+      "loss": 1.1113,
+      "step": 2839
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.88201153098742e-05,
+      "loss": 1.0098,
+      "step": 2840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.881906079958205e-05,
+      "loss": 1.0068,
+      "step": 2841
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.881800583260666e-05,
+      "loss": 1.0205,
+      "step": 2842
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.88169504089736e-05,
+      "loss": 0.9092,
+      "step": 2843
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8815894528708485e-05,
+      "loss": 1.0479,
+      "step": 2844
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8814838191836954e-05,
+      "loss": 0.8779,
+      "step": 2845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.881378139838463e-05,
+      "loss": 1.0977,
+      "step": 2846
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.881272414837717e-05,
+      "loss": 1.0068,
+      "step": 2847
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.881166644184021e-05,
+      "loss": 1.1367,
+      "step": 2848
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.881060827879943e-05,
+      "loss": 1.0967,
+      "step": 2849
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.880954965928051e-05,
+      "loss": 1.084,
+      "step": 2850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8808490583309136e-05,
+      "loss": 0.9795,
+      "step": 2851
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8807431050911005e-05,
+      "loss": 1.0576,
+      "step": 2852
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.880637106211183e-05,
+      "loss": 1.1934,
+      "step": 2853
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8805310616937355e-05,
+      "loss": 0.9434,
+      "step": 2854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.880424971541328e-05,
+      "loss": 0.9189,
+      "step": 2855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.880318835756537e-05,
+      "loss": 1.0244,
+      "step": 2856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.880212654341938e-05,
+      "loss": 0.999,
+      "step": 2857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.880106427300106e-05,
+      "loss": 1.043,
+      "step": 2858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8800001546336214e-05,
+      "loss": 0.9873,
+      "step": 2859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.879893836345062e-05,
+      "loss": 1.0264,
+      "step": 2860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.879787472437007e-05,
+      "loss": 0.9004,
+      "step": 2861
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8796810629120394e-05,
+      "loss": 0.9912,
+      "step": 2862
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.87957460777274e-05,
+      "loss": 0.9561,
+      "step": 2863
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.879468107021692e-05,
+      "loss": 0.9609,
+      "step": 2864
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.879361560661481e-05,
+      "loss": 0.7583,
+      "step": 2865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.879254968694691e-05,
+      "loss": 1.0918,
+      "step": 2866
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.879148331123911e-05,
+      "loss": 1.0156,
+      "step": 2867
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.879041647951726e-05,
+      "loss": 1.0645,
+      "step": 2868
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.878934919180727e-05,
+      "loss": 1.0498,
+      "step": 2869
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.878828144813503e-05,
+      "loss": 1.0488,
+      "step": 2870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8787213248526456e-05,
+      "loss": 1.1367,
+      "step": 2871
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8786144593007466e-05,
+      "loss": 1.0049,
+      "step": 2872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8785075481603995e-05,
+      "loss": 0.9404,
+      "step": 2873
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.878400591434199e-05,
+      "loss": 0.9307,
+      "step": 2874
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8782935891247396e-05,
+      "loss": 1.1562,
+      "step": 2875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.878186541234619e-05,
+      "loss": 0.8931,
+      "step": 2876
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8780794477664346e-05,
+      "loss": 0.9238,
+      "step": 2877
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.877972308722785e-05,
+      "loss": 0.9238,
+      "step": 2878
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.877865124106271e-05,
+      "loss": 0.9346,
+      "step": 2879
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.877757893919493e-05,
+      "loss": 1.0225,
+      "step": 2880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8776506181650525e-05,
+      "loss": 0.834,
+      "step": 2881
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.877543296845554e-05,
+      "loss": 1.1797,
+      "step": 2882
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8774359299636015e-05,
+      "loss": 0.9824,
+      "step": 2883
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8773285175218004e-05,
+      "loss": 0.8691,
+      "step": 2884
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8772210595227564e-05,
+      "loss": 1.0059,
+      "step": 2885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8771135559690784e-05,
+      "loss": 0.9336,
+      "step": 2886
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8770060068633745e-05,
+      "loss": 1.1045,
+      "step": 2887
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876898412208256e-05,
+      "loss": 0.9297,
+      "step": 2888
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8767907720063316e-05,
+      "loss": 1.0312,
+      "step": 2889
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876683086260215e-05,
+      "loss": 1.0732,
+      "step": 2890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876575354972519e-05,
+      "loss": 0.9414,
+      "step": 2891
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876467578145858e-05,
+      "loss": 0.9746,
+      "step": 2892
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876359755782848e-05,
+      "loss": 1.0449,
+      "step": 2893
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876251887886103e-05,
+      "loss": 1.0957,
+      "step": 2894
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876143974458245e-05,
+      "loss": 1.0957,
+      "step": 2895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876036015501889e-05,
+      "loss": 1.1426,
+      "step": 2896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.875928011019656e-05,
+      "loss": 0.9863,
+      "step": 2897
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.875819961014167e-05,
+      "loss": 0.9111,
+      "step": 2898
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8757118654880445e-05,
+      "loss": 0.9316,
+      "step": 2899
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.875603724443911e-05,
+      "loss": 0.9727,
+      "step": 2900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8754955378843916e-05,
+      "loss": 1.0859,
+      "step": 2901
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.875387305812111e-05,
+      "loss": 1.0039,
+      "step": 2902
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.875279028229695e-05,
+      "loss": 1.1855,
+      "step": 2903
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.875170705139773e-05,
+      "loss": 1.0039,
+      "step": 2904
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.875062336544973e-05,
+      "loss": 1.0459,
+      "step": 2905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8749539224479234e-05,
+      "loss": 0.9609,
+      "step": 2906
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.874845462851256e-05,
+      "loss": 1.085,
+      "step": 2907
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8747369577576035e-05,
+      "loss": 1.1699,
+      "step": 2908
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.874628407169598e-05,
+      "loss": 0.9814,
+      "step": 2909
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.874519811089875e-05,
+      "loss": 1.0566,
+      "step": 2910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.874411169521068e-05,
+      "loss": 1.0879,
+      "step": 2911
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8743024824658156e-05,
+      "loss": 1.0371,
+      "step": 2912
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.874193749926753e-05,
+      "loss": 1.0576,
+      "step": 2913
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8740849719065196e-05,
+      "loss": 1.1123,
+      "step": 2914
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8739761484077565e-05,
+      "loss": 0.8252,
+      "step": 2915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8738672794331024e-05,
+      "loss": 1.0068,
+      "step": 2916
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.873758364985201e-05,
+      "loss": 1.0635,
+      "step": 2917
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8736494050666944e-05,
+      "loss": 1.0117,
+      "step": 2918
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8735403996802274e-05,
+      "loss": 1.0146,
+      "step": 2919
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8734313488284436e-05,
+      "loss": 1.0391,
+      "step": 2920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.873322252513992e-05,
+      "loss": 1.2227,
+      "step": 2921
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.873213110739517e-05,
+      "loss": 1.1504,
+      "step": 2922
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.87310392350767e-05,
+      "loss": 1.0684,
+      "step": 2923
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.872994690821099e-05,
+      "loss": 0.8809,
+      "step": 2924
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.872885412682455e-05,
+      "loss": 1.0479,
+      "step": 2925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.87277608909439e-05,
+      "loss": 1.0186,
+      "step": 2926
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.872666720059557e-05,
+      "loss": 0.7959,
+      "step": 2927
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.87255730558061e-05,
+      "loss": 1.1562,
+      "step": 2928
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8724478456602046e-05,
+      "loss": 1.0615,
+      "step": 2929
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.872338340300996e-05,
+      "loss": 0.9619,
+      "step": 2930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.872228789505642e-05,
+      "loss": 1.1123,
+      "step": 2931
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.872119193276802e-05,
+      "loss": 0.9531,
+      "step": 2932
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.872009551617135e-05,
+      "loss": 0.9092,
+      "step": 2933
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.871899864529301e-05,
+      "loss": 1.0859,
+      "step": 2934
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.871790132015963e-05,
+      "loss": 1.0039,
+      "step": 2935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8716803540797824e-05,
+      "loss": 1.0508,
+      "step": 2936
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8715705307234245e-05,
+      "loss": 1.0889,
+      "step": 2937
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8714606619495544e-05,
+      "loss": 0.9092,
+      "step": 2938
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.871350747760837e-05,
+      "loss": 1.1982,
+      "step": 2939
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8712407881599404e-05,
+      "loss": 1.0156,
+      "step": 2940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8711307831495336e-05,
+      "loss": 1.1484,
+      "step": 2941
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.871020732732285e-05,
+      "loss": 0.834,
+      "step": 2942
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.870910636910866e-05,
+      "loss": 1.0498,
+      "step": 2943
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.870800495687948e-05,
+      "loss": 0.9355,
+      "step": 2944
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.870690309066204e-05,
+      "loss": 0.9668,
+      "step": 2945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.870580077048307e-05,
+      "loss": 1.0391,
+      "step": 2946
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.870469799636933e-05,
+      "loss": 1.0918,
+      "step": 2947
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8703594768347574e-05,
+      "loss": 1.0215,
+      "step": 2948
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8702491086444584e-05,
+      "loss": 0.9141,
+      "step": 2949
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.870138695068714e-05,
+      "loss": 0.9678,
+      "step": 2950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.870028236110203e-05,
+      "loss": 0.8457,
+      "step": 2951
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.869917731771606e-05,
+      "loss": 1.0,
+      "step": 2952
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8698071820556055e-05,
+      "loss": 1.1816,
+      "step": 2953
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8696965869648825e-05,
+      "loss": 1.002,
+      "step": 2954
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.869585946502123e-05,
+      "loss": 1.001,
+      "step": 2955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.869475260670011e-05,
+      "loss": 1.0381,
+      "step": 2956
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8693645294712315e-05,
+      "loss": 1.0781,
+      "step": 2957
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.869253752908472e-05,
+      "loss": 0.9834,
+      "step": 2958
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8691429309844225e-05,
+      "loss": 1.1562,
+      "step": 2959
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.869032063701771e-05,
+      "loss": 0.8867,
+      "step": 2960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.868921151063207e-05,
+      "loss": 1.0879,
+      "step": 2961
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8688101930714235e-05,
+      "loss": 1.0303,
+      "step": 2962
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8686991897291125e-05,
+      "loss": 1.0527,
+      "step": 2963
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8685881410389675e-05,
+      "loss": 1.1699,
+      "step": 2964
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.868477047003684e-05,
+      "loss": 1.1016,
+      "step": 2965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8683659076259574e-05,
+      "loss": 1.1182,
+      "step": 2966
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8682547229084854e-05,
+      "loss": 0.8232,
+      "step": 2967
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8681434928539646e-05,
+      "loss": 0.9028,
+      "step": 2968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.868032217465097e-05,
+      "loss": 0.9668,
+      "step": 2969
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.867920896744579e-05,
+      "loss": 1.2031,
+      "step": 2970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.867809530695116e-05,
+      "loss": 0.957,
+      "step": 2971
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.867698119319408e-05,
+      "loss": 0.9414,
+      "step": 2972
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8675866626201585e-05,
+      "loss": 1.2969,
+      "step": 2973
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.867475160600074e-05,
+      "loss": 1.0078,
+      "step": 2974
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8673636132618594e-05,
+      "loss": 1.0469,
+      "step": 2975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8672520206082216e-05,
+      "loss": 0.9922,
+      "step": 2976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8671403826418685e-05,
+      "loss": 1.0186,
+      "step": 2977
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.867028699365509e-05,
+      "loss": 1.1201,
+      "step": 2978
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.866916970781854e-05,
+      "loss": 1.0,
+      "step": 2979
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.866805196893615e-05,
+      "loss": 1.1309,
+      "step": 2980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8666933777035037e-05,
+      "loss": 1.0049,
+      "step": 2981
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8665815132142337e-05,
+      "loss": 0.998,
+      "step": 2982
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8664696034285195e-05,
+      "loss": 1.04,
+      "step": 2983
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.866357648349078e-05,
+      "loss": 1.0166,
+      "step": 2984
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.866245647978624e-05,
+      "loss": 1.1689,
+      "step": 2985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8661336023198776e-05,
+      "loss": 1.0068,
+      "step": 2986
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.866021511375556e-05,
+      "loss": 1.1025,
+      "step": 2987
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.865909375148381e-05,
+      "loss": 0.957,
+      "step": 2988
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.865797193641072e-05,
+      "loss": 1.0742,
+      "step": 2989
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8656849668563525e-05,
+      "loss": 1.0547,
+      "step": 2990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.865572694796946e-05,
+      "loss": 0.812,
+      "step": 2991
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.865460377465576e-05,
+      "loss": 1.0469,
+      "step": 2992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.865348014864969e-05,
+      "loss": 0.9854,
+      "step": 2993
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.865235606997852e-05,
+      "loss": 0.9863,
+      "step": 2994
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.865123153866951e-05,
+      "loss": 1.0215,
+      "step": 2995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.865010655474997e-05,
+      "loss": 1.1221,
+      "step": 2996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.86489811182472e-05,
+      "loss": 1.0088,
+      "step": 2997
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.864785522918849e-05,
+      "loss": 1.0225,
+      "step": 2998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.864672888760118e-05,
+      "loss": 0.9736,
+      "step": 2999
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.86456020935126e-05,
+      "loss": 1.1016,
+      "step": 3000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.864447484695009e-05,
+      "loss": 1.0938,
+      "step": 3001
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8643347147941004e-05,
+      "loss": 0.835,
+      "step": 3002
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.864221899651272e-05,
+      "loss": 1.1191,
+      "step": 3003
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.864109039269259e-05,
+      "loss": 1.0713,
+      "step": 3004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.863996133650803e-05,
+      "loss": 1.1592,
+      "step": 3005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.863883182798643e-05,
+      "loss": 1.0566,
+      "step": 3006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8637701867155184e-05,
+      "loss": 1.0117,
+      "step": 3007
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8636571454041735e-05,
+      "loss": 1.0645,
+      "step": 3008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8635440588673494e-05,
+      "loss": 0.7896,
+      "step": 3009
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.863430927107793e-05,
+      "loss": 0.8945,
+      "step": 3010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8633177501282465e-05,
+      "loss": 1.041,
+      "step": 3011
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.863204527931459e-05,
+      "loss": 0.9551,
+      "step": 3012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.863091260520177e-05,
+      "loss": 1.04,
+      "step": 3013
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8629779478971494e-05,
+      "loss": 1.0225,
+      "step": 3014
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.862864590065126e-05,
+      "loss": 0.9258,
+      "step": 3015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.862751187026857e-05,
+      "loss": 0.875,
+      "step": 3016
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.862637738785095e-05,
+      "loss": 1.1006,
+      "step": 3017
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8625242453425924e-05,
+      "loss": 0.9189,
+      "step": 3018
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.862410706702105e-05,
+      "loss": 1.1084,
+      "step": 3019
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8622971228663856e-05,
+      "loss": 0.8818,
+      "step": 3020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.862183493838193e-05,
+      "loss": 1.0703,
+      "step": 3021
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8620698196202825e-05,
+      "loss": 1.0117,
+      "step": 3022
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.861956100215414e-05,
+      "loss": 0.999,
+      "step": 3023
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8618423356263466e-05,
+      "loss": 1.127,
+      "step": 3024
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.861728525855842e-05,
+      "loss": 1.1396,
+      "step": 3025
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8616146709066605e-05,
+      "loss": 1.1006,
+      "step": 3026
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8615007707815664e-05,
+      "loss": 1.0068,
+      "step": 3027
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.861386825483322e-05,
+      "loss": 0.9307,
+      "step": 3028
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.861272835014695e-05,
+      "loss": 1.0635,
+      "step": 3029
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.861158799378449e-05,
+      "loss": 0.999,
+      "step": 3030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.861044718577352e-05,
+      "loss": 1.0713,
+      "step": 3031
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8609305926141735e-05,
+      "loss": 0.8271,
+      "step": 3032
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8608164214916825e-05,
+      "loss": 1.0986,
+      "step": 3033
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.860702205212649e-05,
+      "loss": 1.0371,
+      "step": 3034
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.860587943779846e-05,
+      "loss": 1.0205,
+      "step": 3035
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8604736371960436e-05,
+      "loss": 0.9443,
+      "step": 3036
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8603592854640186e-05,
+      "loss": 1.1025,
+      "step": 3037
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.860244888586545e-05,
+      "loss": 1.0859,
+      "step": 3038
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.860130446566398e-05,
+      "loss": 1.0176,
+      "step": 3039
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.860015959406356e-05,
+      "loss": 1.0488,
+      "step": 3040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.859901427109197e-05,
+      "loss": 0.9336,
+      "step": 3041
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8597868496777e-05,
+      "loss": 1.0986,
+      "step": 3042
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.859672227114646e-05,
+      "loss": 0.9668,
+      "step": 3043
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8595575594228156e-05,
+      "loss": 1.1367,
+      "step": 3044
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.859442846604992e-05,
+      "loss": 0.8584,
+      "step": 3045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8593280886639585e-05,
+      "loss": 0.9199,
+      "step": 3046
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8592132856025015e-05,
+      "loss": 1.1953,
+      "step": 3047
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.859098437423405e-05,
+      "loss": 0.9883,
+      "step": 3048
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8589835441294574e-05,
+      "loss": 1.0469,
+      "step": 3049
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.858868605723446e-05,
+      "loss": 1.0869,
+      "step": 3050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.85875362220816e-05,
+      "loss": 1.0605,
+      "step": 3051
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.85863859358639e-05,
+      "loss": 0.9854,
+      "step": 3052
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.858523519860928e-05,
+      "loss": 0.9307,
+      "step": 3053
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8584084010345645e-05,
+      "loss": 0.9668,
+      "step": 3054
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8582932371100964e-05,
+      "loss": 0.8984,
+      "step": 3055
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.858178028090315e-05,
+      "loss": 0.9766,
+      "step": 3056
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.858062773978018e-05,
+      "loss": 1.0859,
+      "step": 3057
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.857947474776002e-05,
+      "loss": 1.0098,
+      "step": 3058
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.857832130487065e-05,
+      "loss": 0.8594,
+      "step": 3059
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.857716741114005e-05,
+      "loss": 1.0791,
+      "step": 3060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.857601306659624e-05,
+      "loss": 1.0693,
+      "step": 3061
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.857485827126722e-05,
+      "loss": 0.7891,
+      "step": 3062
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.857370302518102e-05,
+      "loss": 1.0137,
+      "step": 3063
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8572547328365663e-05,
+      "loss": 0.9932,
+      "step": 3064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.857139118084921e-05,
+      "loss": 0.9854,
+      "step": 3065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8570234582659706e-05,
+      "loss": 1.0117,
+      "step": 3066
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8569077533825215e-05,
+      "loss": 1.0371,
+      "step": 3067
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.856792003437383e-05,
+      "loss": 0.9277,
+      "step": 3068
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.856676208433364e-05,
+      "loss": 1.0723,
+      "step": 3069
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.856560368373272e-05,
+      "loss": 0.7236,
+      "step": 3070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.856444483259921e-05,
+      "loss": 1.0713,
+      "step": 3071
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.856328553096121e-05,
+      "loss": 0.8604,
+      "step": 3072
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.856212577884687e-05,
+      "loss": 1.1299,
+      "step": 3073
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8560965576284324e-05,
+      "loss": 1.082,
+      "step": 3074
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.855980492330173e-05,
+      "loss": 1.0488,
+      "step": 3075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8558643819927254e-05,
+      "loss": 1.0469,
+      "step": 3076
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.855748226618907e-05,
+      "loss": 1.0117,
+      "step": 3077
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.855632026211537e-05,
+      "loss": 1.0635,
+      "step": 3078
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.855515780773435e-05,
+      "loss": 1.0918,
+      "step": 3079
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.855399490307421e-05,
+      "loss": 1.1045,
+      "step": 3080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8552831548163185e-05,
+      "loss": 1.0576,
+      "step": 3081
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.85516677430295e-05,
+      "loss": 1.0498,
+      "step": 3082
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.85505034877014e-05,
+      "loss": 1.042,
+      "step": 3083
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8549338782207135e-05,
+      "loss": 0.9648,
+      "step": 3084
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.854817362657497e-05,
+      "loss": 1.0508,
+      "step": 3085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.854700802083317e-05,
+      "loss": 0.9316,
+      "step": 3086
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.854584196501004e-05,
+      "loss": 1.1309,
+      "step": 3087
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8544675459133866e-05,
+      "loss": 1.0039,
+      "step": 3088
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.854350850323295e-05,
+      "loss": 0.7979,
+      "step": 3089
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.854234109733562e-05,
+      "loss": 1.0254,
+      "step": 3090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8541173241470203e-05,
+      "loss": 1.0586,
+      "step": 3091
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.854000493566504e-05,
+      "loss": 1.0342,
+      "step": 3092
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.853883617994848e-05,
+      "loss": 1.0332,
+      "step": 3093
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.853766697434889e-05,
+      "loss": 1.083,
+      "step": 3094
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.853649731889464e-05,
+      "loss": 1.0479,
+      "step": 3095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.853532721361411e-05,
+      "loss": 1.082,
+      "step": 3096
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8534156658535706e-05,
+      "loss": 0.877,
+      "step": 3097
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.853298565368782e-05,
+      "loss": 1.127,
+      "step": 3098
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8531814199098876e-05,
+      "loss": 1.0898,
+      "step": 3099
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.85306422947973e-05,
+      "loss": 1.0186,
+      "step": 3100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.852946994081154e-05,
+      "loss": 1.0801,
+      "step": 3101
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.852829713717003e-05,
+      "loss": 1.0771,
+      "step": 3102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8527123883901244e-05,
+      "loss": 0.8652,
+      "step": 3103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.852595018103364e-05,
+      "loss": 1.0195,
+      "step": 3104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.852477602859572e-05,
+      "loss": 1.043,
+      "step": 3105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.852360142661596e-05,
+      "loss": 1.1543,
+      "step": 3106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8522426375122866e-05,
+      "loss": 1.0537,
+      "step": 3107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.852125087414496e-05,
+      "loss": 0.8506,
+      "step": 3108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.852007492371076e-05,
+      "loss": 1.0342,
+      "step": 3109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.851889852384881e-05,
+      "loss": 1.0195,
+      "step": 3110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.851772167458765e-05,
+      "loss": 1.0107,
+      "step": 3111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.851654437595585e-05,
+      "loss": 1.1035,
+      "step": 3112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.851536662798197e-05,
+      "loss": 0.9668,
+      "step": 3113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8514188430694595e-05,
+      "loss": 0.9521,
+      "step": 3114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.851300978412231e-05,
+      "loss": 1.0488,
+      "step": 3115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.851183068829373e-05,
+      "loss": 1.1406,
+      "step": 3116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8510651143237455e-05,
+      "loss": 1.1348,
+      "step": 3117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.850947114898212e-05,
+      "loss": 1.1416,
+      "step": 3118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.850829070555635e-05,
+      "loss": 1.0801,
+      "step": 3119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.85071098129888e-05,
+      "loss": 1.1582,
+      "step": 3120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8505928471308116e-05,
+      "loss": 0.9463,
+      "step": 3121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.850474668054298e-05,
+      "loss": 1.0518,
+      "step": 3122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8503564440722054e-05,
+      "loss": 1.0244,
+      "step": 3123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.850238175187405e-05,
+      "loss": 1.0273,
+      "step": 3124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.850119861402764e-05,
+      "loss": 1.0176,
+      "step": 3125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.850001502721155e-05,
+      "loss": 1.1348,
+      "step": 3126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8498830991454504e-05,
+      "loss": 1.1104,
+      "step": 3127
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.849764650678524e-05,
+      "loss": 1.0322,
+      "step": 3128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.849646157323249e-05,
+      "loss": 0.9541,
+      "step": 3129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8495276190825015e-05,
+      "loss": 0.9043,
+      "step": 3130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8494090359591575e-05,
+      "loss": 0.9912,
+      "step": 3131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.849290407956096e-05,
+      "loss": 1.0166,
+      "step": 3132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.849171735076194e-05,
+      "loss": 0.9414,
+      "step": 3133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.849053017322332e-05,
+      "loss": 1.0625,
+      "step": 3134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8489342546973924e-05,
+      "loss": 1.1895,
+      "step": 3135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.848815447204255e-05,
+      "loss": 1.127,
+      "step": 3136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.848696594845804e-05,
+      "loss": 1.0859,
+      "step": 3137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.848577697624924e-05,
+      "loss": 1.0977,
+      "step": 3138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.848458755544499e-05,
+      "loss": 1.1113,
+      "step": 3139
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8483397686074166e-05,
+      "loss": 1.1553,
+      "step": 3140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.848220736816563e-05,
+      "loss": 0.9648,
+      "step": 3141
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.848101660174829e-05,
+      "loss": 1.1953,
+      "step": 3142
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.847982538685101e-05,
+      "loss": 1.1406,
+      "step": 3143
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.847863372350273e-05,
+      "loss": 1.1113,
+      "step": 3144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.847744161173235e-05,
+      "loss": 1.1406,
+      "step": 3145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.84762490515688e-05,
+      "loss": 1.042,
+      "step": 3146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8475056043041024e-05,
+      "loss": 1.0898,
+      "step": 3147
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8473862586177964e-05,
+      "loss": 1.1406,
+      "step": 3148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8472668681008595e-05,
+      "loss": 0.9688,
+      "step": 3149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.847147432756188e-05,
+      "loss": 1.1094,
+      "step": 3150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8470279525866804e-05,
+      "loss": 0.8457,
+      "step": 3151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.846908427595237e-05,
+      "loss": 1.2363,
+      "step": 3152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8467888577847576e-05,
+      "loss": 0.958,
+      "step": 3153
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8466692431581434e-05,
+      "loss": 0.9785,
+      "step": 3154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8465495837182977e-05,
+      "loss": 0.7466,
+      "step": 3155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.846429879468124e-05,
+      "loss": 1.1035,
+      "step": 3156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.846310130410527e-05,
+      "loss": 1.1074,
+      "step": 3157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.846190336548413e-05,
+      "loss": 0.8799,
+      "step": 3158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.84607049788469e-05,
+      "loss": 1.1055,
+      "step": 3159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.845950614422265e-05,
+      "loss": 1.002,
+      "step": 3160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8458306861640465e-05,
+      "loss": 1.043,
+      "step": 3161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8457107131129456e-05,
+      "loss": 1.0645,
+      "step": 3162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.845590695271874e-05,
+      "loss": 1.0381,
+      "step": 3163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.845470632643744e-05,
+      "loss": 1.1562,
+      "step": 3164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.845350525231469e-05,
+      "loss": 1.0557,
+      "step": 3165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.845230373037964e-05,
+      "loss": 1.0488,
+      "step": 3166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8451101760661436e-05,
+      "loss": 0.9707,
+      "step": 3167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8449899343189266e-05,
+      "loss": 0.9629,
+      "step": 3168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.844869647799229e-05,
+      "loss": 0.9453,
+      "step": 3169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8447493165099706e-05,
+      "loss": 1.1035,
+      "step": 3170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.844628940454072e-05,
+      "loss": 0.9678,
+      "step": 3171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8445085196344536e-05,
+      "loss": 0.9346,
+      "step": 3172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.844388054054038e-05,
+      "loss": 1.0713,
+      "step": 3173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.844267543715749e-05,
+      "loss": 0.9805,
+      "step": 3174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.84414698862251e-05,
+      "loss": 1.1426,
+      "step": 3175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8440263887772465e-05,
+      "loss": 0.8301,
+      "step": 3176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8439057441828863e-05,
+      "loss": 0.9912,
+      "step": 3177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.843785054842356e-05,
+      "loss": 0.873,
+      "step": 3178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8436643207585856e-05,
+      "loss": 1.0186,
+      "step": 3179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.843543541934504e-05,
+      "loss": 0.9766,
+      "step": 3180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.843422718373042e-05,
+      "loss": 0.9971,
+      "step": 3181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.843301850077132e-05,
+      "loss": 1.1064,
+      "step": 3182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.843180937049707e-05,
+      "loss": 1.0557,
+      "step": 3183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.843059979293701e-05,
+      "loss": 1.0098,
+      "step": 3184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8429389768120497e-05,
+      "loss": 1.0322,
+      "step": 3185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.842817929607689e-05,
+      "loss": 0.7969,
+      "step": 3186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8426968376835575e-05,
+      "loss": 0.9199,
+      "step": 3187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.842575701042592e-05,
+      "loss": 1.002,
+      "step": 3188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.842454519687734e-05,
+      "loss": 1.0459,
+      "step": 3189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8423332936219224e-05,
+      "loss": 1.001,
+      "step": 3190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8422120228481006e-05,
+      "loss": 0.9473,
+      "step": 3191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8420907073692106e-05,
+      "loss": 1.082,
+      "step": 3192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.841969347188196e-05,
+      "loss": 1.1318,
+      "step": 3193
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.841847942308003e-05,
+      "loss": 1.0947,
+      "step": 3194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.841726492731577e-05,
+      "loss": 1.1387,
+      "step": 3195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.841604998461865e-05,
+      "loss": 1.0586,
+      "step": 3196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.841483459501816e-05,
+      "loss": 1.0771,
+      "step": 3197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8413618758543785e-05,
+      "loss": 0.8291,
+      "step": 3198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.841240247522504e-05,
+      "loss": 0.958,
+      "step": 3199
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.841118574509144e-05,
+      "loss": 1.0078,
+      "step": 3200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.84099685681725e-05,
+      "loss": 1.1152,
+      "step": 3201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.840875094449777e-05,
+      "loss": 1.0996,
+      "step": 3202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.840753287409679e-05,
+      "loss": 0.9951,
+      "step": 3203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.840631435699912e-05,
+      "loss": 1.1094,
+      "step": 3204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.840509539323433e-05,
+      "loss": 1.041,
+      "step": 3205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8403875982832005e-05,
+      "loss": 1.0713,
+      "step": 3206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.840265612582174e-05,
+      "loss": 1.1035,
+      "step": 3207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.840143582223313e-05,
+      "loss": 0.9854,
+      "step": 3208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.840021507209578e-05,
+      "loss": 0.8418,
+      "step": 3209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8398993875439326e-05,
+      "loss": 0.8457,
+      "step": 3210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.83977722322934e-05,
+      "loss": 0.9775,
+      "step": 3211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8396550142687654e-05,
+      "loss": 0.8223,
+      "step": 3212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8395327606651734e-05,
+      "loss": 1.0381,
+      "step": 3213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.839410462421531e-05,
+      "loss": 0.9834,
+      "step": 3214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8392881195408066e-05,
+      "loss": 1.0771,
+      "step": 3215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8391657320259684e-05,
+      "loss": 1.0059,
+      "step": 3216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8390432998799864e-05,
+      "loss": 1.0264,
+      "step": 3217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8389208231058327e-05,
+      "loss": 0.9463,
+      "step": 3218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.838798301706478e-05,
+      "loss": 1.0762,
+      "step": 3219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.838675735684897e-05,
+      "loss": 1.0322,
+      "step": 3220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8385531250440624e-05,
+      "loss": 0.9346,
+      "step": 3221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8384304697869505e-05,
+      "loss": 1.1641,
+      "step": 3222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.838307769916538e-05,
+      "loss": 0.9717,
+      "step": 3223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.838185025435803e-05,
+      "loss": 0.9775,
+      "step": 3224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.838062236347722e-05,
+      "loss": 0.833,
+      "step": 3225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.837939402655277e-05,
+      "loss": 0.9912,
+      "step": 3226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.837816524361447e-05,
+      "loss": 0.9346,
+      "step": 3227
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.837693601469216e-05,
+      "loss": 0.9619,
+      "step": 3228
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.837570633981565e-05,
+      "loss": 1.0156,
+      "step": 3229
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8374476219014786e-05,
+      "loss": 1.0625,
+      "step": 3230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.837324565231942e-05,
+      "loss": 1.1484,
+      "step": 3231
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.837201463975943e-05,
+      "loss": 1.0791,
+      "step": 3232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.837078318136466e-05,
+      "loss": 1.1348,
+      "step": 3233
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8369551277165015e-05,
+      "loss": 0.918,
+      "step": 3234
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.836831892719038e-05,
+      "loss": 1.0635,
+      "step": 3235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.836708613147066e-05,
+      "loss": 0.9893,
+      "step": 3236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.836585289003579e-05,
+      "loss": 0.8594,
+      "step": 3237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8364619202915674e-05,
+      "loss": 1.0059,
+      "step": 3238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8363385070140256e-05,
+      "loss": 1.0488,
+      "step": 3239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.836215049173949e-05,
+      "loss": 0.79,
+      "step": 3240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.836091546774333e-05,
+      "loss": 0.9355,
+      "step": 3241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.835967999818175e-05,
+      "loss": 1.0605,
+      "step": 3242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.835844408308473e-05,
+      "loss": 0.8135,
+      "step": 3243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.835720772248227e-05,
+      "loss": 1.125,
+      "step": 3244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.835597091640436e-05,
+      "loss": 1.1797,
+      "step": 3245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.835473366488101e-05,
+      "loss": 0.7471,
+      "step": 3246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.835349596794226e-05,
+      "loss": 0.8613,
+      "step": 3247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.835225782561814e-05,
+      "loss": 1.042,
+      "step": 3248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8351019237938695e-05,
+      "loss": 0.9541,
+      "step": 3249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.834978020493398e-05,
+      "loss": 1.166,
+      "step": 3250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.834854072663406e-05,
+      "loss": 1.0439,
+      "step": 3251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.834730080306903e-05,
+      "loss": 1.0146,
+      "step": 3252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.834606043426896e-05,
+      "loss": 1.0596,
+      "step": 3253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.834481962026395e-05,
+      "loss": 0.8867,
+      "step": 3254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.834357836108413e-05,
+      "loss": 1.0176,
+      "step": 3255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8342336656759605e-05,
+      "loss": 1.0674,
+      "step": 3256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8341094507320505e-05,
+      "loss": 0.8887,
+      "step": 3257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.833985191279699e-05,
+      "loss": 0.9053,
+      "step": 3258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.83386088732192e-05,
+      "loss": 0.8857,
+      "step": 3259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.833736538861731e-05,
+      "loss": 1.0898,
+      "step": 3260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8336121459021484e-05,
+      "loss": 1.0938,
+      "step": 3261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.833487708446192e-05,
+      "loss": 0.9775,
+      "step": 3262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.833363226496881e-05,
+      "loss": 1.0117,
+      "step": 3263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8332387000572354e-05,
+      "loss": 1.0078,
+      "step": 3264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8331141291302784e-05,
+      "loss": 1.0859,
+      "step": 3265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.832989513719033e-05,
+      "loss": 0.9395,
+      "step": 3266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8328648538265215e-05,
+      "loss": 0.9854,
+      "step": 3267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8327401494557714e-05,
+      "loss": 0.9336,
+      "step": 3268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8326154006098075e-05,
+      "loss": 0.9961,
+      "step": 3269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.832490607291657e-05,
+      "loss": 1.0586,
+      "step": 3270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8323657695043486e-05,
+      "loss": 1.0195,
+      "step": 3271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.832240887250913e-05,
+      "loss": 1.084,
+      "step": 3272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.832115960534378e-05,
+      "loss": 0.8408,
+      "step": 3273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.831990989357778e-05,
+      "loss": 0.9971,
+      "step": 3274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.831865973724143e-05,
+      "loss": 1.0957,
+      "step": 3275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8317409136365086e-05,
+      "loss": 1.0859,
+      "step": 3276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.831615809097909e-05,
+      "loss": 0.8555,
+      "step": 3277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.831490660111381e-05,
+      "loss": 0.9814,
+      "step": 3278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.831365466679961e-05,
+      "loss": 0.917,
+      "step": 3279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.831240228806686e-05,
+      "loss": 1.0254,
+      "step": 3280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.831114946494597e-05,
+      "loss": 1.1973,
+      "step": 3281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.830989619746733e-05,
+      "loss": 0.9092,
+      "step": 3282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.830864248566135e-05,
+      "loss": 0.8496,
+      "step": 3283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.830738832955847e-05,
+      "loss": 1.1006,
+      "step": 3284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8306133729189115e-05,
+      "loss": 0.9961,
+      "step": 3285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.830487868458373e-05,
+      "loss": 1.001,
+      "step": 3286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.830362319577277e-05,
+      "loss": 0.9492,
+      "step": 3287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8302367262786704e-05,
+      "loss": 1.0293,
+      "step": 3288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.830111088565601e-05,
+      "loss": 0.9541,
+      "step": 3289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.829985406441118e-05,
+      "loss": 1.0371,
+      "step": 3290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8298596799082705e-05,
+      "loss": 1.0264,
+      "step": 3291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8297339089701104e-05,
+      "loss": 1.0381,
+      "step": 3292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8296080936296884e-05,
+      "loss": 1.1035,
+      "step": 3293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8294822338900604e-05,
+      "loss": 1.0996,
+      "step": 3294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.829356329754277e-05,
+      "loss": 0.8877,
+      "step": 3295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8292303812253965e-05,
+      "loss": 1.0107,
+      "step": 3296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.829104388306475e-05,
+      "loss": 0.958,
+      "step": 3297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8289783510005674e-05,
+      "loss": 0.9062,
+      "step": 3298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.828852269310735e-05,
+      "loss": 1.0576,
+      "step": 3299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.828726143240037e-05,
+      "loss": 1.1748,
+      "step": 3300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8285999727915324e-05,
+      "loss": 0.8047,
+      "step": 3301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8284737579682845e-05,
+      "loss": 0.8662,
+      "step": 3302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.828347498773356e-05,
+      "loss": 0.9658,
+      "step": 3303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.828221195209811e-05,
+      "loss": 1.0293,
+      "step": 3304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.828094847280714e-05,
+      "loss": 1.1201,
+      "step": 3305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.827968454989131e-05,
+      "loss": 0.9326,
+      "step": 3306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.82784201833813e-05,
+      "loss": 1.0332,
+      "step": 3307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8277155373307786e-05,
+      "loss": 1.0186,
+      "step": 3308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.827589011970146e-05,
+      "loss": 0.9775,
+      "step": 3309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.827462442259303e-05,
+      "loss": 1.083,
+      "step": 3310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8273358282013206e-05,
+      "loss": 1.1387,
+      "step": 3311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8272091697992714e-05,
+      "loss": 1.2207,
+      "step": 3312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8270824670562305e-05,
+      "loss": 1.0684,
+      "step": 3313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.82695571997527e-05,
+      "loss": 0.9375,
+      "step": 3314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.826828928559468e-05,
+      "loss": 1.0742,
+      "step": 3315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8267020928119e-05,
+      "loss": 1.0889,
+      "step": 3316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.826575212735644e-05,
+      "loss": 0.7939,
+      "step": 3317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8264482883337806e-05,
+      "loss": 1.04,
+      "step": 3318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.826321319609387e-05,
+      "loss": 0.876,
+      "step": 3319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.826194306565547e-05,
+      "loss": 1.0625,
+      "step": 3320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8260672492053414e-05,
+      "loss": 0.9434,
+      "step": 3321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8259401475318547e-05,
+      "loss": 0.9482,
+      "step": 3322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8258130015481696e-05,
+      "loss": 1.1855,
+      "step": 3323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.825685811257373e-05,
+      "loss": 1.0283,
+      "step": 3324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.825558576662551e-05,
+      "loss": 0.9238,
+      "step": 3325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.82543129776679e-05,
+      "loss": 1.0693,
+      "step": 3326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.825303974573181e-05,
+      "loss": 1.0342,
+      "step": 3327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8251766070848125e-05,
+      "loss": 1.0381,
+      "step": 3328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.825049195304775e-05,
+      "loss": 0.8926,
+      "step": 3329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.824921739236161e-05,
+      "loss": 1.1465,
+      "step": 3330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8247942388820634e-05,
+      "loss": 0.9287,
+      "step": 3331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.824666694245576e-05,
+      "loss": 0.8398,
+      "step": 3332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8245391053297936e-05,
+      "loss": 1.0703,
+      "step": 3333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.824411472137814e-05,
+      "loss": 0.9307,
+      "step": 3334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.824283794672732e-05,
+      "loss": 1.0723,
+      "step": 3335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.824156072937649e-05,
+      "loss": 0.9346,
+      "step": 3336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.824028306935662e-05,
+      "loss": 1.0039,
+      "step": 3337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8239004966698715e-05,
+      "loss": 1.0566,
+      "step": 3338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.82377264214338e-05,
+      "loss": 0.835,
+      "step": 3339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.823644743359291e-05,
+      "loss": 0.9844,
+      "step": 3340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8235168003207065e-05,
+      "loss": 0.9795,
+      "step": 3341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.823388813030732e-05,
+      "loss": 0.9062,
+      "step": 3342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.823260781492473e-05,
+      "loss": 1.2188,
+      "step": 3343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.823132705709038e-05,
+      "loss": 1.0127,
+      "step": 3344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.823004585683533e-05,
+      "loss": 0.9873,
+      "step": 3345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.822876421419068e-05,
+      "loss": 1.0381,
+      "step": 3346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.822748212918754e-05,
+      "loss": 0.7578,
+      "step": 3347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8226199601857005e-05,
+      "loss": 0.8467,
+      "step": 3348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8224916632230214e-05,
+      "loss": 1.0205,
+      "step": 3349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.822363322033828e-05,
+      "loss": 0.9971,
+      "step": 3350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.822234936621238e-05,
+      "loss": 0.9521,
+      "step": 3351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8221065069883643e-05,
+      "loss": 0.8945,
+      "step": 3352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.821978033138324e-05,
+      "loss": 1.0381,
+      "step": 3353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8218495150742354e-05,
+      "loss": 1.0234,
+      "step": 3354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.821720952799217e-05,
+      "loss": 1.1641,
+      "step": 3355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.821592346316388e-05,
+      "loss": 0.8682,
+      "step": 3356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.82146369562887e-05,
+      "loss": 0.7588,
+      "step": 3357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8213350007397856e-05,
+      "loss": 1.0469,
+      "step": 3358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.821206261652257e-05,
+      "loss": 0.8496,
+      "step": 3359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.821077478369407e-05,
+      "loss": 1.1133,
+      "step": 3360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.820948650894364e-05,
+      "loss": 1.0928,
+      "step": 3361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.820819779230251e-05,
+      "loss": 1.0928,
+      "step": 3362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8206908633801975e-05,
+      "loss": 0.8906,
+      "step": 3363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.820561903347331e-05,
+      "loss": 0.9697,
+      "step": 3364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.820432899134781e-05,
+      "loss": 0.9971,
+      "step": 3365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.820303850745679e-05,
+      "loss": 1.0322,
+      "step": 3366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.820174758183155e-05,
+      "loss": 1.1113,
+      "step": 3367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.820045621450343e-05,
+      "loss": 0.7969,
+      "step": 3368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.819916440550377e-05,
+      "loss": 1.0781,
+      "step": 3369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.819787215486391e-05,
+      "loss": 1.0801,
+      "step": 3370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.81965794626152e-05,
+      "loss": 0.9785,
+      "step": 3371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8195286328789023e-05,
+      "loss": 0.8906,
+      "step": 3372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8193992753416764e-05,
+      "loss": 0.9668,
+      "step": 3373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.819269873652981e-05,
+      "loss": 1.1152,
+      "step": 3374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.819140427815955e-05,
+      "loss": 0.8809,
+      "step": 3375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8190109378337413e-05,
+      "loss": 1.0439,
+      "step": 3376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.818881403709481e-05,
+      "loss": 0.8994,
+      "step": 3377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.818751825446319e-05,
+      "loss": 1.1875,
+      "step": 3378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8186222030473986e-05,
+      "loss": 1.209,
+      "step": 3379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.818492536515866e-05,
+      "loss": 1.0029,
+      "step": 3380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.818362825854867e-05,
+      "loss": 0.8457,
+      "step": 3381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.818233071067551e-05,
+      "loss": 1.0459,
+      "step": 3382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8181032721570644e-05,
+      "loss": 1.0303,
+      "step": 3383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.817973429126558e-05,
+      "loss": 1.0098,
+      "step": 3384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8178435419791835e-05,
+      "loss": 0.8936,
+      "step": 3385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.817713610718092e-05,
+      "loss": 0.9053,
+      "step": 3386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8175836353464375e-05,
+      "loss": 1.2031,
+      "step": 3387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.817453615867373e-05,
+      "loss": 0.9795,
+      "step": 3388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8173235522840544e-05,
+      "loss": 0.9609,
+      "step": 3389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.817193444599638e-05,
+      "loss": 0.9658,
+      "step": 3390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8170632928172806e-05,
+      "loss": 0.9609,
+      "step": 3391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.816933096940141e-05,
+      "loss": 0.9902,
+      "step": 3392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.816802856971378e-05,
+      "loss": 0.8984,
+      "step": 3393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8166725729141536e-05,
+      "loss": 1.123,
+      "step": 3394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.816542244771627e-05,
+      "loss": 0.9697,
+      "step": 3395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8164118725469644e-05,
+      "loss": 1.0713,
+      "step": 3396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.816281456243326e-05,
+      "loss": 1.1289,
+      "step": 3397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.816150995863879e-05,
+      "loss": 1.0791,
+      "step": 3398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.816020491411788e-05,
+      "loss": 0.7129,
+      "step": 3399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.81588994289022e-05,
+      "loss": 1.1426,
+      "step": 3400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.815759350302344e-05,
+      "loss": 0.9756,
+      "step": 3401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.815628713651329e-05,
+      "loss": 1.0879,
+      "step": 3402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.815498032940344e-05,
+      "loss": 0.9932,
+      "step": 3403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.815367308172561e-05,
+      "loss": 0.9121,
+      "step": 3404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.815236539351152e-05,
+      "loss": 1.0898,
+      "step": 3405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.815105726479291e-05,
+      "loss": 0.834,
+      "step": 3406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.814974869560152e-05,
+      "loss": 1.0078,
+      "step": 3407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8148439685969106e-05,
+      "loss": 1.1738,
+      "step": 3408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.814713023592743e-05,
+      "loss": 1.0762,
+      "step": 3409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8145820345508274e-05,
+      "loss": 0.9121,
+      "step": 3410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.814451001474343e-05,
+      "loss": 0.9658,
+      "step": 3411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8143199243664675e-05,
+      "loss": 0.835,
+      "step": 3412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8141888032303846e-05,
+      "loss": 1.0078,
+      "step": 3413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.814057638069274e-05,
+      "loss": 0.9961,
+      "step": 3414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.813926428886319e-05,
+      "loss": 1.0088,
+      "step": 3415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.813795175684705e-05,
+      "loss": 0.8955,
+      "step": 3416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8136638784676165e-05,
+      "loss": 0.8945,
+      "step": 3417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8135325372382384e-05,
+      "loss": 0.9326,
+      "step": 3418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.813401151999759e-05,
+      "loss": 0.9746,
+      "step": 3419
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.813269722755368e-05,
+      "loss": 1.1641,
+      "step": 3420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.813138249508253e-05,
+      "loss": 0.8867,
+      "step": 3421
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8130067322616045e-05,
+      "loss": 1.0498,
+      "step": 3422
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.812875171018614e-05,
+      "loss": 0.8809,
+      "step": 3423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.812743565782475e-05,
+      "loss": 1.0283,
+      "step": 3424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.812611916556381e-05,
+      "loss": 1.0068,
+      "step": 3425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.812480223343527e-05,
+      "loss": 1.0107,
+      "step": 3426
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8123484861471076e-05,
+      "loss": 0.9131,
+      "step": 3427
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.81221670497032e-05,
+      "loss": 0.9072,
+      "step": 3428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.812084879816363e-05,
+      "loss": 0.877,
+      "step": 3429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.811953010688435e-05,
+      "loss": 0.9609,
+      "step": 3430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8118210975897366e-05,
+      "loss": 1.1543,
+      "step": 3431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.811689140523468e-05,
+      "loss": 0.9844,
+      "step": 3432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8115571394928324e-05,
+      "loss": 1.0381,
+      "step": 3433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8114250945010324e-05,
+      "loss": 1.0693,
+      "step": 3434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8112930055512724e-05,
+      "loss": 1.0361,
+      "step": 3435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.811160872646758e-05,
+      "loss": 1.1055,
+      "step": 3436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.811028695790696e-05,
+      "loss": 1.1152,
+      "step": 3437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.810896474986294e-05,
+      "loss": 1.0918,
+      "step": 3438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.810764210236759e-05,
+      "loss": 1.0566,
+      "step": 3439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.810631901545303e-05,
+      "loss": 0.9307,
+      "step": 3440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.810499548915135e-05,
+      "loss": 1.1074,
+      "step": 3441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.810367152349467e-05,
+      "loss": 1.0811,
+      "step": 3442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8102347118515134e-05,
+      "loss": 0.8926,
+      "step": 3443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.810102227424486e-05,
+      "loss": 0.9717,
+      "step": 3444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.809969699071601e-05,
+      "loss": 1.083,
+      "step": 3445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.809837126796075e-05,
+      "loss": 1.001,
+      "step": 3446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.809704510601124e-05,
+      "loss": 0.9492,
+      "step": 3447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8095718504899675e-05,
+      "loss": 1.0469,
+      "step": 3448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8094391464658234e-05,
+      "loss": 1.0801,
+      "step": 3449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.809306398531912e-05,
+      "loss": 0.9053,
+      "step": 3450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.809173606691456e-05,
+      "loss": 1.1289,
+      "step": 3451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.809040770947677e-05,
+      "loss": 1.1309,
+      "step": 3452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8089078913037984e-05,
+      "loss": 0.9697,
+      "step": 3453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.808774967763045e-05,
+      "loss": 1.1406,
+      "step": 3454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.808642000328643e-05,
+      "loss": 1.0918,
+      "step": 3455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.808508989003819e-05,
+      "loss": 1.002,
+      "step": 3456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8083759337918e-05,
+      "loss": 1.1279,
+      "step": 3457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.808242834695815e-05,
+      "loss": 1.0312,
+      "step": 3458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8081096917190944e-05,
+      "loss": 1.0693,
+      "step": 3459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.807976504864869e-05,
+      "loss": 0.9248,
+      "step": 3460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8078432741363716e-05,
+      "loss": 0.8911,
+      "step": 3461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.807709999536834e-05,
+      "loss": 0.8091,
+      "step": 3462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.807576681069491e-05,
+      "loss": 0.9766,
+      "step": 3463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.807443318737578e-05,
+      "loss": 1.2227,
+      "step": 3464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.807309912544331e-05,
+      "loss": 0.9043,
+      "step": 3465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8071764624929875e-05,
+      "loss": 0.8369,
+      "step": 3466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.807042968586786e-05,
+      "loss": 0.9844,
+      "step": 3467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8069094308289664e-05,
+      "loss": 1.0449,
+      "step": 3468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8067758492227685e-05,
+      "loss": 1.2041,
+      "step": 3469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.806642223771434e-05,
+      "loss": 1.0215,
+      "step": 3470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.806508554478206e-05,
+      "loss": 0.9844,
+      "step": 3471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.806374841346329e-05,
+      "loss": 1.1309,
+      "step": 3472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.806241084379046e-05,
+      "loss": 0.9443,
+      "step": 3473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.806107283579604e-05,
+      "loss": 0.9316,
+      "step": 3474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8059734389512496e-05,
+      "loss": 1.0557,
+      "step": 3475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.805839550497232e-05,
+      "loss": 0.9844,
+      "step": 3476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.805705618220798e-05,
+      "loss": 1.042,
+      "step": 3477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8055716421252e-05,
+      "loss": 1.0381,
+      "step": 3478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.805437622213688e-05,
+      "loss": 1.0039,
+      "step": 3479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.805303558489514e-05,
+      "loss": 0.9136,
+      "step": 3480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8051694509559324e-05,
+      "loss": 1.2637,
+      "step": 3481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8050352996161974e-05,
+      "loss": 1.0459,
+      "step": 3482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8049011044735636e-05,
+      "loss": 0.9268,
+      "step": 3483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.804766865531288e-05,
+      "loss": 1.0449,
+      "step": 3484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.804632582792627e-05,
+      "loss": 1.0537,
+      "step": 3485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.804498256260843e-05,
+      "loss": 1.1221,
+      "step": 3486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.804363885939191e-05,
+      "loss": 1.0049,
+      "step": 3487
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.804229471830935e-05,
+      "loss": 1.0918,
+      "step": 3488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.804095013939335e-05,
+      "loss": 1.0342,
+      "step": 3489
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8039605122676555e-05,
+      "loss": 1.1035,
+      "step": 3490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8038259668191596e-05,
+      "loss": 0.9736,
+      "step": 3491
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8036913775971115e-05,
+      "loss": 1.0049,
+      "step": 3492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.803556744604779e-05,
+      "loss": 0.9473,
+      "step": 3493
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.803422067845428e-05,
+      "loss": 0.9297,
+      "step": 3494
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.803287347322327e-05,
+      "loss": 1.1875,
+      "step": 3495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8031525830387454e-05,
+      "loss": 1.1523,
+      "step": 3496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.803017774997954e-05,
+      "loss": 1.0205,
+      "step": 3497
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.802882923203223e-05,
+      "loss": 1.0254,
+      "step": 3498
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.802748027657826e-05,
+      "loss": 0.915,
+      "step": 3499
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.802613088365036e-05,
+      "loss": 0.9126,
+      "step": 3500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8024781053281276e-05,
+      "loss": 1.125,
+      "step": 3501
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8023430785503764e-05,
+      "loss": 0.9414,
+      "step": 3502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.802208008035059e-05,
+      "loss": 0.9863,
+      "step": 3503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8020728937854534e-05,
+      "loss": 1.0479,
+      "step": 3504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.801937735804838e-05,
+      "loss": 0.7627,
+      "step": 3505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.801802534096494e-05,
+      "loss": 1.0596,
+      "step": 3506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.801667288663701e-05,
+      "loss": 1.0986,
+      "step": 3507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.801531999509742e-05,
+      "loss": 0.7607,
+      "step": 3508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8013966666378986e-05,
+      "loss": 1.1465,
+      "step": 3509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8012612900514555e-05,
+      "loss": 1.0684,
+      "step": 3510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8011258697537e-05,
+      "loss": 0.9375,
+      "step": 3511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.800990405747914e-05,
+      "loss": 0.9287,
+      "step": 3512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.800854898037389e-05,
+      "loss": 0.9443,
+      "step": 3513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8007193466254116e-05,
+      "loss": 1.0098,
+      "step": 3514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8005837515152715e-05,
+      "loss": 1.0312,
+      "step": 3515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8004481127102586e-05,
+      "loss": 1.0928,
+      "step": 3516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.800312430213665e-05,
+      "loss": 0.9473,
+      "step": 3517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.800176704028784e-05,
+      "loss": 0.9434,
+      "step": 3518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.800040934158908e-05,
+      "loss": 1.1191,
+      "step": 3519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.799905120607332e-05,
+      "loss": 1.0078,
+      "step": 3520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.799769263377352e-05,
+      "loss": 0.9004,
+      "step": 3521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.799633362472265e-05,
+      "loss": 0.9805,
+      "step": 3522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.799497417895369e-05,
+      "loss": 0.8511,
+      "step": 3523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7993614296499634e-05,
+      "loss": 1.0361,
+      "step": 3524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.799225397739346e-05,
+      "loss": 1.0801,
+      "step": 3525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.799089322166821e-05,
+      "loss": 1.0693,
+      "step": 3526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.798953202935689e-05,
+      "loss": 1.1338,
+      "step": 3527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.798817040049253e-05,
+      "loss": 0.8691,
+      "step": 3528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.798680833510817e-05,
+      "loss": 0.9375,
+      "step": 3529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.798544583323688e-05,
+      "loss": 0.999,
+      "step": 3530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.79840828949117e-05,
+      "loss": 0.9277,
+      "step": 3531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.798271952016573e-05,
+      "loss": 1.1172,
+      "step": 3532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7981355709032034e-05,
+      "loss": 0.9463,
+      "step": 3533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.797999146154372e-05,
+      "loss": 0.9307,
+      "step": 3534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.797862677773389e-05,
+      "loss": 1.1543,
+      "step": 3535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7977261657635655e-05,
+      "loss": 0.915,
+      "step": 3536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.797589610128216e-05,
+      "loss": 0.999,
+      "step": 3537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7974530108706524e-05,
+      "loss": 1.1855,
+      "step": 3538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.797316367994191e-05,
+      "loss": 1.0156,
+      "step": 3539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.797179681502146e-05,
+      "loss": 0.9658,
+      "step": 3540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.797042951397836e-05,
+      "loss": 0.7046,
+      "step": 3541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7969061776845794e-05,
+      "loss": 0.8794,
+      "step": 3542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7967693603656936e-05,
+      "loss": 0.8057,
+      "step": 3543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.796632499444499e-05,
+      "loss": 0.916,
+      "step": 3544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.796495594924318e-05,
+      "loss": 0.9922,
+      "step": 3545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7963586468084723e-05,
+      "loss": 1.0527,
+      "step": 3546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7962216551002844e-05,
+      "loss": 1.0361,
+      "step": 3547
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7960846198030804e-05,
+      "loss": 1.0664,
+      "step": 3548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.795947540920185e-05,
+      "loss": 1.0928,
+      "step": 3549
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.795810418454923e-05,
+      "loss": 1.0215,
+      "step": 3550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.795673252410624e-05,
+      "loss": 1.0879,
+      "step": 3551
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.795536042790617e-05,
+      "loss": 0.9268,
+      "step": 3552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.795398789598229e-05,
+      "loss": 0.9199,
+      "step": 3553
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.795261492836794e-05,
+      "loss": 1.0723,
+      "step": 3554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7951241525096414e-05,
+      "loss": 1.0801,
+      "step": 3555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.794986768620105e-05,
+      "loss": 1.0166,
+      "step": 3556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.794849341171519e-05,
+      "loss": 1.1699,
+      "step": 3557
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7947118701672184e-05,
+      "loss": 1.1152,
+      "step": 3558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7945743556105376e-05,
+      "loss": 1.0146,
+      "step": 3559
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.794436797504815e-05,
+      "loss": 1.0645,
+      "step": 3560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.794299195853389e-05,
+      "loss": 1.0557,
+      "step": 3561
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.794161550659598e-05,
+      "loss": 1.0029,
+      "step": 3562
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.794023861926783e-05,
+      "loss": 0.96,
+      "step": 3563
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.793886129658285e-05,
+      "loss": 0.8242,
+      "step": 3564
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.793748353857446e-05,
+      "loss": 1.0068,
+      "step": 3565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.79361053452761e-05,
+      "loss": 1.041,
+      "step": 3566
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.79347267167212e-05,
+      "loss": 0.9854,
+      "step": 3567
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.793334765294324e-05,
+      "loss": 0.9688,
+      "step": 3568
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.793196815397567e-05,
+      "loss": 0.9717,
+      "step": 3569
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.793058821985197e-05,
+      "loss": 0.9141,
+      "step": 3570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7929207850605626e-05,
+      "loss": 0.9385,
+      "step": 3571
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7927827046270124e-05,
+      "loss": 1.0918,
+      "step": 3572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.792644580687899e-05,
+      "loss": 1.0107,
+      "step": 3573
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.792506413246574e-05,
+      "loss": 1.1045,
+      "step": 3574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7923682023063896e-05,
+      "loss": 1.0381,
+      "step": 3575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7922299478707004e-05,
+      "loss": 1.0283,
+      "step": 3576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7920916499428605e-05,
+      "loss": 1.0713,
+      "step": 3577
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7919533085262274e-05,
+      "loss": 0.9121,
+      "step": 3578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.791814923624157e-05,
+      "loss": 0.9502,
+      "step": 3579
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7916764952400076e-05,
+      "loss": 0.9355,
+      "step": 3580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7915380233771395e-05,
+      "loss": 0.7876,
+      "step": 3581
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.791399508038912e-05,
+      "loss": 1.0469,
+      "step": 3582
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.791260949228687e-05,
+      "loss": 0.9688,
+      "step": 3583
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.791122346949826e-05,
+      "loss": 1.0928,
+      "step": 3584
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.790983701205693e-05,
+      "loss": 0.9258,
+      "step": 3585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.790845011999654e-05,
+      "loss": 1.0596,
+      "step": 3586
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.790706279335072e-05,
+      "loss": 1.0,
+      "step": 3587
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.790567503215316e-05,
+      "loss": 0.9961,
+      "step": 3588
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.790428683643752e-05,
+      "loss": 1.1816,
+      "step": 3589
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7902898206237487e-05,
+      "loss": 1.0527,
+      "step": 3590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.790150914158677e-05,
+      "loss": 1.166,
+      "step": 3591
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.790011964251907e-05,
+      "loss": 1.1182,
+      "step": 3592
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.789872970906811e-05,
+      "loss": 0.8633,
+      "step": 3593
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7897339341267623e-05,
+      "loss": 1.0127,
+      "step": 3594
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.789594853915134e-05,
+      "loss": 1.0156,
+      "step": 3595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.789455730275302e-05,
+      "loss": 1.0625,
+      "step": 3596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.789316563210642e-05,
+      "loss": 0.8838,
+      "step": 3597
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.78917735272453e-05,
+      "loss": 0.8301,
+      "step": 3598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.789038098820347e-05,
+      "loss": 1.0811,
+      "step": 3599
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.78889880150147e-05,
+      "loss": 1.1133,
+      "step": 3600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.78875946077128e-05,
+      "loss": 1.0967,
+      "step": 3601
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.788620076633159e-05,
+      "loss": 0.9902,
+      "step": 3602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.788480649090488e-05,
+      "loss": 0.9482,
+      "step": 3603
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.788341178146652e-05,
+      "loss": 1.0361,
+      "step": 3604
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.788201663805035e-05,
+      "loss": 1.1641,
+      "step": 3605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.788062106069022e-05,
+      "loss": 1.0098,
+      "step": 3606
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7879225049420005e-05,
+      "loss": 0.8506,
+      "step": 3607
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7877828604273584e-05,
+      "loss": 0.998,
+      "step": 3608
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.787643172528482e-05,
+      "loss": 1.0615,
+      "step": 3609
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7875034412487656e-05,
+      "loss": 0.8896,
+      "step": 3610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7873636665915966e-05,
+      "loss": 1.1719,
+      "step": 3611
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7872238485603674e-05,
+      "loss": 1.083,
+      "step": 3612
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7870839871584716e-05,
+      "loss": 0.9346,
+      "step": 3613
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.786944082389303e-05,
+      "loss": 1.1523,
+      "step": 3614
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.786804134256257e-05,
+      "loss": 0.8945,
+      "step": 3615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7866641427627294e-05,
+      "loss": 1.0781,
+      "step": 3616
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7865241079121175e-05,
+      "loss": 0.8721,
+      "step": 3617
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7863840297078196e-05,
+      "loss": 0.873,
+      "step": 3618
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.786243908153235e-05,
+      "loss": 1.1152,
+      "step": 3619
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.786103743251763e-05,
+      "loss": 1.1836,
+      "step": 3620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7859635350068066e-05,
+      "loss": 0.8789,
+      "step": 3621
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7858232834217675e-05,
+      "loss": 0.8555,
+      "step": 3622
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.785682988500049e-05,
+      "loss": 1.0293,
+      "step": 3623
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.785542650245056e-05,
+      "loss": 0.9033,
+      "step": 3624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.785402268660194e-05,
+      "loss": 0.9766,
+      "step": 3625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7852618437488695e-05,
+      "loss": 0.9365,
+      "step": 3626
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.78512137551449e-05,
+      "loss": 1.0459,
+      "step": 3627
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7849808639604655e-05,
+      "loss": 1.0078,
+      "step": 3628
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.784840309090204e-05,
+      "loss": 1.1729,
+      "step": 3629
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.784699710907118e-05,
+      "loss": 0.9717,
+      "step": 3630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.784559069414618e-05,
+      "loss": 0.8633,
+      "step": 3631
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.784418384616118e-05,
+      "loss": 0.8291,
+      "step": 3632
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7842776565150314e-05,
+      "loss": 0.9727,
+      "step": 3633
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.784136885114773e-05,
+      "loss": 1.2051,
+      "step": 3634
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7839960704187604e-05,
+      "loss": 1.1025,
+      "step": 3635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.783855212430409e-05,
+      "loss": 1.1162,
+      "step": 3636
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.783714311153138e-05,
+      "loss": 1.1182,
+      "step": 3637
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.783573366590366e-05,
+      "loss": 1.0684,
+      "step": 3638
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.783432378745514e-05,
+      "loss": 1.0957,
+      "step": 3639
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7832913476220026e-05,
+      "loss": 0.7871,
+      "step": 3640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.783150273223255e-05,
+      "loss": 1.1582,
+      "step": 3641
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.783009155552694e-05,
+      "loss": 0.9688,
+      "step": 3642
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.782867994613745e-05,
+      "loss": 1.0381,
+      "step": 3643
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7827267904098335e-05,
+      "loss": 0.9365,
+      "step": 3644
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7825855429443844e-05,
+      "loss": 1.0176,
+      "step": 3645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.782444252220827e-05,
+      "loss": 0.9395,
+      "step": 3646
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.78230291824259e-05,
+      "loss": 1.0078,
+      "step": 3647
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.782161541013103e-05,
+      "loss": 1.1455,
+      "step": 3648
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.782020120535795e-05,
+      "loss": 1.0996,
+      "step": 3649
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7818786568141e-05,
+      "loss": 0.8965,
+      "step": 3650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7817371498514514e-05,
+      "loss": 0.9268,
+      "step": 3651
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.781595599651281e-05,
+      "loss": 1.0391,
+      "step": 3652
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.781454006217026e-05,
+      "loss": 1.0576,
+      "step": 3653
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.78131236955212e-05,
+      "loss": 1.0391,
+      "step": 3654
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.781170689660002e-05,
+      "loss": 1.1387,
+      "step": 3655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7810289665441094e-05,
+      "loss": 0.9854,
+      "step": 3656
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7808872002078815e-05,
+      "loss": 0.9785,
+      "step": 3657
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.780745390654759e-05,
+      "loss": 1.0381,
+      "step": 3658
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.780603537888183e-05,
+      "loss": 0.959,
+      "step": 3659
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.780461641911595e-05,
+      "loss": 1.0557,
+      "step": 3660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.78031970272844e-05,
+      "loss": 1.0947,
+      "step": 3661
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.780177720342161e-05,
+      "loss": 1.0215,
+      "step": 3662
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.780035694756204e-05,
+      "loss": 1.0986,
+      "step": 3663
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.779893625974016e-05,
+      "loss": 0.9873,
+      "step": 3664
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7797515139990436e-05,
+      "loss": 1.0469,
+      "step": 3665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.779609358834737e-05,
+      "loss": 1.0205,
+      "step": 3666
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.779467160484544e-05,
+      "loss": 1.0059,
+      "step": 3667
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7793249189519167e-05,
+      "loss": 1.0381,
+      "step": 3668
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.779182634240306e-05,
+      "loss": 0.9893,
+      "step": 3669
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.779040306353166e-05,
+      "loss": 0.709,
+      "step": 3670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.778897935293948e-05,
+      "loss": 1.1465,
+      "step": 3671
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7787555210661106e-05,
+      "loss": 1.0977,
+      "step": 3672
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.778613063673107e-05,
+      "loss": 0.9189,
+      "step": 3673
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7784705631183946e-05,
+      "loss": 1.0381,
+      "step": 3674
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7783280194054327e-05,
+      "loss": 1.0996,
+      "step": 3675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7781854325376796e-05,
+      "loss": 1.0107,
+      "step": 3676
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7780428025185954e-05,
+      "loss": 0.8457,
+      "step": 3677
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7779001293516406e-05,
+      "loss": 1.0137,
+      "step": 3678
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.777757413040279e-05,
+      "loss": 1.1895,
+      "step": 3679
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.777614653587973e-05,
+      "loss": 1.0938,
+      "step": 3680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.777471850998187e-05,
+      "loss": 1.0811,
+      "step": 3681
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.777329005274387e-05,
+      "loss": 1.041,
+      "step": 3682
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.777186116420039e-05,
+      "loss": 0.8711,
+      "step": 3683
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.77704318443861e-05,
+      "loss": 1.04,
+      "step": 3684
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7769002093335695e-05,
+      "loss": 0.9092,
+      "step": 3685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.776757191108386e-05,
+      "loss": 1.1143,
+      "step": 3686
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.776614129766532e-05,
+      "loss": 0.9268,
+      "step": 3687
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7764710253114765e-05,
+      "loss": 1.0684,
+      "step": 3688
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7763278777466936e-05,
+      "loss": 0.8813,
+      "step": 3689
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.776184687075658e-05,
+      "loss": 1.0469,
+      "step": 3690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.776041453301843e-05,
+      "loss": 0.9961,
+      "step": 3691
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.775898176428726e-05,
+      "loss": 1.0332,
+      "step": 3692
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.775754856459782e-05,
+      "loss": 0.9863,
+      "step": 3693
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.77561149339849e-05,
+      "loss": 1.0957,
+      "step": 3694
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.775468087248329e-05,
+      "loss": 0.9795,
+      "step": 3695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7753246380127796e-05,
+      "loss": 0.873,
+      "step": 3696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.775181145695322e-05,
+      "loss": 1.0244,
+      "step": 3697
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.775037610299438e-05,
+      "loss": 0.9092,
+      "step": 3698
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.774894031828612e-05,
+      "loss": 1.1406,
+      "step": 3699
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.774750410286327e-05,
+      "loss": 0.9033,
+      "step": 3700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.77460674567607e-05,
+      "loss": 1.1309,
+      "step": 3701
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.774463038001325e-05,
+      "loss": 0.9072,
+      "step": 3702
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7743192872655815e-05,
+      "loss": 0.9678,
+      "step": 3703
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7741754934723266e-05,
+      "loss": 0.9756,
+      "step": 3704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.77403165662505e-05,
+      "loss": 0.6831,
+      "step": 3705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7738877767272424e-05,
+      "loss": 0.9873,
+      "step": 3706
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7737438537823956e-05,
+      "loss": 0.915,
+      "step": 3707
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7735998877940014e-05,
+      "loss": 1.0859,
+      "step": 3708
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.773455878765554e-05,
+      "loss": 0.876,
+      "step": 3709
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.773311826700548e-05,
+      "loss": 0.9775,
+      "step": 3710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7731677316024795e-05,
+      "loss": 1.0898,
+      "step": 3711
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.773023593474845e-05,
+      "loss": 0.9492,
+      "step": 3712
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.772879412321142e-05,
+      "loss": 0.877,
+      "step": 3713
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.772735188144869e-05,
+      "loss": 0.8477,
+      "step": 3714
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.772590920949526e-05,
+      "loss": 1.1191,
+      "step": 3715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.772446610738615e-05,
+      "loss": 0.8984,
+      "step": 3716
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7723022575156376e-05,
+      "loss": 1.0156,
+      "step": 3717
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7721578612840956e-05,
+      "loss": 1.2168,
+      "step": 3718
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7720134220474956e-05,
+      "loss": 0.8296,
+      "step": 3719
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.77186893980934e-05,
+      "loss": 1.04,
+      "step": 3720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.771724414573136e-05,
+      "loss": 0.9854,
+      "step": 3721
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.771579846342391e-05,
+      "loss": 0.9482,
+      "step": 3722
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.771435235120614e-05,
+      "loss": 1.2168,
+      "step": 3723
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.771290580911313e-05,
+      "loss": 0.9756,
+      "step": 3724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.771145883717999e-05,
+      "loss": 0.9268,
+      "step": 3725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.771001143544182e-05,
+      "loss": 0.9619,
+      "step": 3726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.770856360393377e-05,
+      "loss": 1.0137,
+      "step": 3727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7707115342690955e-05,
+      "loss": 1.1875,
+      "step": 3728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.770566665174852e-05,
+      "loss": 1.0137,
+      "step": 3729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.770421753114163e-05,
+      "loss": 0.9307,
+      "step": 3730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.770276798090545e-05,
+      "loss": 1.2002,
+      "step": 3731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.770131800107515e-05,
+      "loss": 0.7954,
+      "step": 3732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.769986759168592e-05,
+      "loss": 1.019,
+      "step": 3733
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7698416752772954e-05,
+      "loss": 1.0713,
+      "step": 3734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7696965484371474e-05,
+      "loss": 0.9668,
+      "step": 3735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.769551378651667e-05,
+      "loss": 0.8516,
+      "step": 3736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.76940616592438e-05,
+      "loss": 0.9482,
+      "step": 3737
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.769260910258808e-05,
+      "loss": 1.1289,
+      "step": 3738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7691156116584784e-05,
+      "loss": 1.0303,
+      "step": 3739
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7689702701269146e-05,
+      "loss": 0.8506,
+      "step": 3740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7688248856676445e-05,
+      "loss": 0.96,
+      "step": 3741
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7686794582841965e-05,
+      "loss": 1.1416,
+      "step": 3742
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7685339879801e-05,
+      "loss": 1.1162,
+      "step": 3743
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.768388474758884e-05,
+      "loss": 0.9678,
+      "step": 3744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.768242918624081e-05,
+      "loss": 1.082,
+      "step": 3745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.768097319579223e-05,
+      "loss": 1.0781,
+      "step": 3746
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7679516776278416e-05,
+      "loss": 0.9609,
+      "step": 3747
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7678059927734725e-05,
+      "loss": 1.0186,
+      "step": 3748
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.767660265019652e-05,
+      "loss": 0.9521,
+      "step": 3749
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.767514494369915e-05,
+      "loss": 1.0869,
+      "step": 3750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7673686808277985e-05,
+      "loss": 1.0684,
+      "step": 3751
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.767222824396842e-05,
+      "loss": 0.9473,
+      "step": 3752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.767076925080585e-05,
+      "loss": 0.9941,
+      "step": 3753
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.766930982882568e-05,
+      "loss": 1.0225,
+      "step": 3754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7667849978063326e-05,
+      "loss": 1.002,
+      "step": 3755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.766638969855421e-05,
+      "loss": 0.9131,
+      "step": 3756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.766492899033377e-05,
+      "loss": 1.085,
+      "step": 3757
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.766346785343745e-05,
+      "loss": 1.0518,
+      "step": 3758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7662006287900715e-05,
+      "loss": 1.0469,
+      "step": 3759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7660544293759026e-05,
+      "loss": 1.0146,
+      "step": 3760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.765908187104788e-05,
+      "loss": 1.1406,
+      "step": 3761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.765761901980273e-05,
+      "loss": 0.9873,
+      "step": 3762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.765615574005911e-05,
+      "loss": 1.0732,
+      "step": 3763
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7654692031852507e-05,
+      "loss": 1.1484,
+      "step": 3764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.765322789521845e-05,
+      "loss": 1.0244,
+      "step": 3765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7651763330192474e-05,
+      "loss": 0.9609,
+      "step": 3766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.765029833681011e-05,
+      "loss": 1.0381,
+      "step": 3767
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.764883291510691e-05,
+      "loss": 0.9531,
+      "step": 3768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.764736706511844e-05,
+      "loss": 0.9434,
+      "step": 3769
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.764590078688027e-05,
+      "loss": 0.958,
+      "step": 3770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.764443408042799e-05,
+      "loss": 0.9326,
+      "step": 3771
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.764296694579718e-05,
+      "loss": 1.1465,
+      "step": 3772
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7641499383023446e-05,
+      "loss": 0.959,
+      "step": 3773
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7640031392142404e-05,
+      "loss": 1.0342,
+      "step": 3774
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.763856297318968e-05,
+      "loss": 0.8945,
+      "step": 3775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7637094126200905e-05,
+      "loss": 0.8916,
+      "step": 3776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.763562485121173e-05,
+      "loss": 1.0352,
+      "step": 3777
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.76341551482578e-05,
+      "loss": 1.0957,
+      "step": 3778
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7632685017374776e-05,
+      "loss": 1.0527,
+      "step": 3779
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7631214458598354e-05,
+      "loss": 0.9863,
+      "step": 3780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7629743471964204e-05,
+      "loss": 1.1377,
+      "step": 3781
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.762827205750803e-05,
+      "loss": 1.0703,
+      "step": 3782
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7626800215265526e-05,
+      "loss": 1.042,
+      "step": 3783
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.762532794527244e-05,
+      "loss": 0.9932,
+      "step": 3784
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7623855247564455e-05,
+      "loss": 1.1162,
+      "step": 3785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7622382122177345e-05,
+      "loss": 1.0049,
+      "step": 3786
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.762090856914685e-05,
+      "loss": 1.0264,
+      "step": 3787
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.761943458850872e-05,
+      "loss": 0.9492,
+      "step": 3788
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.761796018029874e-05,
+      "loss": 0.7725,
+      "step": 3789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.761648534455266e-05,
+      "loss": 1.0234,
+      "step": 3790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.76150100813063e-05,
+      "loss": 0.7998,
+      "step": 3791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.761353439059545e-05,
+      "loss": 0.9395,
+      "step": 3792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.761205827245592e-05,
+      "loss": 1.0654,
+      "step": 3793
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.761058172692352e-05,
+      "loss": 1.0322,
+      "step": 3794
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.76091047540341e-05,
+      "loss": 1.0137,
+      "step": 3795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.76076273538235e-05,
+      "loss": 1.1797,
+      "step": 3796
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.760614952632756e-05,
+      "loss": 1.0898,
+      "step": 3797
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.760467127158215e-05,
+      "loss": 0.9648,
+      "step": 3798
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.760319258962314e-05,
+      "loss": 1.0625,
+      "step": 3799
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.760171348048642e-05,
+      "loss": 0.9297,
+      "step": 3800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.760023394420787e-05,
+      "loss": 0.9609,
+      "step": 3801
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7598753980823415e-05,
+      "loss": 1.1807,
+      "step": 3802
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7597273590368947e-05,
+      "loss": 0.9541,
+      "step": 3803
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7595792772880405e-05,
+      "loss": 1.1162,
+      "step": 3804
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.759431152839372e-05,
+      "loss": 0.9102,
+      "step": 3805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7592829856944835e-05,
+      "loss": 1.126,
+      "step": 3806
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.75913477585697e-05,
+      "loss": 1.0293,
+      "step": 3807
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7589865233304304e-05,
+      "loss": 0.9932,
+      "step": 3808
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7588382281184606e-05,
+      "loss": 0.8135,
+      "step": 3809
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.758689890224659e-05,
+      "loss": 1.0283,
+      "step": 3810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.758541509652626e-05,
+      "loss": 0.9023,
+      "step": 3811
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7583930864059624e-05,
+      "loss": 1.0537,
+      "step": 3812
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7582446204882695e-05,
+      "loss": 1.0615,
+      "step": 3813
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.758096111903151e-05,
+      "loss": 0.9727,
+      "step": 3814
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.75794756065421e-05,
+      "loss": 1.0957,
+      "step": 3815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.757798966745052e-05,
+      "loss": 1.0928,
+      "step": 3816
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7576503301792815e-05,
+      "loss": 0.8477,
+      "step": 3817
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7575016509605076e-05,
+      "loss": 1.0938,
+      "step": 3818
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.757352929092337e-05,
+      "loss": 0.9414,
+      "step": 3819
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7572041645783785e-05,
+      "loss": 1.1045,
+      "step": 3820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.757055357422243e-05,
+      "loss": 1.0303,
+      "step": 3821
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.756906507627542e-05,
+      "loss": 0.9268,
+      "step": 3822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.756757615197886e-05,
+      "loss": 1.0112,
+      "step": 3823
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7566086801368894e-05,
+      "loss": 1.0508,
+      "step": 3824
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.756459702448166e-05,
+      "loss": 0.832,
+      "step": 3825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7563106821353317e-05,
+      "loss": 1.0371,
+      "step": 3826
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.756161619202002e-05,
+      "loss": 1.0391,
+      "step": 3827
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.756012513651794e-05,
+      "loss": 0.9375,
+      "step": 3828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.755863365488327e-05,
+      "loss": 1.0645,
+      "step": 3829
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.75571417471522e-05,
+      "loss": 1.1797,
+      "step": 3830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.755564941336093e-05,
+      "loss": 0.9854,
+      "step": 3831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7554156653545685e-05,
+      "loss": 1.1279,
+      "step": 3832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.755266346774268e-05,
+      "loss": 1.0195,
+      "step": 3833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.755116985598815e-05,
+      "loss": 0.9443,
+      "step": 3834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.754967581831834e-05,
+      "loss": 1.0918,
+      "step": 3835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.754818135476952e-05,
+      "loss": 1.0557,
+      "step": 3836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.754668646537794e-05,
+      "loss": 1.0264,
+      "step": 3837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7545191150179886e-05,
+      "loss": 0.9863,
+      "step": 3838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7543695409211636e-05,
+      "loss": 0.8408,
+      "step": 3839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7542199242509495e-05,
+      "loss": 1.041,
+      "step": 3840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7540702650109764e-05,
+      "loss": 1.0391,
+      "step": 3841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.753920563204878e-05,
+      "loss": 0.9854,
+      "step": 3842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.753770818836284e-05,
+      "loss": 1.042,
+      "step": 3843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7536210319088305e-05,
+      "loss": 0.9229,
+      "step": 3844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7534712024261515e-05,
+      "loss": 0.9873,
+      "step": 3845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7533213303918834e-05,
+      "loss": 0.9932,
+      "step": 3846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.753171415809662e-05,
+      "loss": 1.0537,
+      "step": 3847
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.753021458683128e-05,
+      "loss": 0.9102,
+      "step": 3848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.752871459015918e-05,
+      "loss": 0.8926,
+      "step": 3849
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.752721416811672e-05,
+      "loss": 0.9482,
+      "step": 3850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.752571332074033e-05,
+      "loss": 1.0527,
+      "step": 3851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.752421204806641e-05,
+      "loss": 0.8936,
+      "step": 3852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.75227103501314e-05,
+      "loss": 0.998,
+      "step": 3853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7521208226971746e-05,
+      "loss": 0.8789,
+      "step": 3854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.751970567862389e-05,
+      "loss": 0.9873,
+      "step": 3855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.751820270512431e-05,
+      "loss": 0.6938,
+      "step": 3856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.751669930650947e-05,
+      "loss": 0.998,
+      "step": 3857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7515195482815843e-05,
+      "loss": 0.8672,
+      "step": 3858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.751369123407994e-05,
+      "loss": 0.8486,
+      "step": 3859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.751218656033825e-05,
+      "loss": 1.0273,
+      "step": 3860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.75106814616273e-05,
+      "loss": 0.9072,
+      "step": 3861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7509175937983606e-05,
+      "loss": 1.0322,
+      "step": 3862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7507669989443704e-05,
+      "loss": 0.9209,
+      "step": 3863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.750616361604413e-05,
+      "loss": 1.0029,
+      "step": 3864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7504656817821464e-05,
+      "loss": 0.9307,
+      "step": 3865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.750314959481225e-05,
+      "loss": 1.0684,
+      "step": 3866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.750164194705307e-05,
+      "loss": 1.0693,
+      "step": 3867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.750013387458051e-05,
+      "loss": 1.0615,
+      "step": 3868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7498625377431175e-05,
+      "loss": 0.9736,
+      "step": 3869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7497116455641656e-05,
+      "loss": 0.9902,
+      "step": 3870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7495607109248575e-05,
+      "loss": 0.8311,
+      "step": 3871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.749409733828856e-05,
+      "loss": 0.957,
+      "step": 3872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.749258714279826e-05,
+      "loss": 1.0547,
+      "step": 3873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7491076522814314e-05,
+      "loss": 0.9277,
+      "step": 3874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7489565478373375e-05,
+      "loss": 1.0186,
+      "step": 3875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.748805400951212e-05,
+      "loss": 0.9922,
+      "step": 3876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.748654211626722e-05,
+      "loss": 0.7949,
+      "step": 3877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.748502979867538e-05,
+      "loss": 0.9229,
+      "step": 3878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.748351705677328e-05,
+      "loss": 1.0176,
+      "step": 3879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.748200389059763e-05,
+      "loss": 1.0654,
+      "step": 3880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.748049030018517e-05,
+      "loss": 1.0742,
+      "step": 3881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.747897628557262e-05,
+      "loss": 1.0107,
+      "step": 3882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7477461846796715e-05,
+      "loss": 1.1035,
+      "step": 3883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7475946983894204e-05,
+      "loss": 1.0781,
+      "step": 3884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7474431696901866e-05,
+      "loss": 0.9824,
+      "step": 3885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7472915985856456e-05,
+      "loss": 0.75,
+      "step": 3886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.747139985079476e-05,
+      "loss": 0.8848,
+      "step": 3887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.746988329175357e-05,
+      "loss": 1.0557,
+      "step": 3888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7468366308769695e-05,
+      "loss": 0.9893,
+      "step": 3889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.746684890187994e-05,
+      "loss": 1.0654,
+      "step": 3890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.746533107112113e-05,
+      "loss": 0.7383,
+      "step": 3891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.74638128165301e-05,
+      "loss": 0.9678,
+      "step": 3892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7462294138143687e-05,
+      "loss": 1.0703,
+      "step": 3893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7460775035998755e-05,
+      "loss": 1.1299,
+      "step": 3894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7459255510132165e-05,
+      "loss": 0.9873,
+      "step": 3895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7457735560580785e-05,
+      "loss": 1.0615,
+      "step": 3896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7456215187381506e-05,
+      "loss": 0.9395,
+      "step": 3897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.745469439057122e-05,
+      "loss": 1.124,
+      "step": 3898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.745317317018684e-05,
+      "loss": 1.1523,
+      "step": 3899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.745165152626526e-05,
+      "loss": 1.0273,
+      "step": 3900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7450129458843435e-05,
+      "loss": 0.9951,
+      "step": 3901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.744860696795828e-05,
+      "loss": 0.9131,
+      "step": 3902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.744708405364675e-05,
+      "loss": 1.0918,
+      "step": 3903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7445560715945806e-05,
+      "loss": 1.002,
+      "step": 3904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.74440369548924e-05,
+      "loss": 0.8701,
+      "step": 3905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.744251277052352e-05,
+      "loss": 1.0254,
+      "step": 3906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.744098816287616e-05,
+      "loss": 0.9541,
+      "step": 3907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7439463131987295e-05,
+      "loss": 0.8867,
+      "step": 3908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.743793767789396e-05,
+      "loss": 1.0723,
+      "step": 3909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7436411800633156e-05,
+      "loss": 1.042,
+      "step": 3910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.743488550024191e-05,
+      "loss": 0.9424,
+      "step": 3911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.743335877675728e-05,
+      "loss": 1.0645,
+      "step": 3912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.74318316302163e-05,
+      "loss": 0.9043,
+      "step": 3913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.743030406065602e-05,
+      "loss": 1.0898,
+      "step": 3914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.742877606811353e-05,
+      "loss": 0.9922,
+      "step": 3915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.74272476526259e-05,
+      "loss": 1.0371,
+      "step": 3916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.742571881423022e-05,
+      "loss": 0.9551,
+      "step": 3917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7424189552963596e-05,
+      "loss": 1.0732,
+      "step": 3918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7422659868863125e-05,
+      "loss": 0.9238,
+      "step": 3919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.742112976196595e-05,
+      "loss": 0.9951,
+      "step": 3920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7419599232309174e-05,
+      "loss": 0.9922,
+      "step": 3921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.741806827992996e-05,
+      "loss": 1.1719,
+      "step": 3922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.741653690486546e-05,
+      "loss": 0.9473,
+      "step": 3923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.741500510715282e-05,
+      "loss": 1.1035,
+      "step": 3924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.741347288682923e-05,
+      "loss": 1.0322,
+      "step": 3925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.741194024393186e-05,
+      "loss": 0.9785,
+      "step": 3926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7410407178497905e-05,
+      "loss": 0.8818,
+      "step": 3927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7408873690564576e-05,
+      "loss": 1.0205,
+      "step": 3928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.740733978016907e-05,
+      "loss": 0.751,
+      "step": 3929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7405805447348634e-05,
+      "loss": 0.9873,
+      "step": 3930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.740427069214048e-05,
+      "loss": 1.1064,
+      "step": 3931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.740273551458186e-05,
+      "loss": 0.958,
+      "step": 3932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7401199914710034e-05,
+      "loss": 1.0,
+      "step": 3933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7399663892562255e-05,
+      "loss": 1.0605,
+      "step": 3934
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.739812744817581e-05,
+      "loss": 1.0898,
+      "step": 3935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.739659058158798e-05,
+      "loss": 1.0488,
+      "step": 3936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.739505329283604e-05,
+      "loss": 0.9814,
+      "step": 3937
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7393515581957335e-05,
+      "loss": 0.9409,
+      "step": 3938
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7391977448989154e-05,
+      "loss": 0.8774,
+      "step": 3939
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.739043889396882e-05,
+      "loss": 0.9922,
+      "step": 3940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.738889991693368e-05,
+      "loss": 1.0469,
+      "step": 3941
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7387360517921085e-05,
+      "loss": 1.1914,
+      "step": 3942
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.738582069696838e-05,
+      "loss": 0.9287,
+      "step": 3943
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7384280454112936e-05,
+      "loss": 0.7764,
+      "step": 3944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.738273978939213e-05,
+      "loss": 1.0762,
+      "step": 3945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.738119870284335e-05,
+      "loss": 0.998,
+      "step": 3946
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7379657194504e-05,
+      "loss": 1.0557,
+      "step": 3947
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.737811526441147e-05,
+      "loss": 1.0908,
+      "step": 3948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7376572912603205e-05,
+      "loss": 1.0605,
+      "step": 3949
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7375030139116606e-05,
+      "loss": 1.1865,
+      "step": 3950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.737348694398913e-05,
+      "loss": 0.9795,
+      "step": 3951
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.737194332725822e-05,
+      "loss": 0.7969,
+      "step": 3952
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.737039928896133e-05,
+      "loss": 0.9727,
+      "step": 3953
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.736885482913593e-05,
+      "loss": 0.9092,
+      "step": 3954
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.736730994781951e-05,
+      "loss": 1.1123,
+      "step": 3955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7365764645049554e-05,
+      "loss": 0.9375,
+      "step": 3956
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.736421892086355e-05,
+      "loss": 0.874,
+      "step": 3957
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.736267277529903e-05,
+      "loss": 0.9072,
+      "step": 3958
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7361126208393505e-05,
+      "loss": 0.8936,
+      "step": 3959
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.73595792201845e-05,
+      "loss": 0.8711,
+      "step": 3960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7358031810709554e-05,
+      "loss": 0.9629,
+      "step": 3961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.735648398000623e-05,
+      "loss": 0.8789,
+      "step": 3962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.735493572811209e-05,
+      "loss": 1.0547,
+      "step": 3963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.735338705506469e-05,
+      "loss": 0.8027,
+      "step": 3964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7351837960901624e-05,
+      "loss": 1.0137,
+      "step": 3965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7350288445660484e-05,
+      "loss": 1.1484,
+      "step": 3966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.734873850937886e-05,
+      "loss": 0.9619,
+      "step": 3967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.734718815209438e-05,
+      "loss": 1.0479,
+      "step": 3968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.734563737384466e-05,
+      "loss": 0.8281,
+      "step": 3969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7344086174667324e-05,
+      "loss": 1.2285,
+      "step": 3970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.734253455460003e-05,
+      "loss": 0.9805,
+      "step": 3971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7340982513680425e-05,
+      "loss": 0.9824,
+      "step": 3972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.733943005194618e-05,
+      "loss": 1.0381,
+      "step": 3973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7337877169434946e-05,
+      "loss": 0.9854,
+      "step": 3974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7336323866184437e-05,
+      "loss": 1.0635,
+      "step": 3975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7334770142232314e-05,
+      "loss": 1.0088,
+      "step": 3976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7333215997616314e-05,
+      "loss": 1.0195,
+      "step": 3977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.733166143237413e-05,
+      "loss": 0.9502,
+      "step": 3978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.73301064465435e-05,
+      "loss": 1.0479,
+      "step": 3979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.732855104016214e-05,
+      "loss": 0.9775,
+      "step": 3980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.732699521326782e-05,
+      "loss": 0.8506,
+      "step": 3981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7325438965898284e-05,
+      "loss": 0.9707,
+      "step": 3982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.732388229809129e-05,
+      "loss": 1.2793,
+      "step": 3983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.732232520988462e-05,
+      "loss": 0.9512,
+      "step": 3984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.732076770131607e-05,
+      "loss": 0.8701,
+      "step": 3985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.731920977242341e-05,
+      "loss": 1.1797,
+      "step": 3986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.731765142324448e-05,
+      "loss": 1.0713,
+      "step": 3987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7316092653817066e-05,
+      "loss": 1.1084,
+      "step": 3988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7314533464179016e-05,
+      "loss": 1.0479,
+      "step": 3989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.731297385436816e-05,
+      "loss": 1.1865,
+      "step": 3990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.731141382442234e-05,
+      "loss": 0.999,
+      "step": 3991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.730985337437941e-05,
+      "loss": 0.8496,
+      "step": 3992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.730829250427726e-05,
+      "loss": 0.7891,
+      "step": 3993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7306731214153746e-05,
+      "loss": 0.8818,
+      "step": 3994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.730516950404676e-05,
+      "loss": 0.9961,
+      "step": 3995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.73036073739942e-05,
+      "loss": 0.9756,
+      "step": 3996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.730204482403399e-05,
+      "loss": 0.8955,
+      "step": 3997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7300481854204024e-05,
+      "loss": 1.0547,
+      "step": 3998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.729891846454225e-05,
+      "loss": 0.9355,
+      "step": 3999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7297354655086595e-05,
+      "loss": 1.0098,
+      "step": 4000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.729579042587501e-05,
+      "loss": 1.1738,
+      "step": 4001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7294225776945465e-05,
+      "loss": 1.0352,
+      "step": 4002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.729266070833591e-05,
+      "loss": 1.1973,
+      "step": 4003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7291095220084346e-05,
+      "loss": 1.041,
+      "step": 4004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.728952931222874e-05,
+      "loss": 1.1152,
+      "step": 4005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7287962984807114e-05,
+      "loss": 0.9873,
+      "step": 4006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.728639623785746e-05,
+      "loss": 0.9697,
+      "step": 4007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.728482907141782e-05,
+      "loss": 1.1309,
+      "step": 4008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.72832614855262e-05,
+      "loss": 0.8682,
+      "step": 4009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7281693480220654e-05,
+      "loss": 0.9414,
+      "step": 4010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.728012505553923e-05,
+      "loss": 1.0557,
+      "step": 4011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7278556211519996e-05,
+      "loss": 0.917,
+      "step": 4012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7276986948201015e-05,
+      "loss": 1.0293,
+      "step": 4013
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.727541726562037e-05,
+      "loss": 0.7751,
+      "step": 4014
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.727384716381615e-05,
+      "loss": 0.9395,
+      "step": 4015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.727227664282646e-05,
+      "loss": 1.0527,
+      "step": 4016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7270705702689415e-05,
+      "loss": 1.1201,
+      "step": 4017
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7269134343443134e-05,
+      "loss": 1.0703,
+      "step": 4018
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.726756256512575e-05,
+      "loss": 1.1465,
+      "step": 4019
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7265990367775395e-05,
+      "loss": 0.9111,
+      "step": 4020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7264417751430244e-05,
+      "loss": 1.0293,
+      "step": 4021
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7262844716128436e-05,
+      "loss": 1.0039,
+      "step": 4022
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.726127126190817e-05,
+      "loss": 0.876,
+      "step": 4023
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.72596973888076e-05,
+      "loss": 0.8066,
+      "step": 4024
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.725812309686494e-05,
+      "loss": 0.9268,
+      "step": 4025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7256548386118386e-05,
+      "loss": 1.0215,
+      "step": 4026
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.725497325660615e-05,
+      "loss": 0.8887,
+      "step": 4027
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.725339770836646e-05,
+      "loss": 0.9053,
+      "step": 4028
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.725182174143755e-05,
+      "loss": 0.9307,
+      "step": 4029
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.725024535585766e-05,
+      "loss": 1.0703,
+      "step": 4030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7248668551665044e-05,
+      "loss": 0.9443,
+      "step": 4031
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7247091328897974e-05,
+      "loss": 0.8672,
+      "step": 4032
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.724551368759471e-05,
+      "loss": 0.9365,
+      "step": 4033
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.724393562779355e-05,
+      "loss": 1.04,
+      "step": 4034
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7242357149532794e-05,
+      "loss": 0.8779,
+      "step": 4035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.724077825285073e-05,
+      "loss": 1.0625,
+      "step": 4036
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.723919893778568e-05,
+      "loss": 1.0801,
+      "step": 4037
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7237619204375975e-05,
+      "loss": 1.084,
+      "step": 4038
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.723603905265994e-05,
+      "loss": 1.1152,
+      "step": 4039
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7234458482675934e-05,
+      "loss": 0.9971,
+      "step": 4040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.72328774944623e-05,
+      "loss": 0.9902,
+      "step": 4041
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.72312960880574e-05,
+      "loss": 0.9375,
+      "step": 4042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7229714263499636e-05,
+      "loss": 1.252,
+      "step": 4043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.722813202082737e-05,
+      "loss": 0.9258,
+      "step": 4044
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.722654936007901e-05,
+      "loss": 0.9473,
+      "step": 4045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.722496628129296e-05,
+      "loss": 1.1113,
+      "step": 4046
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.722338278450763e-05,
+      "loss": 1.1104,
+      "step": 4047
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.722179886976145e-05,
+      "loss": 1.043,
+      "step": 4048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.722021453709286e-05,
+      "loss": 0.8457,
+      "step": 4049
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.721862978654031e-05,
+      "loss": 0.9287,
+      "step": 4050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7217044618142256e-05,
+      "loss": 0.9375,
+      "step": 4051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.721545903193716e-05,
+      "loss": 1.0566,
+      "step": 4052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.721387302796349e-05,
+      "loss": 0.8975,
+      "step": 4053
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.721228660625976e-05,
+      "loss": 0.9141,
+      "step": 4054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.721069976686445e-05,
+      "loss": 0.875,
+      "step": 4055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.720911250981607e-05,
+      "loss": 1.0439,
+      "step": 4056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.720752483515314e-05,
+      "loss": 1.0791,
+      "step": 4057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.720593674291419e-05,
+      "loss": 1.0967,
+      "step": 4058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.720434823313775e-05,
+      "loss": 0.9961,
+      "step": 4059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.720275930586238e-05,
+      "loss": 1.0264,
+      "step": 4060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.720116996112663e-05,
+      "loss": 0.9023,
+      "step": 4061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.719958019896908e-05,
+      "loss": 1.0488,
+      "step": 4062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.719799001942829e-05,
+      "loss": 1.0498,
+      "step": 4063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7196399422542866e-05,
+      "loss": 1.0869,
+      "step": 4064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.71948084083514e-05,
+      "loss": 0.9521,
+      "step": 4065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.71932169768925e-05,
+      "loss": 1.0107,
+      "step": 4066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7191625128204786e-05,
+      "loss": 0.9326,
+      "step": 4067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7190032862326895e-05,
+      "loss": 0.8496,
+      "step": 4068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7188440179297455e-05,
+      "loss": 1.0205,
+      "step": 4069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.718684707915513e-05,
+      "loss": 0.834,
+      "step": 4070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.718525356193856e-05,
+      "loss": 0.9053,
+      "step": 4071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.718365962768643e-05,
+      "loss": 0.9746,
+      "step": 4072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.718206527643742e-05,
+      "loss": 0.8965,
+      "step": 4073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.718047050823021e-05,
+      "loss": 0.9922,
+      "step": 4074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.717887532310351e-05,
+      "loss": 0.8975,
+      "step": 4075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7177279721096024e-05,
+      "loss": 1.0264,
+      "step": 4076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.717568370224648e-05,
+      "loss": 0.9863,
+      "step": 4077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.71740872665936e-05,
+      "loss": 1.1523,
+      "step": 4078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7172490414176136e-05,
+      "loss": 1.0312,
+      "step": 4079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.717089314503283e-05,
+      "loss": 0.9404,
+      "step": 4080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.716929545920244e-05,
+      "loss": 0.9561,
+      "step": 4081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.716769735672374e-05,
+      "loss": 1.0557,
+      "step": 4082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.716609883763553e-05,
+      "loss": 0.8652,
+      "step": 4083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.716449990197657e-05,
+      "loss": 0.9795,
+      "step": 4084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7162900549785684e-05,
+      "loss": 1.2754,
+      "step": 4085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.716130078110167e-05,
+      "loss": 1.0879,
+      "step": 4086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.715970059596336e-05,
+      "loss": 1.0771,
+      "step": 4087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.715809999440958e-05,
+      "loss": 1.0156,
+      "step": 4088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.715649897647917e-05,
+      "loss": 1.0547,
+      "step": 4089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.715489754221099e-05,
+      "loss": 1.0586,
+      "step": 4090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7153295691643894e-05,
+      "loss": 1.2285,
+      "step": 4091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.715169342481676e-05,
+      "loss": 0.9082,
+      "step": 4092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7150090741768464e-05,
+      "loss": 0.7358,
+      "step": 4093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7148487642537906e-05,
+      "loss": 0.8965,
+      "step": 4094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.714688412716398e-05,
+      "loss": 1.0908,
+      "step": 4095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.714528019568561e-05,
+      "loss": 1.0898,
+      "step": 4096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7143675848141706e-05,
+      "loss": 0.9922,
+      "step": 4097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.714207108457122e-05,
+      "loss": 0.9287,
+      "step": 4098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.714046590501306e-05,
+      "loss": 1.0977,
+      "step": 4099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.713886030950622e-05,
+      "loss": 0.9893,
+      "step": 4100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.713725429808964e-05,
+      "loss": 1.0566,
+      "step": 4101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.713564787080229e-05,
+      "loss": 0.9541,
+      "step": 4102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7134041027683164e-05,
+      "loss": 1.0205,
+      "step": 4103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.713243376877125e-05,
+      "loss": 0.9551,
+      "step": 4104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7130826094105555e-05,
+      "loss": 1.0977,
+      "step": 4105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7129218003725094e-05,
+      "loss": 0.9717,
+      "step": 4106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.712760949766888e-05,
+      "loss": 0.9551,
+      "step": 4107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.712600057597596e-05,
+      "loss": 1.123,
+      "step": 4108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.712439123868537e-05,
+      "loss": 0.9336,
+      "step": 4109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.712278148583616e-05,
+      "loss": 0.9316,
+      "step": 4110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7121171317467414e-05,
+      "loss": 0.873,
+      "step": 4111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.711956073361818e-05,
+      "loss": 1.0,
+      "step": 4112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7117949734327556e-05,
+      "loss": 0.9648,
+      "step": 4113
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.711633831963463e-05,
+      "loss": 1.0557,
+      "step": 4114
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.711472648957852e-05,
+      "loss": 0.9121,
+      "step": 4115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7113114244198325e-05,
+      "loss": 0.7207,
+      "step": 4116
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7111501583533176e-05,
+      "loss": 0.8633,
+      "step": 4117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7109888507622206e-05,
+      "loss": 0.9707,
+      "step": 4118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7108275016504556e-05,
+      "loss": 1.0039,
+      "step": 4119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.710666111021939e-05,
+      "loss": 1.1211,
+      "step": 4120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.710504678880587e-05,
+      "loss": 1.0342,
+      "step": 4121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.710343205230317e-05,
+      "loss": 1.0068,
+      "step": 4122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7101816900750464e-05,
+      "loss": 1.0078,
+      "step": 4123
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.710020133418696e-05,
+      "loss": 1.0361,
+      "step": 4124
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.709858535265187e-05,
+      "loss": 1.0918,
+      "step": 4125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7096968956184384e-05,
+      "loss": 1.0,
+      "step": 4126
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.709535214482375e-05,
+      "loss": 0.9434,
+      "step": 4127
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.709373491860919e-05,
+      "loss": 1.0947,
+      "step": 4128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7092117277579964e-05,
+      "loss": 0.9824,
+      "step": 4129
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.709049922177531e-05,
+      "loss": 1.0811,
+      "step": 4130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.70888807512345e-05,
+      "loss": 1.2188,
+      "step": 4131
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7087261865996815e-05,
+      "loss": 0.9023,
+      "step": 4132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.708564256610153e-05,
+      "loss": 1.0195,
+      "step": 4133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.708402285158795e-05,
+      "loss": 0.9609,
+      "step": 4134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.708240272249538e-05,
+      "loss": 0.957,
+      "step": 4135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.708078217886313e-05,
+      "loss": 1.1133,
+      "step": 4136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7079161220730534e-05,
+      "loss": 1.1709,
+      "step": 4137
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.707753984813692e-05,
+      "loss": 0.8643,
+      "step": 4138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.707591806112164e-05,
+      "loss": 0.8984,
+      "step": 4139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.707429585972405e-05,
+      "loss": 0.8555,
+      "step": 4140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.70726732439835e-05,
+      "loss": 0.7109,
+      "step": 4141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.707105021393939e-05,
+      "loss": 1.1328,
+      "step": 4142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7069426769631096e-05,
+      "loss": 0.9414,
+      "step": 4143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.706780291109801e-05,
+      "loss": 0.9209,
+      "step": 4144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.706617863837954e-05,
+      "loss": 0.8184,
+      "step": 4145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.706455395151511e-05,
+      "loss": 1.043,
+      "step": 4146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.706292885054413e-05,
+      "loss": 1.0762,
+      "step": 4147
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.706130333550606e-05,
+      "loss": 1.0049,
+      "step": 4148
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.705967740644033e-05,
+      "loss": 1.1973,
+      "step": 4149
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.70580510633864e-05,
+      "loss": 0.8896,
+      "step": 4150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7056424306383735e-05,
+      "loss": 1.1016,
+      "step": 4151
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.705479713547181e-05,
+      "loss": 1.0137,
+      "step": 4152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.705316955069012e-05,
+      "loss": 0.9189,
+      "step": 4153
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.705154155207815e-05,
+      "loss": 0.9727,
+      "step": 4154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.704991313967542e-05,
+      "loss": 0.918,
+      "step": 4155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.704828431352144e-05,
+      "loss": 1.0039,
+      "step": 4156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.704665507365573e-05,
+      "loss": 1.1094,
+      "step": 4157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.704502542011784e-05,
+      "loss": 0.9463,
+      "step": 4158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.704339535294732e-05,
+      "loss": 1.084,
+      "step": 4159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.70417648721837e-05,
+      "loss": 0.873,
+      "step": 4160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.704013397786657e-05,
+      "loss": 0.835,
+      "step": 4161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7038502670035506e-05,
+      "loss": 0.873,
+      "step": 4162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.703687094873009e-05,
+      "loss": 1.1045,
+      "step": 4163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.703523881398991e-05,
+      "loss": 1.2021,
+      "step": 4164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.703360626585459e-05,
+      "loss": 1.1416,
+      "step": 4165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7031973304363746e-05,
+      "loss": 1.083,
+      "step": 4166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.703033992955699e-05,
+      "loss": 0.8105,
+      "step": 4167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7028706141473974e-05,
+      "loss": 1.0234,
+      "step": 4168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.702707194015434e-05,
+      "loss": 0.8413,
+      "step": 4169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.702543732563774e-05,
+      "loss": 0.9561,
+      "step": 4170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7023802297963844e-05,
+      "loss": 1.0342,
+      "step": 4171
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7022166857172345e-05,
+      "loss": 0.8018,
+      "step": 4172
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.702053100330291e-05,
+      "loss": 1.0732,
+      "step": 4173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7018894736395234e-05,
+      "loss": 0.9902,
+      "step": 4174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.701725805648904e-05,
+      "loss": 1.0244,
+      "step": 4175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.701562096362404e-05,
+      "loss": 0.9736,
+      "step": 4176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.701398345783996e-05,
+      "loss": 0.9561,
+      "step": 4177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.701234553917654e-05,
+      "loss": 0.8789,
+      "step": 4178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.701070720767351e-05,
+      "loss": 1.1172,
+      "step": 4179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7009068463370665e-05,
+      "loss": 1.0078,
+      "step": 4180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7007429306307734e-05,
+      "loss": 0.9023,
+      "step": 4181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.700578973652452e-05,
+      "loss": 1.1133,
+      "step": 4182
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.70041497540608e-05,
+      "loss": 0.9854,
+      "step": 4183
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7002509358956376e-05,
+      "loss": 1.0264,
+      "step": 4184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.700086855125104e-05,
+      "loss": 0.9746,
+      "step": 4185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.699922733098463e-05,
+      "loss": 0.9805,
+      "step": 4186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6997585698196965e-05,
+      "loss": 1.0176,
+      "step": 4187
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.699594365292789e-05,
+      "loss": 1.1211,
+      "step": 4188
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.699430119521724e-05,
+      "loss": 1.127,
+      "step": 4189
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.699265832510487e-05,
+      "loss": 1.0713,
+      "step": 4190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6991015042630667e-05,
+      "loss": 0.9355,
+      "step": 4191
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6989371347834496e-05,
+      "loss": 1.0596,
+      "step": 4192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6987727240756246e-05,
+      "loss": 0.9531,
+      "step": 4193
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6986082721435816e-05,
+      "loss": 0.8247,
+      "step": 4194
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.698443778991311e-05,
+      "loss": 1.0127,
+      "step": 4195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.698279244622805e-05,
+      "loss": 0.9912,
+      "step": 4196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.698114669042056e-05,
+      "loss": 1.1338,
+      "step": 4197
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.697950052253058e-05,
+      "loss": 1.001,
+      "step": 4198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.697785394259806e-05,
+      "loss": 1.0381,
+      "step": 4199
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.697620695066295e-05,
+      "loss": 1.1641,
+      "step": 4200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.697455954676523e-05,
+      "loss": 0.959,
+      "step": 4201
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.697291173094486e-05,
+      "loss": 0.9336,
+      "step": 4202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.697126350324184e-05,
+      "loss": 1.1572,
+      "step": 4203
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.696961486369617e-05,
+      "loss": 1.0898,
+      "step": 4204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6967965812347856e-05,
+      "loss": 1.0732,
+      "step": 4205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.696631634923691e-05,
+      "loss": 0.9355,
+      "step": 4206
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6964666474403356e-05,
+      "loss": 1.0781,
+      "step": 4207
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6963016187887245e-05,
+      "loss": 0.8809,
+      "step": 4208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.696136548972861e-05,
+      "loss": 0.9404,
+      "step": 4209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6959714379967524e-05,
+      "loss": 1.0859,
+      "step": 4210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6958062858644045e-05,
+      "loss": 0.8594,
+      "step": 4211
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6956410925798254e-05,
+      "loss": 1.0869,
+      "step": 4212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6954758581470235e-05,
+      "loss": 1.0654,
+      "step": 4213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.695310582570009e-05,
+      "loss": 1.0625,
+      "step": 4214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6951452658527915e-05,
+      "loss": 0.9902,
+      "step": 4215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6949799079993846e-05,
+      "loss": 0.9521,
+      "step": 4216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6948145090137996e-05,
+      "loss": 0.9668,
+      "step": 4217
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.694649068900051e-05,
+      "loss": 0.9033,
+      "step": 4218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.694483587662153e-05,
+      "loss": 1.0176,
+      "step": 4219
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6943180653041214e-05,
+      "loss": 0.8682,
+      "step": 4220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6941525018299734e-05,
+      "loss": 0.9854,
+      "step": 4221
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.693986897243727e-05,
+      "loss": 0.8701,
+      "step": 4222
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6938212515494004e-05,
+      "loss": 1.0938,
+      "step": 4223
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.693655564751012e-05,
+      "loss": 1.1318,
+      "step": 4224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6934898368525854e-05,
+      "loss": 1.0342,
+      "step": 4225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.69332406785814e-05,
+      "loss": 0.999,
+      "step": 4226
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6931582577717e-05,
+      "loss": 0.9365,
+      "step": 4227
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.692992406597288e-05,
+      "loss": 1.0264,
+      "step": 4228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6928265143389296e-05,
+      "loss": 0.8979,
+      "step": 4229
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6926605810006495e-05,
+      "loss": 1.0625,
+      "step": 4230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.692494606586476e-05,
+      "loss": 0.832,
+      "step": 4231
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.692328591100436e-05,
+      "loss": 1.1309,
+      "step": 4232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.692162534546556e-05,
+      "loss": 1.1045,
+      "step": 4233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.69199643692887e-05,
+      "loss": 1.0527,
+      "step": 4234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.691830298251406e-05,
+      "loss": 1.1113,
+      "step": 4235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.691664118518195e-05,
+      "loss": 1.0693,
+      "step": 4236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.691497897733272e-05,
+      "loss": 0.7598,
+      "step": 4237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.691331635900669e-05,
+      "loss": 0.8682,
+      "step": 4238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.691165333024421e-05,
+      "loss": 0.9941,
+      "step": 4239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.690998989108564e-05,
+      "loss": 0.9033,
+      "step": 4240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.690832604157135e-05,
+      "loss": 1.0439,
+      "step": 4241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.690666178174171e-05,
+      "loss": 0.9434,
+      "step": 4242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.690499711163711e-05,
+      "loss": 0.958,
+      "step": 4243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.690333203129795e-05,
+      "loss": 0.9609,
+      "step": 4244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6901666540764624e-05,
+      "loss": 1.0508,
+      "step": 4245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.690000064007756e-05,
+      "loss": 1.0342,
+      "step": 4246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.689833432927719e-05,
+      "loss": 0.9922,
+      "step": 4247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.689666760840394e-05,
+      "loss": 0.9072,
+      "step": 4248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.689500047749825e-05,
+      "loss": 1.0967,
+      "step": 4249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6893332936600585e-05,
+      "loss": 1.0762,
+      "step": 4250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.689166498575142e-05,
+      "loss": 1.1016,
+      "step": 4251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.688999662499122e-05,
+      "loss": 0.8809,
+      "step": 4252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6888327854360464e-05,
+      "loss": 1.0293,
+      "step": 4253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.688665867389967e-05,
+      "loss": 1.0732,
+      "step": 4254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.688498908364932e-05,
+      "loss": 0.9414,
+      "step": 4255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.688331908364995e-05,
+      "loss": 0.9795,
+      "step": 4256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.688164867394208e-05,
+      "loss": 1.0508,
+      "step": 4257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.687997785456624e-05,
+      "loss": 0.9541,
+      "step": 4258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6878306625562975e-05,
+      "loss": 0.9326,
+      "step": 4259
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.687663498697285e-05,
+      "loss": 0.9746,
+      "step": 4260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.687496293883642e-05,
+      "loss": 0.8613,
+      "step": 4261
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.687329048119428e-05,
+      "loss": 1.0684,
+      "step": 4262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6871617614086995e-05,
+      "loss": 1.0557,
+      "step": 4263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.686994433755517e-05,
+      "loss": 0.9893,
+      "step": 4264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6868270651639406e-05,
+      "loss": 1.1523,
+      "step": 4265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6866596556380326e-05,
+      "loss": 0.9824,
+      "step": 4266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.686492205181854e-05,
+      "loss": 0.8652,
+      "step": 4267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.68632471379947e-05,
+      "loss": 0.9307,
+      "step": 4268
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6861571814949454e-05,
+      "loss": 1.0283,
+      "step": 4269
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.685989608272344e-05,
+      "loss": 1.0645,
+      "step": 4270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.685821994135733e-05,
+      "loss": 0.8271,
+      "step": 4271
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6856543390891807e-05,
+      "loss": 1.0557,
+      "step": 4272
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.685486643136754e-05,
+      "loss": 0.834,
+      "step": 4273
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.685318906282524e-05,
+      "loss": 1.002,
+      "step": 4274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6851511285305594e-05,
+      "loss": 0.8896,
+      "step": 4275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.684983309884934e-05,
+      "loss": 0.8486,
+      "step": 4276
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6848154503497185e-05,
+      "loss": 1.2559,
+      "step": 4277
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.684647549928987e-05,
+      "loss": 0.9326,
+      "step": 4278
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6844796086268136e-05,
+      "loss": 1.0723,
+      "step": 4279
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.684311626447274e-05,
+      "loss": 1.1836,
+      "step": 4280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6841436033944454e-05,
+      "loss": 1.1064,
+      "step": 4281
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.683975539472403e-05,
+      "loss": 1.0312,
+      "step": 4282
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.683807434685228e-05,
+      "loss": 1.0449,
+      "step": 4283
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6836392890369975e-05,
+      "loss": 1.1514,
+      "step": 4284
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.683471102531794e-05,
+      "loss": 1.082,
+      "step": 4285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6833028751736965e-05,
+      "loss": 0.8242,
+      "step": 4286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.683134606966789e-05,
+      "loss": 1.0479,
+      "step": 4287
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.682966297915155e-05,
+      "loss": 0.9238,
+      "step": 4288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.682797948022878e-05,
+      "loss": 0.6743,
+      "step": 4289
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6826295572940434e-05,
+      "loss": 1.0391,
+      "step": 4290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6824611257327386e-05,
+      "loss": 1.1406,
+      "step": 4291
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.68229265334305e-05,
+      "loss": 1.1094,
+      "step": 4292
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.682124140129066e-05,
+      "loss": 0.8564,
+      "step": 4293
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.681955586094876e-05,
+      "loss": 1.0127,
+      "step": 4294
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.681786991244571e-05,
+      "loss": 1.0127,
+      "step": 4295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6816183555822405e-05,
+      "loss": 0.9775,
+      "step": 4296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6814496791119795e-05,
+      "loss": 1.0146,
+      "step": 4297
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.681280961837878e-05,
+      "loss": 1.0908,
+      "step": 4298
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.681112203764033e-05,
+      "loss": 0.8594,
+      "step": 4299
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.680943404894539e-05,
+      "loss": 0.7734,
+      "step": 4300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6807745652334916e-05,
+      "loss": 1.0977,
+      "step": 4301
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6806056847849886e-05,
+      "loss": 1.1562,
+      "step": 4302
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6804367635531286e-05,
+      "loss": 0.9873,
+      "step": 4303
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.68026780154201e-05,
+      "loss": 1.0215,
+      "step": 4304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.680098798755733e-05,
+      "loss": 1.083,
+      "step": 4305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.679929755198399e-05,
+      "loss": 0.876,
+      "step": 4306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.679760670874111e-05,
+      "loss": 1.1553,
+      "step": 4307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.679591545786971e-05,
+      "loss": 1.207,
+      "step": 4308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.679422379941084e-05,
+      "loss": 1.0508,
+      "step": 4309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6792531733405546e-05,
+      "loss": 0.9697,
+      "step": 4310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6790839259894896e-05,
+      "loss": 0.8569,
+      "step": 4311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6789146378919954e-05,
+      "loss": 0.9707,
+      "step": 4312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6787453090521806e-05,
+      "loss": 0.9189,
+      "step": 4313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.678575939474154e-05,
+      "loss": 0.9023,
+      "step": 4314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.678406529162026e-05,
+      "loss": 1.0234,
+      "step": 4315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.678237078119907e-05,
+      "loss": 1.0605,
+      "step": 4316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6780675863519105e-05,
+      "loss": 1.0479,
+      "step": 4317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.677898053862149e-05,
+      "loss": 0.9951,
+      "step": 4318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.677728480654735e-05,
+      "loss": 1.082,
+      "step": 4319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6775588667337856e-05,
+      "loss": 1.0322,
+      "step": 4320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.677389212103416e-05,
+      "loss": 0.9395,
+      "step": 4321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.677219516767743e-05,
+      "loss": 1.0029,
+      "step": 4322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6770497807308845e-05,
+      "loss": 0.8604,
+      "step": 4323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.67688000399696e-05,
+      "loss": 0.9668,
+      "step": 4324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.67671018657009e-05,
+      "loss": 1.0615,
+      "step": 4325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.676540328454394e-05,
+      "loss": 1.1621,
+      "step": 4326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.676370429653995e-05,
+      "loss": 1.0908,
+      "step": 4327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6762004901730156e-05,
+      "loss": 1.0439,
+      "step": 4328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6760305100155806e-05,
+      "loss": 0.8477,
+      "step": 4329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6758604891858136e-05,
+      "loss": 0.9111,
+      "step": 4330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.675690427687841e-05,
+      "loss": 1.0586,
+      "step": 4331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6755203255257903e-05,
+      "loss": 1.0254,
+      "step": 4332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.675350182703788e-05,
+      "loss": 0.9932,
+      "step": 4333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6751799992259646e-05,
+      "loss": 1.0293,
+      "step": 4334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.675009775096448e-05,
+      "loss": 1.0391,
+      "step": 4335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6748395103193714e-05,
+      "loss": 1.1133,
+      "step": 4336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.674669204898865e-05,
+      "loss": 1.0557,
+      "step": 4337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.674498858839061e-05,
+      "loss": 0.9863,
+      "step": 4338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.674328472144096e-05,
+      "loss": 1.0508,
+      "step": 4339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6741580448181015e-05,
+      "loss": 1.0771,
+      "step": 4340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.673987576865216e-05,
+      "loss": 0.7363,
+      "step": 4341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.673817068289574e-05,
+      "loss": 1.1211,
+      "step": 4342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.673646519095315e-05,
+      "loss": 1.0615,
+      "step": 4343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.673475929286577e-05,
+      "loss": 1.0386,
+      "step": 4344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6733052988674994e-05,
+      "loss": 1.0801,
+      "step": 4345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.673134627842223e-05,
+      "loss": 1.0488,
+      "step": 4346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6729639162148894e-05,
+      "loss": 1.0352,
+      "step": 4347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.672793163989642e-05,
+      "loss": 0.8945,
+      "step": 4348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6726223711706244e-05,
+      "loss": 0.9775,
+      "step": 4349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.672451537761981e-05,
+      "loss": 0.8633,
+      "step": 4350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6722806637678566e-05,
+      "loss": 0.915,
+      "step": 4351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.672109749192399e-05,
+      "loss": 0.8711,
+      "step": 4352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6719387940397544e-05,
+      "loss": 0.7656,
+      "step": 4353
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6717677983140726e-05,
+      "loss": 1.1914,
+      "step": 4354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.671596762019504e-05,
+      "loss": 0.8374,
+      "step": 4355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6714256851601966e-05,
+      "loss": 0.9424,
+      "step": 4356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.671254567740304e-05,
+      "loss": 1.1562,
+      "step": 4357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.671083409763978e-05,
+      "loss": 1.0332,
+      "step": 4358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6709122112353714e-05,
+      "loss": 1.04,
+      "step": 4359
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6707409721586404e-05,
+      "loss": 0.9717,
+      "step": 4360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.670569692537939e-05,
+      "loss": 0.7925,
+      "step": 4361
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.670398372377424e-05,
+      "loss": 0.9473,
+      "step": 4362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.670227011681253e-05,
+      "loss": 1.0322,
+      "step": 4363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6700556104535845e-05,
+      "loss": 0.835,
+      "step": 4364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.669884168698578e-05,
+      "loss": 1.0205,
+      "step": 4365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6697126864203934e-05,
+      "loss": 0.8662,
+      "step": 4366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.669541163623192e-05,
+      "loss": 1.0039,
+      "step": 4367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6693696003111375e-05,
+      "loss": 1.0186,
+      "step": 4368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.669197996488391e-05,
+      "loss": 1.1592,
+      "step": 4369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6690263521591195e-05,
+      "loss": 1.1172,
+      "step": 4370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.668854667327486e-05,
+      "loss": 0.9902,
+      "step": 4371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6686829419976576e-05,
+      "loss": 0.9277,
+      "step": 4372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.668511176173802e-05,
+      "loss": 0.9668,
+      "step": 4373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.668339369860086e-05,
+      "loss": 0.9834,
+      "step": 4374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.668167523060681e-05,
+      "loss": 0.9365,
+      "step": 4375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.667995635779756e-05,
+      "loss": 1.0332,
+      "step": 4376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.667823708021482e-05,
+      "loss": 1.1152,
+      "step": 4377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.667651739790032e-05,
+      "loss": 1.0947,
+      "step": 4378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.667479731089579e-05,
+      "loss": 0.9443,
+      "step": 4379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.667307681924295e-05,
+      "loss": 1.0439,
+      "step": 4380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6671355922983586e-05,
+      "loss": 0.8916,
+      "step": 4381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.666963462215944e-05,
+      "loss": 1.0127,
+      "step": 4382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.666791291681228e-05,
+      "loss": 0.9717,
+      "step": 4383
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.666619080698389e-05,
+      "loss": 0.9619,
+      "step": 4384
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.666446829271607e-05,
+      "loss": 0.8467,
+      "step": 4385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.666274537405061e-05,
+      "loss": 0.8965,
+      "step": 4386
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6661022051029324e-05,
+      "loss": 0.8857,
+      "step": 4387
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.665929832369403e-05,
+      "loss": 0.8135,
+      "step": 4388
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6657574192086556e-05,
+      "loss": 0.9863,
+      "step": 4389
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6655849656248754e-05,
+      "loss": 0.8809,
+      "step": 4390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6654124716222455e-05,
+      "loss": 0.8657,
+      "step": 4391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.665239937204953e-05,
+      "loss": 0.8965,
+      "step": 4392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.665067362377185e-05,
+      "loss": 1.0176,
+      "step": 4393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.664894747143129e-05,
+      "loss": 0.9062,
+      "step": 4394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.664722091506972e-05,
+      "loss": 0.9717,
+      "step": 4395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6645493954729074e-05,
+      "loss": 1.0625,
+      "step": 4396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6643766590451235e-05,
+      "loss": 1.0234,
+      "step": 4397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.664203882227813e-05,
+      "loss": 0.9297,
+      "step": 4398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6640310650251686e-05,
+      "loss": 1.1152,
+      "step": 4399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.663858207441384e-05,
+      "loss": 0.8389,
+      "step": 4400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.663685309480654e-05,
+      "loss": 1.1162,
+      "step": 4401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6635123711471735e-05,
+      "loss": 1.1621,
+      "step": 4402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.663339392445141e-05,
+      "loss": 0.9775,
+      "step": 4403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6631663733787525e-05,
+      "loss": 0.9561,
+      "step": 4404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.662993313952208e-05,
+      "loss": 1.0303,
+      "step": 4405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.662820214169705e-05,
+      "loss": 0.8887,
+      "step": 4406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.662647074035447e-05,
+      "loss": 0.9834,
+      "step": 4407
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6624738935536336e-05,
+      "loss": 0.8286,
+      "step": 4408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.662300672728468e-05,
+      "loss": 1.0205,
+      "step": 4409
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.662127411564154e-05,
+      "loss": 0.9814,
+      "step": 4410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.661954110064896e-05,
+      "loss": 1.1973,
+      "step": 4411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6617807682348986e-05,
+      "loss": 0.8257,
+      "step": 4412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6616073860783697e-05,
+      "loss": 1.0342,
+      "step": 4413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.661433963599516e-05,
+      "loss": 1.127,
+      "step": 4414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6612605008025464e-05,
+      "loss": 1.0166,
+      "step": 4415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6610869976916695e-05,
+      "loss": 1.0156,
+      "step": 4416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6609134542710965e-05,
+      "loss": 1.0264,
+      "step": 4417
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6607398705450386e-05,
+      "loss": 1.0986,
+      "step": 4418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.660566246517708e-05,
+      "loss": 0.9688,
+      "step": 4419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6603925821933185e-05,
+      "loss": 1.1191,
+      "step": 4420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.660218877576084e-05,
+      "loss": 1.1719,
+      "step": 4421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6600451326702196e-05,
+      "loss": 1.0693,
+      "step": 4422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6598713474799414e-05,
+      "loss": 1.0518,
+      "step": 4423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6596975220094674e-05,
+      "loss": 1.0215,
+      "step": 4424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.659523656263016e-05,
+      "loss": 1.0186,
+      "step": 4425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.659349750244806e-05,
+      "loss": 0.9893,
+      "step": 4426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6591758039590567e-05,
+      "loss": 1.1934,
+      "step": 4427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.659001817409991e-05,
+      "loss": 0.958,
+      "step": 4428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.658827790601829e-05,
+      "loss": 0.9561,
+      "step": 4429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.658653723538796e-05,
+      "loss": 1.123,
+      "step": 4430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6584796162251145e-05,
+      "loss": 1.0576,
+      "step": 4431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6583054686650104e-05,
+      "loss": 0.6323,
+      "step": 4432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.658131280862709e-05,
+      "loss": 1.0801,
+      "step": 4433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6579570528224385e-05,
+      "loss": 0.9492,
+      "step": 4434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.657782784548427e-05,
+      "loss": 0.877,
+      "step": 4435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6576084760449016e-05,
+      "loss": 1.0674,
+      "step": 4436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.657434127316093e-05,
+      "loss": 1.0078,
+      "step": 4437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.657259738366233e-05,
+      "loss": 1.0498,
+      "step": 4438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.657085309199553e-05,
+      "loss": 1.1484,
+      "step": 4439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.656910839820286e-05,
+      "loss": 0.8711,
+      "step": 4440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.656736330232665e-05,
+      "loss": 0.8848,
+      "step": 4441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6565617804409254e-05,
+      "loss": 0.9712,
+      "step": 4442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.656387190449304e-05,
+      "loss": 1.0654,
+      "step": 4443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.656212560262036e-05,
+      "loss": 1.0811,
+      "step": 4444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.656037889883361e-05,
+      "loss": 1.0713,
+      "step": 4445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.655863179317516e-05,
+      "loss": 1.0547,
+      "step": 4446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6556884285687406e-05,
+      "loss": 1.0391,
+      "step": 4447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.655513637641277e-05,
+      "loss": 1.002,
+      "step": 4448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.655338806539367e-05,
+      "loss": 0.9014,
+      "step": 4449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.655163935267251e-05,
+      "loss": 0.8467,
+      "step": 4450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.654989023829174e-05,
+      "loss": 1.0635,
+      "step": 4451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.654814072229381e-05,
+      "loss": 0.9795,
+      "step": 4452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.654639080472117e-05,
+      "loss": 0.8076,
+      "step": 4453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6544640485616286e-05,
+      "loss": 1.0049,
+      "step": 4454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.654288976502164e-05,
+      "loss": 1.1172,
+      "step": 4455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.65411386429797e-05,
+      "loss": 1.0439,
+      "step": 4456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.653938711953297e-05,
+      "loss": 1.0488,
+      "step": 4457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.653763519472397e-05,
+      "loss": 1.1094,
+      "step": 4458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.653588286859519e-05,
+      "loss": 0.9648,
+      "step": 4459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.653413014118916e-05,
+      "loss": 1.0225,
+      "step": 4460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.653237701254842e-05,
+      "loss": 0.9434,
+      "step": 4461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6530623482715506e-05,
+      "loss": 1.0107,
+      "step": 4462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.652886955173298e-05,
+      "loss": 1.0137,
+      "step": 4463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.65271152196434e-05,
+      "loss": 0.8701,
+      "step": 4464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.652536048648934e-05,
+      "loss": 0.9561,
+      "step": 4465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.652360535231337e-05,
+      "loss": 0.7959,
+      "step": 4466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.65218498171581e-05,
+      "loss": 0.877,
+      "step": 4467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6520093881066125e-05,
+      "loss": 1.082,
+      "step": 4468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6518337544080054e-05,
+      "loss": 1.1211,
+      "step": 4469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6516580806242504e-05,
+      "loss": 0.9219,
+      "step": 4470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.651482366759612e-05,
+      "loss": 0.9785,
+      "step": 4471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.651306612818353e-05,
+      "loss": 0.916,
+      "step": 4472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6511308188047385e-05,
+      "loss": 1.0439,
+      "step": 4473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6509549847230354e-05,
+      "loss": 0.9062,
+      "step": 4474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6507791105775094e-05,
+      "loss": 0.958,
+      "step": 4475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.65060319637243e-05,
+      "loss": 0.8926,
+      "step": 4476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.650427242112064e-05,
+      "loss": 0.998,
+      "step": 4477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.650251247800684e-05,
+      "loss": 0.9287,
+      "step": 4478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.650075213442559e-05,
+      "loss": 1.0645,
+      "step": 4479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.649899139041961e-05,
+      "loss": 1.0205,
+      "step": 4480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.649723024603163e-05,
+      "loss": 0.876,
+      "step": 4481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6495468701304394e-05,
+      "loss": 1.0791,
+      "step": 4482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.649370675628064e-05,
+      "loss": 0.9502,
+      "step": 4483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.649194441100313e-05,
+      "loss": 1.0068,
+      "step": 4484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6490181665514633e-05,
+      "loss": 0.9336,
+      "step": 4485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.648841851985792e-05,
+      "loss": 0.9512,
+      "step": 4486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.648665497407578e-05,
+      "loss": 1.0879,
+      "step": 4487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.648489102821101e-05,
+      "loss": 1.0996,
+      "step": 4488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6483126682306405e-05,
+      "loss": 0.9639,
+      "step": 4489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.64813619364048e-05,
+      "loss": 1.1992,
+      "step": 4490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6479596790549014e-05,
+      "loss": 1.0352,
+      "step": 4491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.647783124478188e-05,
+      "loss": 0.9346,
+      "step": 4492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.647606529914623e-05,
+      "loss": 1.0645,
+      "step": 4493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6474298953684945e-05,
+      "loss": 0.8281,
+      "step": 4494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6472532208440864e-05,
+      "loss": 0.9629,
+      "step": 4495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.647076506345687e-05,
+      "loss": 0.7588,
+      "step": 4496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.646899751877585e-05,
+      "loss": 0.9707,
+      "step": 4497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.646722957444069e-05,
+      "loss": 0.7793,
+      "step": 4498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.646546123049429e-05,
+      "loss": 1.0713,
+      "step": 4499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.646369248697958e-05,
+      "loss": 1.0371,
+      "step": 4500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.646192334393947e-05,
+      "loss": 0.9648,
+      "step": 4501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6460153801416894e-05,
+      "loss": 0.9824,
+      "step": 4502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.645838385945479e-05,
+      "loss": 1.0449,
+      "step": 4503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.645661351809611e-05,
+      "loss": 0.9365,
+      "step": 4504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6454842777383824e-05,
+      "loss": 1.0977,
+      "step": 4505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.645307163736089e-05,
+      "loss": 1.0107,
+      "step": 4506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6451300098070296e-05,
+      "loss": 1.1592,
+      "step": 4507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.644952815955503e-05,
+      "loss": 1.0293,
+      "step": 4508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6447755821858095e-05,
+      "loss": 0.7026,
+      "step": 4509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6445983085022494e-05,
+      "loss": 1.0254,
+      "step": 4510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6444209949091246e-05,
+      "loss": 0.8584,
+      "step": 4511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.644243641410739e-05,
+      "loss": 0.9424,
+      "step": 4512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.644066248011395e-05,
+      "loss": 1.0469,
+      "step": 4513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6438888147153985e-05,
+      "loss": 1.1699,
+      "step": 4514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.643711341527055e-05,
+      "loss": 1.0547,
+      "step": 4515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6435338284506716e-05,
+      "loss": 1.0215,
+      "step": 4516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.643356275490555e-05,
+      "loss": 1.0703,
+      "step": 4517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6431786826510146e-05,
+      "loss": 1.0674,
+      "step": 4518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.643001049936361e-05,
+      "loss": 0.9482,
+      "step": 4519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6428233773509025e-05,
+      "loss": 0.8994,
+      "step": 4520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.642645664898953e-05,
+      "loss": 0.9414,
+      "step": 4521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.642467912584824e-05,
+      "loss": 1.0,
+      "step": 4522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.642290120412828e-05,
+      "loss": 0.9375,
+      "step": 4523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.642112288387281e-05,
+      "loss": 1.3066,
+      "step": 4524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.641934416512499e-05,
+      "loss": 1.0723,
+      "step": 4525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.641756504792797e-05,
+      "loss": 0.9385,
+      "step": 4526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.641578553232493e-05,
+      "loss": 0.9258,
+      "step": 4527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.641400561835904e-05,
+      "loss": 1.1699,
+      "step": 4528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.641222530607352e-05,
+      "loss": 0.8916,
+      "step": 4529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.641044459551155e-05,
+      "loss": 0.876,
+      "step": 4530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.640866348671636e-05,
+      "loss": 0.9648,
+      "step": 4531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6406881979731155e-05,
+      "loss": 1.0742,
+      "step": 4532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6405100074599176e-05,
+      "loss": 1.0918,
+      "step": 4533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.640331777136366e-05,
+      "loss": 1.0518,
+      "step": 4534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6401535070067876e-05,
+      "loss": 1.0391,
+      "step": 4535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.639975197075506e-05,
+      "loss": 1.1113,
+      "step": 4536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.639796847346849e-05,
+      "loss": 1.0293,
+      "step": 4537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.639618457825146e-05,
+      "loss": 0.9727,
+      "step": 4538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.639440028514725e-05,
+      "loss": 1.0947,
+      "step": 4539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.639261559419916e-05,
+      "loss": 0.8325,
+      "step": 4540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6390830505450485e-05,
+      "loss": 1.0088,
+      "step": 4541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6389045018944566e-05,
+      "loss": 0.8926,
+      "step": 4542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.638725913472472e-05,
+      "loss": 0.835,
+      "step": 4543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.638547285283429e-05,
+      "loss": 1.0566,
+      "step": 4544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6383686173316626e-05,
+      "loss": 0.9053,
+      "step": 4545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.638189909621508e-05,
+      "loss": 1.1025,
+      "step": 4546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6380111621573015e-05,
+      "loss": 0.9346,
+      "step": 4547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.637832374943382e-05,
+      "loss": 1.0469,
+      "step": 4548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.637653547984087e-05,
+      "loss": 0.9434,
+      "step": 4549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6374746812837565e-05,
+      "loss": 1.0625,
+      "step": 4550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.637295774846731e-05,
+      "loss": 0.9551,
+      "step": 4551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.637116828677353e-05,
+      "loss": 1.0107,
+      "step": 4552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.636937842779963e-05,
+      "loss": 1.0645,
+      "step": 4553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.636758817158906e-05,
+      "loss": 0.9941,
+      "step": 4554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.636579751818526e-05,
+      "loss": 0.9043,
+      "step": 4555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6364006467631694e-05,
+      "loss": 1.1504,
+      "step": 4556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.636221501997181e-05,
+      "loss": 1.0635,
+      "step": 4557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6360423175249085e-05,
+      "loss": 0.9277,
+      "step": 4558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6358630933507e-05,
+      "loss": 1.2129,
+      "step": 4559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6356838294789056e-05,
+      "loss": 0.8164,
+      "step": 4560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.635504525913875e-05,
+      "loss": 0.9355,
+      "step": 4561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6353251826599594e-05,
+      "loss": 1.0596,
+      "step": 4562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.635145799721511e-05,
+      "loss": 1.1211,
+      "step": 4563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.634966377102883e-05,
+      "loss": 1.0449,
+      "step": 4564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.634786914808428e-05,
+      "loss": 1.1289,
+      "step": 4565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6346074128425045e-05,
+      "loss": 0.9688,
+      "step": 4566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6344278712094646e-05,
+      "loss": 0.8486,
+      "step": 4567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.634248289913668e-05,
+      "loss": 1.0293,
+      "step": 4568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.634068668959471e-05,
+      "loss": 0.9561,
+      "step": 4569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.633889008351233e-05,
+      "loss": 1.0742,
+      "step": 4570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.633709308093313e-05,
+      "loss": 1.1523,
+      "step": 4571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.633529568190074e-05,
+      "loss": 0.9629,
+      "step": 4572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.633349788645876e-05,
+      "loss": 1.1279,
+      "step": 4573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6331699694650826e-05,
+      "loss": 1.0234,
+      "step": 4574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.632990110652056e-05,
+      "loss": 1.0811,
+      "step": 4575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6328102122111625e-05,
+      "loss": 1.0234,
+      "step": 4576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6326302741467675e-05,
+      "loss": 0.9863,
+      "step": 4577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.632450296463237e-05,
+      "loss": 0.8682,
+      "step": 4578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6322702791649375e-05,
+      "loss": 1.1514,
+      "step": 4579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6320902222562405e-05,
+      "loss": 1.0488,
+      "step": 4580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.631910125741512e-05,
+      "loss": 1.0957,
+      "step": 4581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.631729989625124e-05,
+      "loss": 0.9414,
+      "step": 4582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.631549813911449e-05,
+      "loss": 1.0586,
+      "step": 4583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.631369598604858e-05,
+      "loss": 1.0635,
+      "step": 4584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.631189343709724e-05,
+      "loss": 1.0225,
+      "step": 4585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6310090492304225e-05,
+      "loss": 1.0322,
+      "step": 4586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.630828715171327e-05,
+      "loss": 1.1074,
+      "step": 4587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6306483415368154e-05,
+      "loss": 1.0264,
+      "step": 4588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.630467928331264e-05,
+      "loss": 1.0527,
+      "step": 4589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.630287475559051e-05,
+      "loss": 1.0381,
+      "step": 4590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6301069832245553e-05,
+      "loss": 0.9863,
+      "step": 4591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.629926451332158e-05,
+      "loss": 0.8877,
+      "step": 4592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6297458798862374e-05,
+      "loss": 1.0859,
+      "step": 4593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6295652688911785e-05,
+      "loss": 0.9111,
+      "step": 4594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6293846183513626e-05,
+      "loss": 1.0371,
+      "step": 4595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.629203928271174e-05,
+      "loss": 1.043,
+      "step": 4596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.629023198654997e-05,
+      "loss": 0.8418,
+      "step": 4597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.628842429507218e-05,
+      "loss": 1.0215,
+      "step": 4598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6286616208322234e-05,
+      "loss": 1.0283,
+      "step": 4599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6284807726344015e-05,
+      "loss": 1.0479,
+      "step": 4600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.62829988491814e-05,
+      "loss": 1.0068,
+      "step": 4601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6281189576878285e-05,
+      "loss": 0.8877,
+      "step": 4602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.627937990947859e-05,
+      "loss": 0.8809,
+      "step": 4603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.627756984702621e-05,
+      "loss": 0.9326,
+      "step": 4604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6275759389565086e-05,
+      "loss": 1.0576,
+      "step": 4605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.627394853713915e-05,
+      "loss": 0.6934,
+      "step": 4606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.627213728979234e-05,
+      "loss": 0.9707,
+      "step": 4607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.627032564756861e-05,
+      "loss": 0.9854,
+      "step": 4608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.626851361051193e-05,
+      "loss": 0.873,
+      "step": 4609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.626670117866627e-05,
+      "loss": 0.7803,
+      "step": 4610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.626488835207561e-05,
+      "loss": 0.999,
+      "step": 4611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6263075130783934e-05,
+      "loss": 0.9072,
+      "step": 4612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6261261514835256e-05,
+      "loss": 0.8428,
+      "step": 4613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.625944750427359e-05,
+      "loss": 0.9502,
+      "step": 4614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6257633099142943e-05,
+      "loss": 0.9863,
+      "step": 4615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.625581829948736e-05,
+      "loss": 0.959,
+      "step": 4616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.625400310535087e-05,
+      "loss": 0.9487,
+      "step": 4617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.625218751677753e-05,
+      "loss": 1.1133,
+      "step": 4618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6250371533811393e-05,
+      "loss": 1.0361,
+      "step": 4619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.624855515649652e-05,
+      "loss": 1.0166,
+      "step": 4620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6246738384877016e-05,
+      "loss": 0.9307,
+      "step": 4621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.624492121899694e-05,
+      "loss": 0.7812,
+      "step": 4622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.62431036589004e-05,
+      "loss": 1.082,
+      "step": 4623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.62412857046315e-05,
+      "loss": 1.082,
+      "step": 4624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.623946735623437e-05,
+      "loss": 1.1514,
+      "step": 4625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.623764861375312e-05,
+      "loss": 1.1465,
+      "step": 4626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.62358294772319e-05,
+      "loss": 1.0225,
+      "step": 4627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.623400994671484e-05,
+      "loss": 1.1035,
+      "step": 4628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.62321900222461e-05,
+      "loss": 0.9453,
+      "step": 4629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6230369703869844e-05,
+      "loss": 1.0107,
+      "step": 4630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6228548991630245e-05,
+      "loss": 0.9492,
+      "step": 4631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.622672788557149e-05,
+      "loss": 1.0195,
+      "step": 4632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.622490638573778e-05,
+      "loss": 0.9629,
+      "step": 4633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.62230844921733e-05,
+      "loss": 0.9199,
+      "step": 4634
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.622126220492227e-05,
+      "loss": 0.9922,
+      "step": 4635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.621943952402891e-05,
+      "loss": 0.9248,
+      "step": 4636
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.621761644953745e-05,
+      "loss": 1.0,
+      "step": 4637
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6215792981492134e-05,
+      "loss": 0.9951,
+      "step": 4638
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.621396911993722e-05,
+      "loss": 1.0098,
+      "step": 4639
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6212144864916956e-05,
+      "loss": 0.8193,
+      "step": 4640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.621032021647561e-05,
+      "loss": 1.1914,
+      "step": 4641
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6208495174657466e-05,
+      "loss": 0.8174,
+      "step": 4642
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.620666973950681e-05,
+      "loss": 1.0391,
+      "step": 4643
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.620484391106795e-05,
+      "loss": 0.9519,
+      "step": 4644
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.620301768938517e-05,
+      "loss": 0.9971,
+      "step": 4645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.620119107450282e-05,
+      "loss": 0.998,
+      "step": 4646
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.61993640664652e-05,
+      "loss": 0.9473,
+      "step": 4647
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.619753666531665e-05,
+      "loss": 0.9756,
+      "step": 4648
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.619570887110152e-05,
+      "loss": 1.041,
+      "step": 4649
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.619388068386418e-05,
+      "loss": 0.9766,
+      "step": 4650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6192052103648966e-05,
+      "loss": 1.0332,
+      "step": 4651
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.619022313050027e-05,
+      "loss": 0.8838,
+      "step": 4652
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.618839376446247e-05,
+      "loss": 1.1738,
+      "step": 4653
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.618656400557997e-05,
+      "loss": 0.8901,
+      "step": 4654
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6184733853897156e-05,
+      "loss": 0.7637,
+      "step": 4655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.618290330945845e-05,
+      "loss": 1.0342,
+      "step": 4656
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6181072372308285e-05,
+      "loss": 0.9131,
+      "step": 4657
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.617924104249107e-05,
+      "loss": 1.0234,
+      "step": 4658
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.617740932005125e-05,
+      "loss": 1.1074,
+      "step": 4659
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.617557720503329e-05,
+      "loss": 1.1348,
+      "step": 4660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.617374469748164e-05,
+      "loss": 1.0312,
+      "step": 4661
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6171911797440776e-05,
+      "loss": 1.0752,
+      "step": 4662
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.617007850495517e-05,
+      "loss": 0.9316,
+      "step": 4663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.616824482006931e-05,
+      "loss": 1.1865,
+      "step": 4664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.61664107428277e-05,
+      "loss": 0.9346,
+      "step": 4665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.616457627327484e-05,
+      "loss": 1.0928,
+      "step": 4666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6162741411455264e-05,
+      "loss": 1.0449,
+      "step": 4667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6160906157413475e-05,
+      "loss": 1.043,
+      "step": 4668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.615907051119403e-05,
+      "loss": 0.9258,
+      "step": 4669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.615723447284146e-05,
+      "loss": 0.7646,
+      "step": 4670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6155398042400325e-05,
+      "loss": 0.9873,
+      "step": 4671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.61535612199152e-05,
+      "loss": 1.0117,
+      "step": 4672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.615172400543064e-05,
+      "loss": 1.0479,
+      "step": 4673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6149886398991244e-05,
+      "loss": 0.8916,
+      "step": 4674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6148048400641595e-05,
+      "loss": 0.9648,
+      "step": 4675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.614621001042631e-05,
+      "loss": 0.7432,
+      "step": 4676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.614437122838998e-05,
+      "loss": 1.043,
+      "step": 4677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.614253205457725e-05,
+      "loss": 0.9756,
+      "step": 4678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.614069248903273e-05,
+      "loss": 0.9072,
+      "step": 4679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.613885253180108e-05,
+      "loss": 1.1064,
+      "step": 4680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6137012182926946e-05,
+      "loss": 1.0439,
+      "step": 4681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.613517144245498e-05,
+      "loss": 1.1494,
+      "step": 4682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.613333031042984e-05,
+      "loss": 0.7744,
+      "step": 4683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6131488786896236e-05,
+      "loss": 0.8506,
+      "step": 4684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.612964687189884e-05,
+      "loss": 1.0146,
+      "step": 4685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.612780456548234e-05,
+      "loss": 1.0068,
+      "step": 4686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6125961867691464e-05,
+      "loss": 0.9609,
+      "step": 4687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.612411877857091e-05,
+      "loss": 0.9453,
+      "step": 4688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.612227529816542e-05,
+      "loss": 1.0527,
+      "step": 4689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.612043142651971e-05,
+      "loss": 0.9648,
+      "step": 4690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6118587163678554e-05,
+      "loss": 0.8369,
+      "step": 4691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.611674250968668e-05,
+      "loss": 1.0039,
+      "step": 4692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6114897464588865e-05,
+      "loss": 1.0078,
+      "step": 4693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.611305202842989e-05,
+      "loss": 1.0654,
+      "step": 4694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.611120620125452e-05,
+      "loss": 1.0723,
+      "step": 4695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6109359983107555e-05,
+      "loss": 0.999,
+      "step": 4696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.61075133740338e-05,
+      "loss": 1.0273,
+      "step": 4697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.610566637407806e-05,
+      "loss": 1.1025,
+      "step": 4698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.610381898328518e-05,
+      "loss": 1.0615,
+      "step": 4699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.610197120169996e-05,
+      "loss": 1.0576,
+      "step": 4700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.610012302936725e-05,
+      "loss": 0.9521,
+      "step": 4701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.609827446633191e-05,
+      "loss": 1.0078,
+      "step": 4702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.609642551263879e-05,
+      "loss": 0.875,
+      "step": 4703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.609457616833276e-05,
+      "loss": 0.9854,
+      "step": 4704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.609272643345871e-05,
+      "loss": 1.0547,
+      "step": 4705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6090876308061503e-05,
+      "loss": 0.9219,
+      "step": 4706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6089025792186054e-05,
+      "loss": 1.1025,
+      "step": 4707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6087174885877266e-05,
+      "loss": 0.9424,
+      "step": 4708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6085323589180046e-05,
+      "loss": 1.1289,
+      "step": 4709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.608347190213934e-05,
+      "loss": 1.042,
+      "step": 4710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.608161982480006e-05,
+      "loss": 0.8184,
+      "step": 4711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6079767357207166e-05,
+      "loss": 0.8672,
+      "step": 4712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.607791449940561e-05,
+      "loss": 1.1406,
+      "step": 4713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6076061251440346e-05,
+      "loss": 0.8809,
+      "step": 4714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.607420761335636e-05,
+      "loss": 0.8447,
+      "step": 4715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6072353585198626e-05,
+      "loss": 1.0596,
+      "step": 4716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.607049916701214e-05,
+      "loss": 0.8555,
+      "step": 4717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.606864435884189e-05,
+      "loss": 0.9502,
+      "step": 4718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6066789160732906e-05,
+      "loss": 1.0977,
+      "step": 4719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6064933572730195e-05,
+      "loss": 1.0195,
+      "step": 4720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.60630775948788e-05,
+      "loss": 0.9883,
+      "step": 4721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.606122122722375e-05,
+      "loss": 1.0029,
+      "step": 4722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.605936446981009e-05,
+      "loss": 0.8574,
+      "step": 4723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6057507322682884e-05,
+      "loss": 0.959,
+      "step": 4724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.60556497858872e-05,
+      "loss": 0.9551,
+      "step": 4725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6053791859468115e-05,
+      "loss": 0.9727,
+      "step": 4726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.605193354347071e-05,
+      "loss": 0.9873,
+      "step": 4727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6050074837940094e-05,
+      "loss": 1.0898,
+      "step": 4728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6048215742921355e-05,
+      "loss": 0.9238,
+      "step": 4729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6046356258459616e-05,
+      "loss": 0.6631,
+      "step": 4730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6044496384600005e-05,
+      "loss": 0.8291,
+      "step": 4731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6042636121387646e-05,
+      "loss": 0.9785,
+      "step": 4732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6040775468867696e-05,
+      "loss": 0.9121,
+      "step": 4733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.60389144270853e-05,
+      "loss": 0.9971,
+      "step": 4734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.603705299608561e-05,
+      "loss": 1.0957,
+      "step": 4735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6035191175913814e-05,
+      "loss": 0.9824,
+      "step": 4736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.603332896661508e-05,
+      "loss": 1.0244,
+      "step": 4737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.603146636823461e-05,
+      "loss": 0.9443,
+      "step": 4738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6029603380817597e-05,
+      "loss": 1.1504,
+      "step": 4739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.602774000440925e-05,
+      "loss": 0.7603,
+      "step": 4740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.602587623905479e-05,
+      "loss": 0.9424,
+      "step": 4741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.602401208479945e-05,
+      "loss": 0.8574,
+      "step": 4742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.602214754168845e-05,
+      "loss": 1.0996,
+      "step": 4743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6020282609767054e-05,
+      "loss": 0.8857,
+      "step": 4744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.601841728908052e-05,
+      "loss": 0.9082,
+      "step": 4745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6016551579674095e-05,
+      "loss": 0.9028,
+      "step": 4746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6014685481593076e-05,
+      "loss": 0.9746,
+      "step": 4747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.601281899488273e-05,
+      "loss": 0.959,
+      "step": 4748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.601095211958837e-05,
+      "loss": 0.9873,
+      "step": 4749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.600908485575527e-05,
+      "loss": 0.9307,
+      "step": 4750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6007217203428775e-05,
+      "loss": 0.9795,
+      "step": 4751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.600534916265419e-05,
+      "loss": 0.9092,
+      "step": 4752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.600348073347685e-05,
+      "loss": 0.9365,
+      "step": 4753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.600161191594211e-05,
+      "loss": 0.7686,
+      "step": 4754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5999742710095294e-05,
+      "loss": 0.9434,
+      "step": 4755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.599787311598178e-05,
+      "loss": 1.126,
+      "step": 4756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.599600313364693e-05,
+      "loss": 0.8115,
+      "step": 4757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.599413276313613e-05,
+      "loss": 1.0342,
+      "step": 4758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5992262004494764e-05,
+      "loss": 1.1035,
+      "step": 4759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.599039085776823e-05,
+      "loss": 0.9121,
+      "step": 4760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5988519323001935e-05,
+      "loss": 1.0938,
+      "step": 4761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5986647400241296e-05,
+      "loss": 1.1006,
+      "step": 4762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.598477508953175e-05,
+      "loss": 1.0361,
+      "step": 4763
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.598290239091871e-05,
+      "loss": 1.0537,
+      "step": 4764
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5981029304447635e-05,
+      "loss": 1.042,
+      "step": 4765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.597915583016398e-05,
+      "loss": 0.9336,
+      "step": 4766
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5977281968113203e-05,
+      "loss": 1.0205,
+      "step": 4767
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5975407718340786e-05,
+      "loss": 0.9727,
+      "step": 4768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.59735330808922e-05,
+      "loss": 0.9678,
+      "step": 4769
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.597165805581295e-05,
+      "loss": 1.0137,
+      "step": 4770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.596978264314852e-05,
+      "loss": 0.9893,
+      "step": 4771
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.596790684294444e-05,
+      "loss": 0.9424,
+      "step": 4772
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.596603065524622e-05,
+      "loss": 1.0684,
+      "step": 4773
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5964154080099384e-05,
+      "loss": 0.8975,
+      "step": 4774
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5962277117549474e-05,
+      "loss": 1.0469,
+      "step": 4775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.596039976764205e-05,
+      "loss": 1.042,
+      "step": 4776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.595852203042266e-05,
+      "loss": 0.8848,
+      "step": 4777
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.595664390593687e-05,
+      "loss": 1.0615,
+      "step": 4778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.595476539423026e-05,
+      "loss": 0.9951,
+      "step": 4779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.595288649534842e-05,
+      "loss": 1.0361,
+      "step": 4780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5951007209336937e-05,
+      "loss": 0.8379,
+      "step": 4781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.594912753624142e-05,
+      "loss": 1.0742,
+      "step": 4782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.594724747610748e-05,
+      "loss": 0.9453,
+      "step": 4783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5945367028980744e-05,
+      "loss": 0.8159,
+      "step": 4784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.594348619490684e-05,
+      "loss": 1.0732,
+      "step": 4785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5941604973931425e-05,
+      "loss": 0.8574,
+      "step": 4786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.593972336610013e-05,
+      "loss": 0.9551,
+      "step": 4787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5937841371458634e-05,
+      "loss": 1.0312,
+      "step": 4788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.593595899005259e-05,
+      "loss": 1.0391,
+      "step": 4789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5934076221927694e-05,
+      "loss": 1.084,
+      "step": 4790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5932193067129627e-05,
+      "loss": 0.9756,
+      "step": 4791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.593030952570409e-05,
+      "loss": 1.0918,
+      "step": 4792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5928425597696786e-05,
+      "loss": 1.0752,
+      "step": 4793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.592654128315344e-05,
+      "loss": 0.9668,
+      "step": 4794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.592465658211977e-05,
+      "loss": 1.0664,
+      "step": 4795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.592277149464153e-05,
+      "loss": 0.9346,
+      "step": 4796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.592088602076444e-05,
+      "loss": 0.9639,
+      "step": 4797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.591900016053428e-05,
+      "loss": 1.0078,
+      "step": 4798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5917113913996795e-05,
+      "loss": 0.9082,
+      "step": 4799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.591522728119776e-05,
+      "loss": 1.0801,
+      "step": 4800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.591334026218297e-05,
+      "loss": 0.8086,
+      "step": 4801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5911452856998214e-05,
+      "loss": 1.0537,
+      "step": 4802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5909565065689286e-05,
+      "loss": 1.0547,
+      "step": 4803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.590767688830201e-05,
+      "loss": 0.9316,
+      "step": 4804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.590578832488219e-05,
+      "loss": 1.0713,
+      "step": 4805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.590389937547566e-05,
+      "loss": 0.998,
+      "step": 4806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5902010040128284e-05,
+      "loss": 1.2188,
+      "step": 4807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.590012031888587e-05,
+      "loss": 0.9639,
+      "step": 4808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.589823021179431e-05,
+      "loss": 1.0156,
+      "step": 4809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.589633971889945e-05,
+      "loss": 1.0186,
+      "step": 4810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5894448840247174e-05,
+      "loss": 0.8438,
+      "step": 4811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5892557575883374e-05,
+      "loss": 0.9883,
+      "step": 4812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.589066592585393e-05,
+      "loss": 1.0693,
+      "step": 4813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5888773890204774e-05,
+      "loss": 1.0264,
+      "step": 4814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5886881468981785e-05,
+      "loss": 0.9697,
+      "step": 4815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.588498866223092e-05,
+      "loss": 0.877,
+      "step": 4816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5883095469998086e-05,
+      "loss": 1.0342,
+      "step": 4817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.588120189232924e-05,
+      "loss": 1.0645,
+      "step": 4818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5879307929270325e-05,
+      "loss": 0.8975,
+      "step": 4819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.587741358086731e-05,
+      "loss": 0.9053,
+      "step": 4820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.587551884716616e-05,
+      "loss": 0.8652,
+      "step": 4821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.587362372821285e-05,
+      "loss": 0.9727,
+      "step": 4822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.587172822405338e-05,
+      "loss": 1.0078,
+      "step": 4823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.586983233473374e-05,
+      "loss": 1.0645,
+      "step": 4824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5867936060299945e-05,
+      "loss": 0.9453,
+      "step": 4825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.586603940079801e-05,
+      "loss": 0.8916,
+      "step": 4826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.586414235627396e-05,
+      "loss": 1.0459,
+      "step": 4827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.586224492677382e-05,
+      "loss": 0.8687,
+      "step": 4828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.586034711234365e-05,
+      "loss": 0.9971,
+      "step": 4829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.58584489130295e-05,
+      "loss": 0.915,
+      "step": 4830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.585655032887743e-05,
+      "loss": 0.9795,
+      "step": 4831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.585465135993352e-05,
+      "loss": 1.0596,
+      "step": 4832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5852752006243844e-05,
+      "loss": 1.0049,
+      "step": 4833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.58508522678545e-05,
+      "loss": 1.0586,
+      "step": 4834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.584895214481159e-05,
+      "loss": 0.7993,
+      "step": 4835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.584705163716122e-05,
+      "loss": 1.0879,
+      "step": 4836
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.584515074494951e-05,
+      "loss": 0.8662,
+      "step": 4837
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5843249468222585e-05,
+      "loss": 1.1045,
+      "step": 4838
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.58413478070266e-05,
+      "loss": 1.0244,
+      "step": 4839
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5839445761407685e-05,
+      "loss": 0.9209,
+      "step": 4840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5837543331412e-05,
+      "loss": 0.8145,
+      "step": 4841
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5835640517085726e-05,
+      "loss": 0.9111,
+      "step": 4842
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.583373731847502e-05,
+      "loss": 1.0498,
+      "step": 4843
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.583183373562607e-05,
+      "loss": 0.9463,
+      "step": 4844
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5829929768585086e-05,
+      "loss": 0.9551,
+      "step": 4845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.582802541739825e-05,
+      "loss": 1.1221,
+      "step": 4846
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.582612068211179e-05,
+      "loss": 0.9668,
+      "step": 4847
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5824215562771925e-05,
+      "loss": 1.1016,
+      "step": 4848
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.582231005942489e-05,
+      "loss": 1.0283,
+      "step": 4849
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5820404172116916e-05,
+      "loss": 0.9219,
+      "step": 4850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.581849790089426e-05,
+      "loss": 1.0801,
+      "step": 4851
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.581659124580318e-05,
+      "loss": 1.0391,
+      "step": 4852
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.581468420688995e-05,
+      "loss": 1.0088,
+      "step": 4853
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.581277678420084e-05,
+      "loss": 0.7979,
+      "step": 4854
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.581086897778214e-05,
+      "loss": 1.0166,
+      "step": 4855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.580896078768015e-05,
+      "loss": 0.8799,
+      "step": 4856
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.580705221394117e-05,
+      "loss": 1.1025,
+      "step": 4857
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.580514325661152e-05,
+      "loss": 1.0303,
+      "step": 4858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.580323391573753e-05,
+      "loss": 0.9336,
+      "step": 4859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5801324191365523e-05,
+      "loss": 0.8857,
+      "step": 4860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.579941408354185e-05,
+      "loss": 0.9375,
+      "step": 4861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5797503592312866e-05,
+      "loss": 1.165,
+      "step": 4862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.579559271772492e-05,
+      "loss": 1.0146,
+      "step": 4863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5793681459824396e-05,
+      "loss": 1.0322,
+      "step": 4864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.579176981865767e-05,
+      "loss": 0.8457,
+      "step": 4865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.578985779427114e-05,
+      "loss": 0.9199,
+      "step": 4866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.578794538671119e-05,
+      "loss": 0.9521,
+      "step": 4867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5786032596024234e-05,
+      "loss": 0.917,
+      "step": 4868
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.578411942225669e-05,
+      "loss": 0.9209,
+      "step": 4869
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5782205865455e-05,
+      "loss": 1.0088,
+      "step": 4870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.578029192566557e-05,
+      "loss": 0.9658,
+      "step": 4871
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.577837760293488e-05,
+      "loss": 1.125,
+      "step": 4872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.577646289730936e-05,
+      "loss": 0.832,
+      "step": 4873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5774547808835485e-05,
+      "loss": 0.9756,
+      "step": 4874
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5772632337559724e-05,
+      "loss": 1.0166,
+      "step": 4875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5770716483528565e-05,
+      "loss": 1.0098,
+      "step": 4876
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5768800246788486e-05,
+      "loss": 1.001,
+      "step": 4877
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.576688362738601e-05,
+      "loss": 1.0381,
+      "step": 4878
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.576496662536763e-05,
+      "loss": 1.1455,
+      "step": 4879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.576304924077988e-05,
+      "loss": 1.0332,
+      "step": 4880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5761131473669285e-05,
+      "loss": 0.9185,
+      "step": 4881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.575921332408237e-05,
+      "loss": 0.9268,
+      "step": 4882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.57572947920657e-05,
+      "loss": 1.0234,
+      "step": 4883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.575537587766583e-05,
+      "loss": 0.877,
+      "step": 4884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5753456580929316e-05,
+      "loss": 1.0752,
+      "step": 4885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.575153690190274e-05,
+      "loss": 1.0156,
+      "step": 4886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.574961684063269e-05,
+      "loss": 1.0352,
+      "step": 4887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.574769639716576e-05,
+      "loss": 1.2031,
+      "step": 4888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.574577557154854e-05,
+      "loss": 1.0254,
+      "step": 4889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5743854363827664e-05,
+      "loss": 1.1562,
+      "step": 4890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.574193277404974e-05,
+      "loss": 1.0645,
+      "step": 4891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.57400108022614e-05,
+      "loss": 0.8877,
+      "step": 4892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5738088448509295e-05,
+      "loss": 1.0986,
+      "step": 4893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.573616571284006e-05,
+      "loss": 1.0898,
+      "step": 4894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.573424259530037e-05,
+      "loss": 1.085,
+      "step": 4895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.573231909593688e-05,
+      "loss": 1.0273,
+      "step": 4896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5730395214796265e-05,
+      "loss": 0.9844,
+      "step": 4897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.572847095192523e-05,
+      "loss": 0.8906,
+      "step": 4898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.572654630737046e-05,
+      "loss": 1.0107,
+      "step": 4899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5724621281178656e-05,
+      "loss": 1.0566,
+      "step": 4900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.572269587339654e-05,
+      "loss": 1.0068,
+      "step": 4901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5720770084070835e-05,
+      "loss": 1.0996,
+      "step": 4902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.571884391324827e-05,
+      "loss": 1.0,
+      "step": 4903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5716917360975596e-05,
+      "loss": 1.04,
+      "step": 4904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.571499042729956e-05,
+      "loss": 0.9727,
+      "step": 4905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.571306311226691e-05,
+      "loss": 1.0732,
+      "step": 4906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.571113541592444e-05,
+      "loss": 0.8818,
+      "step": 4907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.570920733831892e-05,
+      "loss": 0.9902,
+      "step": 4908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.570727887949713e-05,
+      "loss": 0.9688,
+      "step": 4909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.570535003950588e-05,
+      "loss": 0.8096,
+      "step": 4910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5703420818391974e-05,
+      "loss": 0.8389,
+      "step": 4911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.570149121620222e-05,
+      "loss": 1.0137,
+      "step": 4912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.569956123298345e-05,
+      "loss": 0.9365,
+      "step": 4913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.569763086878251e-05,
+      "loss": 1.0225,
+      "step": 4914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.569570012364622e-05,
+      "loss": 1.165,
+      "step": 4915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.569376899762146e-05,
+      "loss": 0.8613,
+      "step": 4916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.569183749075508e-05,
+      "loss": 0.9912,
+      "step": 4917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.568990560309394e-05,
+      "loss": 0.9141,
+      "step": 4918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.568797333468494e-05,
+      "loss": 0.9688,
+      "step": 4919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.568604068557496e-05,
+      "loss": 1.043,
+      "step": 4920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5684107655810914e-05,
+      "loss": 1.1045,
+      "step": 4921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.56821742454397e-05,
+      "loss": 0.7666,
+      "step": 4922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.568024045450823e-05,
+      "loss": 1.1221,
+      "step": 4923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.567830628306344e-05,
+      "loss": 1.0039,
+      "step": 4924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.567637173115226e-05,
+      "loss": 0.8936,
+      "step": 4925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5674436798821645e-05,
+      "loss": 1.002,
+      "step": 4926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.567250148611854e-05,
+      "loss": 1.0127,
+      "step": 4927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5670565793089926e-05,
+      "loss": 1.04,
+      "step": 4928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.566862971978276e-05,
+      "loss": 0.9443,
+      "step": 4929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.566669326624403e-05,
+      "loss": 0.8516,
+      "step": 4930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.566475643252073e-05,
+      "loss": 1.0889,
+      "step": 4931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.566281921865985e-05,
+      "loss": 1.085,
+      "step": 4932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.566088162470843e-05,
+      "loss": 0.9961,
+      "step": 4933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.565894365071346e-05,
+      "loss": 1.0938,
+      "step": 4934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.565700529672198e-05,
+      "loss": 1.041,
+      "step": 4935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.565506656278102e-05,
+      "loss": 0.9756,
+      "step": 4936
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.565312744893764e-05,
+      "loss": 1.083,
+      "step": 4937
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5651187955238885e-05,
+      "loss": 0.8857,
+      "step": 4938
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.564924808173184e-05,
+      "loss": 1.0498,
+      "step": 4939
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.564730782846356e-05,
+      "loss": 1.0605,
+      "step": 4940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5645367195481136e-05,
+      "loss": 0.9375,
+      "step": 4941
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.564342618283166e-05,
+      "loss": 0.9727,
+      "step": 4942
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.564148479056224e-05,
+      "loss": 1.0527,
+      "step": 4943
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.563954301871998e-05,
+      "loss": 1.0146,
+      "step": 4944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5637600867352e-05,
+      "loss": 1.0293,
+      "step": 4945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.563565833650544e-05,
+      "loss": 0.9307,
+      "step": 4946
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.563371542622744e-05,
+      "loss": 0.9561,
+      "step": 4947
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.563177213656514e-05,
+      "loss": 1.0781,
+      "step": 4948
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5629828467565696e-05,
+      "loss": 1.1582,
+      "step": 4949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5627884419276296e-05,
+      "loss": 0.9521,
+      "step": 4950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.562593999174409e-05,
+      "loss": 0.8691,
+      "step": 4951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.562399518501628e-05,
+      "loss": 1.0791,
+      "step": 4952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.562204999914005e-05,
+      "loss": 0.7671,
+      "step": 4953
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.562010443416261e-05,
+      "loss": 1.1777,
+      "step": 4954
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.561815849013117e-05,
+      "loss": 1.0645,
+      "step": 4955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5616212167092955e-05,
+      "loss": 1.0195,
+      "step": 4956
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5614265465095207e-05,
+      "loss": 1.0645,
+      "step": 4957
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.561231838418515e-05,
+      "loss": 1.0801,
+      "step": 4958
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5610370924410036e-05,
+      "loss": 0.8652,
+      "step": 4959
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5608423085817135e-05,
+      "loss": 1.1289,
+      "step": 4960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.56064748684537e-05,
+      "loss": 0.9307,
+      "step": 4961
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.560452627236703e-05,
+      "loss": 0.9434,
+      "step": 4962
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.560257729760438e-05,
+      "loss": 0.7754,
+      "step": 4963
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.560062794421308e-05,
+      "loss": 1.0137,
+      "step": 4964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.559867821224041e-05,
+      "loss": 0.957,
+      "step": 4965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5596728101733706e-05,
+      "loss": 0.9277,
+      "step": 4966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5594777612740263e-05,
+      "loss": 1.0127,
+      "step": 4967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.559282674530744e-05,
+      "loss": 0.8779,
+      "step": 4968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5590875499482566e-05,
+      "loss": 1.0791,
+      "step": 4969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5588923875313e-05,
+      "loss": 0.8457,
+      "step": 4970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.558697187284609e-05,
+      "loss": 1.1836,
+      "step": 4971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5585019492129216e-05,
+      "loss": 1.0693,
+      "step": 4972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.558306673320975e-05,
+      "loss": 1.0625,
+      "step": 4973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5581113596135086e-05,
+      "loss": 1.001,
+      "step": 4974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.557916008095261e-05,
+      "loss": 1.1211,
+      "step": 4975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5577206187709743e-05,
+      "loss": 0.8984,
+      "step": 4976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5575251916453895e-05,
+      "loss": 0.9609,
+      "step": 4977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.557329726723248e-05,
+      "loss": 1.0527,
+      "step": 4978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5571342240092943e-05,
+      "loss": 0.9443,
+      "step": 4979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.556938683508273e-05,
+      "loss": 0.9551,
+      "step": 4980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5567431052249274e-05,
+      "loss": 0.8926,
+      "step": 4981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.556547489164006e-05,
+      "loss": 0.7732,
+      "step": 4982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.556351835330255e-05,
+      "loss": 0.8975,
+      "step": 4983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.55615614372842e-05,
+      "loss": 0.8311,
+      "step": 4984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.555960414363254e-05,
+      "loss": 0.8291,
+      "step": 4985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.555764647239503e-05,
+      "loss": 0.7793,
+      "step": 4986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5555688423619205e-05,
+      "loss": 1.0127,
+      "step": 4987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.555372999735257e-05,
+      "loss": 0.9814,
+      "step": 4988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.555177119364264e-05,
+      "loss": 1.1309,
+      "step": 4989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5549812012536964e-05,
+      "loss": 1.0225,
+      "step": 4990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.554785245408308e-05,
+      "loss": 1.083,
+      "step": 4991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5545892518328544e-05,
+      "loss": 0.998,
+      "step": 4992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.554393220532092e-05,
+      "loss": 0.8154,
+      "step": 4993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5541971515107766e-05,
+      "loss": 0.9658,
+      "step": 4994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.554001044773667e-05,
+      "loss": 0.9736,
+      "step": 4995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5538049003255234e-05,
+      "loss": 0.9619,
+      "step": 4996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.553608718171103e-05,
+      "loss": 0.918,
+      "step": 4997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5534124983151685e-05,
+      "loss": 1.0439,
+      "step": 4998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5532162407624815e-05,
+      "loss": 1.042,
+      "step": 4999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.553019945517804e-05,
+      "loss": 1.1328,
+      "step": 5000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5528236125858994e-05,
+      "loss": 0.9854,
+      "step": 5001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.552627241971533e-05,
+      "loss": 0.9873,
+      "step": 5002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.552430833679469e-05,
+      "loss": 0.7363,
+      "step": 5003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5522343877144746e-05,
+      "loss": 0.8076,
+      "step": 5004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.552037904081316e-05,
+      "loss": 1.042,
+      "step": 5005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.551841382784762e-05,
+      "loss": 1.0469,
+      "step": 5006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5516448238295815e-05,
+      "loss": 1.1133,
+      "step": 5007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5514482272205446e-05,
+      "loss": 1.0879,
+      "step": 5008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.551251592962422e-05,
+      "loss": 1.0098,
+      "step": 5009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.551054921059985e-05,
+      "loss": 0.9619,
+      "step": 5010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.550858211518007e-05,
+      "loss": 0.9092,
+      "step": 5011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.55066146434126e-05,
+      "loss": 0.9775,
+      "step": 5012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5504646795345206e-05,
+      "loss": 1.0068,
+      "step": 5013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.550267857102562e-05,
+      "loss": 1.1182,
+      "step": 5014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5500709970501635e-05,
+      "loss": 1.1113,
+      "step": 5015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5498740993820995e-05,
+      "loss": 1.0234,
+      "step": 5016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.549677164103149e-05,
+      "loss": 0.9443,
+      "step": 5017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5494801912180915e-05,
+      "loss": 0.9521,
+      "step": 5018
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.549283180731706e-05,
+      "loss": 0.9561,
+      "step": 5019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.549086132648775e-05,
+      "loss": 0.9893,
+      "step": 5020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.548889046974078e-05,
+      "loss": 1.0273,
+      "step": 5021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5486919237124e-05,
+      "loss": 1.0303,
+      "step": 5022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.548494762868523e-05,
+      "loss": 0.9824,
+      "step": 5023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.548297564447233e-05,
+      "loss": 0.7861,
+      "step": 5024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5481003284533135e-05,
+      "loss": 0.9629,
+      "step": 5025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.547903054891552e-05,
+      "loss": 0.9082,
+      "step": 5026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5477057437667354e-05,
+      "loss": 0.9102,
+      "step": 5027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5475083950836525e-05,
+      "loss": 1.1074,
+      "step": 5028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5473110088470916e-05,
+      "loss": 0.9258,
+      "step": 5029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.547113585061843e-05,
+      "loss": 1.0117,
+      "step": 5030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.546916123732698e-05,
+      "loss": 1.0303,
+      "step": 5031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5467186248644475e-05,
+      "loss": 0.8525,
+      "step": 5032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.546521088461884e-05,
+      "loss": 0.7944,
+      "step": 5033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.546323514529804e-05,
+      "loss": 1.1816,
+      "step": 5034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.546125903072998e-05,
+      "loss": 0.8916,
+      "step": 5035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5459282540962645e-05,
+      "loss": 0.9629,
+      "step": 5036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5457305676043975e-05,
+      "loss": 0.9717,
+      "step": 5037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.545532843602196e-05,
+      "loss": 0.9121,
+      "step": 5038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.545335082094458e-05,
+      "loss": 1.0283,
+      "step": 5039
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.545137283085982e-05,
+      "loss": 0.9072,
+      "step": 5040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5449394465815676e-05,
+      "loss": 1.0391,
+      "step": 5041
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.544741572586017e-05,
+      "loss": 0.8799,
+      "step": 5042
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.544543661104132e-05,
+      "loss": 0.9219,
+      "step": 5043
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.544345712140714e-05,
+      "loss": 1.1914,
+      "step": 5044
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.544147725700567e-05,
+      "loss": 1.0693,
+      "step": 5045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.543949701788496e-05,
+      "loss": 1.0342,
+      "step": 5046
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5437516404093064e-05,
+      "loss": 0.6094,
+      "step": 5047
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.543553541567804e-05,
+      "loss": 1.0586,
+      "step": 5048
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.543355405268798e-05,
+      "loss": 0.916,
+      "step": 5049
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.543157231517094e-05,
+      "loss": 1.0986,
+      "step": 5050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5429590203175025e-05,
+      "loss": 1.0312,
+      "step": 5051
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.542760771674833e-05,
+      "loss": 0.9395,
+      "step": 5052
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.542562485593897e-05,
+      "loss": 0.8779,
+      "step": 5053
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5423641620795065e-05,
+      "loss": 0.9717,
+      "step": 5054
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.542165801136473e-05,
+      "loss": 1.0576,
+      "step": 5055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.541967402769611e-05,
+      "loss": 0.8994,
+      "step": 5056
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.541768966983734e-05,
+      "loss": 0.9775,
+      "step": 5057
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.54157049378366e-05,
+      "loss": 0.9521,
+      "step": 5058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.541371983174203e-05,
+      "loss": 0.9434,
+      "step": 5059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.541173435160181e-05,
+      "loss": 1.0342,
+      "step": 5060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.540974849746412e-05,
+      "loss": 1.0811,
+      "step": 5061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.540776226937715e-05,
+      "loss": 0.9434,
+      "step": 5062
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.540577566738911e-05,
+      "loss": 1.0771,
+      "step": 5063
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.54037886915482e-05,
+      "loss": 0.9443,
+      "step": 5064
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.540180134190263e-05,
+      "loss": 1.0791,
+      "step": 5065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.539981361850065e-05,
+      "loss": 1.0586,
+      "step": 5066
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.539782552139048e-05,
+      "loss": 0.9775,
+      "step": 5067
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5395837050620355e-05,
+      "loss": 0.8477,
+      "step": 5068
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5393848206238555e-05,
+      "loss": 0.8701,
+      "step": 5069
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5391858988293324e-05,
+      "loss": 0.8594,
+      "step": 5070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5389869396832944e-05,
+      "loss": 1.0459,
+      "step": 5071
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5387879431905694e-05,
+      "loss": 0.9023,
+      "step": 5072
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5385889093559864e-05,
+      "loss": 0.8984,
+      "step": 5073
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.538389838184375e-05,
+      "loss": 0.8066,
+      "step": 5074
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.538190729680567e-05,
+      "loss": 0.9688,
+      "step": 5075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.537991583849394e-05,
+      "loss": 0.958,
+      "step": 5076
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5377924006956876e-05,
+      "loss": 0.8872,
+      "step": 5077
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.537593180224282e-05,
+      "loss": 1.0732,
+      "step": 5078
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5373939224400126e-05,
+      "loss": 0.9707,
+      "step": 5079
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5371946273477135e-05,
+      "loss": 1.0488,
+      "step": 5080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.536995294952222e-05,
+      "loss": 0.8101,
+      "step": 5081
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.536795925258374e-05,
+      "loss": 0.9062,
+      "step": 5082
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5365965182710086e-05,
+      "loss": 1.0205,
+      "step": 5083
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.536397073994965e-05,
+      "loss": 1.0117,
+      "step": 5084
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.536197592435083e-05,
+      "loss": 1.1133,
+      "step": 5085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.535998073596203e-05,
+      "loss": 0.9502,
+      "step": 5086
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.535798517483167e-05,
+      "loss": 1.1641,
+      "step": 5087
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.535598924100818e-05,
+      "loss": 1.0312,
+      "step": 5088
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.535399293453998e-05,
+      "loss": 1.0,
+      "step": 5089
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.535199625547554e-05,
+      "loss": 1.1191,
+      "step": 5090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.534999920386329e-05,
+      "loss": 1.0068,
+      "step": 5091
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.534800177975171e-05,
+      "loss": 0.9873,
+      "step": 5092
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.534600398318925e-05,
+      "loss": 0.9688,
+      "step": 5093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5344005814224406e-05,
+      "loss": 1.0742,
+      "step": 5094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.534200727290568e-05,
+      "loss": 1.168,
+      "step": 5095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5340008359281546e-05,
+      "loss": 0.9307,
+      "step": 5096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5338009073400525e-05,
+      "loss": 0.8262,
+      "step": 5097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5336009415311136e-05,
+      "loss": 0.9365,
+      "step": 5098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.533400938506189e-05,
+      "loss": 0.8916,
+      "step": 5099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.533200898270134e-05,
+      "loss": 0.7173,
+      "step": 5100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.533000820827802e-05,
+      "loss": 0.8975,
+      "step": 5101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.532800706184048e-05,
+      "loss": 1.0596,
+      "step": 5102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.53260055434373e-05,
+      "loss": 1.0254,
+      "step": 5103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.532400365311702e-05,
+      "loss": 1.0518,
+      "step": 5104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.532200139092825e-05,
+      "loss": 1.1035,
+      "step": 5105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.531999875691957e-05,
+      "loss": 0.959,
+      "step": 5106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5317995751139564e-05,
+      "loss": 0.958,
+      "step": 5107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5315992373636855e-05,
+      "loss": 1.0664,
+      "step": 5108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5313988624460046e-05,
+      "loss": 1.0557,
+      "step": 5109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.531198450365778e-05,
+      "loss": 1.1348,
+      "step": 5110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5309980011278675e-05,
+      "loss": 1.0557,
+      "step": 5111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.530797514737138e-05,
+      "loss": 0.9482,
+      "step": 5112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.530596991198455e-05,
+      "loss": 0.9229,
+      "step": 5113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.530396430516684e-05,
+      "loss": 1.084,
+      "step": 5114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.530195832696693e-05,
+      "loss": 0.9785,
+      "step": 5115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.529995197743349e-05,
+      "loss": 1.0459,
+      "step": 5116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5297945256615214e-05,
+      "loss": 1.1182,
+      "step": 5117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5295938164560795e-05,
+      "loss": 0.9873,
+      "step": 5118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.529393070131893e-05,
+      "loss": 0.8672,
+      "step": 5119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.529192286693836e-05,
+      "loss": 0.8633,
+      "step": 5120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.528991466146778e-05,
+      "loss": 1.0566,
+      "step": 5121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.528790608495594e-05,
+      "loss": 0.9219,
+      "step": 5122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.528589713745159e-05,
+      "loss": 1.0,
+      "step": 5123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.528388781900347e-05,
+      "loss": 0.9414,
+      "step": 5124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.528187812966033e-05,
+      "loss": 1.123,
+      "step": 5125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.527986806947095e-05,
+      "loss": 1.1104,
+      "step": 5126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5277857638484124e-05,
+      "loss": 1.0508,
+      "step": 5127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.527584683674861e-05,
+      "loss": 1.1309,
+      "step": 5128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.527383566431322e-05,
+      "loss": 0.9385,
+      "step": 5129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5271824121226755e-05,
+      "loss": 1.085,
+      "step": 5130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.526981220753804e-05,
+      "loss": 1.002,
+      "step": 5131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.526779992329588e-05,
+      "loss": 0.9717,
+      "step": 5132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.526578726854912e-05,
+      "loss": 0.9062,
+      "step": 5133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5263774243346604e-05,
+      "loss": 1.1035,
+      "step": 5134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.526176084773717e-05,
+      "loss": 1.0166,
+      "step": 5135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.525974708176969e-05,
+      "loss": 1.0762,
+      "step": 5136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.525773294549302e-05,
+      "loss": 0.7866,
+      "step": 5137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.525571843895604e-05,
+      "loss": 1.0947,
+      "step": 5138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.525370356220764e-05,
+      "loss": 1.1455,
+      "step": 5139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.525168831529672e-05,
+      "loss": 1.0303,
+      "step": 5140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5249672698272175e-05,
+      "loss": 1.1914,
+      "step": 5141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.524765671118292e-05,
+      "loss": 0.9561,
+      "step": 5142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.524564035407788e-05,
+      "loss": 0.813,
+      "step": 5143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.524362362700598e-05,
+      "loss": 0.8965,
+      "step": 5144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.524160653001617e-05,
+      "loss": 1.0625,
+      "step": 5145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5239589063157396e-05,
+      "loss": 0.9023,
+      "step": 5146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.523757122647861e-05,
+      "loss": 1.0244,
+      "step": 5147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.523555302002878e-05,
+      "loss": 0.9443,
+      "step": 5148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5233534443856886e-05,
+      "loss": 0.8633,
+      "step": 5149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.523151549801191e-05,
+      "loss": 1.0127,
+      "step": 5150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.522949618254285e-05,
+      "loss": 0.958,
+      "step": 5151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.52274764974987e-05,
+      "loss": 0.7563,
+      "step": 5152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5225456442928486e-05,
+      "loss": 0.9131,
+      "step": 5153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.522343601888121e-05,
+      "loss": 0.7656,
+      "step": 5154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5221415225405925e-05,
+      "loss": 1.1484,
+      "step": 5155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.521939406255165e-05,
+      "loss": 1.0127,
+      "step": 5156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.521737253036744e-05,
+      "loss": 1.0254,
+      "step": 5157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.521535062890236e-05,
+      "loss": 0.8369,
+      "step": 5158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.521332835820545e-05,
+      "loss": 0.9453,
+      "step": 5159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5211305718325816e-05,
+      "loss": 0.9561,
+      "step": 5160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.520928270931252e-05,
+      "loss": 0.9023,
+      "step": 5161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.520725933121466e-05,
+      "loss": 1.0635,
+      "step": 5162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.520523558408134e-05,
+      "loss": 0.9062,
+      "step": 5163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5203211467961673e-05,
+      "loss": 0.9834,
+      "step": 5164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.520118698290478e-05,
+      "loss": 0.9766,
+      "step": 5165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.519916212895977e-05,
+      "loss": 1.0332,
+      "step": 5166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5197136906175804e-05,
+      "loss": 1.0029,
+      "step": 5167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.519511131460202e-05,
+      "loss": 1.0225,
+      "step": 5168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.519308535428756e-05,
+      "loss": 0.834,
+      "step": 5169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.519105902528161e-05,
+      "loss": 0.7827,
+      "step": 5170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.518903232763333e-05,
+      "loss": 0.9736,
+      "step": 5171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.518700526139191e-05,
+      "loss": 0.8887,
+      "step": 5172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.518497782660653e-05,
+      "loss": 0.8418,
+      "step": 5173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5182950023326395e-05,
+      "loss": 1.0127,
+      "step": 5174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5180921851600716e-05,
+      "loss": 0.9131,
+      "step": 5175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5178893311478704e-05,
+      "loss": 1.0479,
+      "step": 5176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5176864403009596e-05,
+      "loss": 1.0078,
+      "step": 5177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.517483512624263e-05,
+      "loss": 1.0332,
+      "step": 5178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.517280548122703e-05,
+      "loss": 0.9121,
+      "step": 5179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.517077546801207e-05,
+      "loss": 1.0146,
+      "step": 5180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5168745086647e-05,
+      "loss": 0.9932,
+      "step": 5181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5166714337181104e-05,
+      "loss": 0.9902,
+      "step": 5182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.516468321966365e-05,
+      "loss": 1.1455,
+      "step": 5183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5162651734143934e-05,
+      "loss": 1.0537,
+      "step": 5184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.516061988067124e-05,
+      "loss": 0.9131,
+      "step": 5185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.515858765929491e-05,
+      "loss": 1.0586,
+      "step": 5186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.515655507006422e-05,
+      "loss": 0.96,
+      "step": 5187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5154522113028514e-05,
+      "loss": 0.9814,
+      "step": 5188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5152488788237124e-05,
+      "loss": 1.0791,
+      "step": 5189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.51504550957394e-05,
+      "loss": 1.0186,
+      "step": 5190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.514842103558468e-05,
+      "loss": 1.0996,
+      "step": 5191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.514638660782233e-05,
+      "loss": 0.7939,
+      "step": 5192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.514435181250173e-05,
+      "loss": 1.0498,
+      "step": 5193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5142316649672246e-05,
+      "loss": 1.0928,
+      "step": 5194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.514028111938327e-05,
+      "loss": 0.9961,
+      "step": 5195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5138245221684194e-05,
+      "loss": 0.9414,
+      "step": 5196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.513620895662443e-05,
+      "loss": 0.9062,
+      "step": 5197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.513417232425338e-05,
+      "loss": 1.0117,
+      "step": 5198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.513213532462048e-05,
+      "loss": 0.9746,
+      "step": 5199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.513009795777516e-05,
+      "loss": 0.876,
+      "step": 5200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.512806022376686e-05,
+      "loss": 0.8984,
+      "step": 5201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.512602212264503e-05,
+      "loss": 1.0508,
+      "step": 5202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.512398365445912e-05,
+      "loss": 1.1592,
+      "step": 5203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.51219448192586e-05,
+      "loss": 0.9775,
+      "step": 5204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.511990561709296e-05,
+      "loss": 0.8652,
+      "step": 5205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.511786604801168e-05,
+      "loss": 0.9951,
+      "step": 5206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.511582611206425e-05,
+      "loss": 0.999,
+      "step": 5207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.511378580930017e-05,
+      "loss": 1.0127,
+      "step": 5208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5111745139768955e-05,
+      "loss": 1.0674,
+      "step": 5209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5109704103520136e-05,
+      "loss": 0.9014,
+      "step": 5210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.510766270060322e-05,
+      "loss": 0.7979,
+      "step": 5211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.510562093106778e-05,
+      "loss": 0.9688,
+      "step": 5212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.510357879496333e-05,
+      "loss": 0.96,
+      "step": 5213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5101536292339443e-05,
+      "loss": 1.1133,
+      "step": 5214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.509949342324568e-05,
+      "loss": 1.1631,
+      "step": 5215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.509745018773162e-05,
+      "loss": 1.0918,
+      "step": 5216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.509540658584684e-05,
+      "loss": 1.0479,
+      "step": 5217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.509336261764094e-05,
+      "loss": 1.0117,
+      "step": 5218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5091318283163524e-05,
+      "loss": 0.8682,
+      "step": 5219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.508927358246419e-05,
+      "loss": 0.9639,
+      "step": 5220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.508722851559256e-05,
+      "loss": 0.9551,
+      "step": 5221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.508518308259826e-05,
+      "loss": 0.8613,
+      "step": 5222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.508313728353094e-05,
+      "loss": 0.9834,
+      "step": 5223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.508109111844023e-05,
+      "loss": 0.9922,
+      "step": 5224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.50790445873758e-05,
+      "loss": 0.9189,
+      "step": 5225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.507699769038729e-05,
+      "loss": 1.0625,
+      "step": 5226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.507495042752439e-05,
+      "loss": 1.1035,
+      "step": 5227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.507290279883678e-05,
+      "loss": 1.0068,
+      "step": 5228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.507085480437414e-05,
+      "loss": 1.1045,
+      "step": 5229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.506880644418619e-05,
+      "loss": 0.8096,
+      "step": 5230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.506675771832261e-05,
+      "loss": 0.7998,
+      "step": 5231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.506470862683314e-05,
+      "loss": 1.0137,
+      "step": 5232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.506265916976749e-05,
+      "loss": 1.0312,
+      "step": 5233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.50606093471754e-05,
+      "loss": 0.8682,
+      "step": 5234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5058559159106606e-05,
+      "loss": 0.9346,
+      "step": 5235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.505650860561088e-05,
+      "loss": 1.0098,
+      "step": 5236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.505445768673796e-05,
+      "loss": 1.1289,
+      "step": 5237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.505240640253763e-05,
+      "loss": 1.0078,
+      "step": 5238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5050354753059656e-05,
+      "loss": 0.9951,
+      "step": 5239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.504830273835384e-05,
+      "loss": 0.9932,
+      "step": 5240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5046250358469974e-05,
+      "loss": 0.9346,
+      "step": 5241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.504419761345786e-05,
+      "loss": 1.1279,
+      "step": 5242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5042144503367315e-05,
+      "loss": 0.9756,
+      "step": 5243
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5040091028248155e-05,
+      "loss": 1.0518,
+      "step": 5244
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.503803718815022e-05,
+      "loss": 1.082,
+      "step": 5245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.503598298312335e-05,
+      "loss": 1.0391,
+      "step": 5246
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.503392841321739e-05,
+      "loss": 0.8306,
+      "step": 5247
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.50318734784822e-05,
+      "loss": 0.8877,
+      "step": 5248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.502981817896765e-05,
+      "loss": 0.9434,
+      "step": 5249
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.502776251472361e-05,
+      "loss": 1.0938,
+      "step": 5250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5025706485799974e-05,
+      "loss": 1.0117,
+      "step": 5251
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.502365009224662e-05,
+      "loss": 1.0107,
+      "step": 5252
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.502159333411347e-05,
+      "loss": 1.0391,
+      "step": 5253
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.501953621145043e-05,
+      "loss": 0.9668,
+      "step": 5254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5017478724307415e-05,
+      "loss": 1.0732,
+      "step": 5255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.501542087273436e-05,
+      "loss": 0.9414,
+      "step": 5256
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5013362656781184e-05,
+      "loss": 0.8818,
+      "step": 5257
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.501130407649786e-05,
+      "loss": 1.0547,
+      "step": 5258
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.500924513193433e-05,
+      "loss": 1.1074,
+      "step": 5259
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5007185823140565e-05,
+      "loss": 0.8408,
+      "step": 5260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.500512615016653e-05,
+      "loss": 0.9111,
+      "step": 5261
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.500306611306221e-05,
+      "loss": 0.9082,
+      "step": 5262
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5001005711877605e-05,
+      "loss": 1.0381,
+      "step": 5263
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4998944946662706e-05,
+      "loss": 1.0391,
+      "step": 5264
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.499688381746752e-05,
+      "loss": 1.1133,
+      "step": 5265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.499482232434207e-05,
+      "loss": 0.9834,
+      "step": 5266
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.499276046733638e-05,
+      "loss": 1.0332,
+      "step": 5267
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4990698246500484e-05,
+      "loss": 1.0264,
+      "step": 5268
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.498863566188443e-05,
+      "loss": 1.0508,
+      "step": 5269
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.498657271353827e-05,
+      "loss": 0.9023,
+      "step": 5270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4984509401512064e-05,
+      "loss": 1.0527,
+      "step": 5271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.498244572585588e-05,
+      "loss": 0.9824,
+      "step": 5272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.49803816866198e-05,
+      "loss": 0.9707,
+      "step": 5273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.497831728385391e-05,
+      "loss": 1.1465,
+      "step": 5274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.497625251760832e-05,
+      "loss": 0.9561,
+      "step": 5275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4974187387933106e-05,
+      "loss": 0.9229,
+      "step": 5276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.497212189487842e-05,
+      "loss": 1.0752,
+      "step": 5277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.497005603849435e-05,
+      "loss": 1.0898,
+      "step": 5278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.496798981883106e-05,
+      "loss": 0.9336,
+      "step": 5279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4965923235938676e-05,
+      "loss": 0.9863,
+      "step": 5280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.496385628986734e-05,
+      "loss": 0.8994,
+      "step": 5281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.496178898066722e-05,
+      "loss": 0.8672,
+      "step": 5282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.495972130838849e-05,
+      "loss": 0.9395,
+      "step": 5283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4957653273081305e-05,
+      "loss": 1.0098,
+      "step": 5284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.495558487479588e-05,
+      "loss": 0.7686,
+      "step": 5285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.495351611358238e-05,
+      "loss": 0.792,
+      "step": 5286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.495144698949103e-05,
+      "loss": 1.0049,
+      "step": 5287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.494937750257203e-05,
+      "loss": 0.8896,
+      "step": 5288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.49473076528756e-05,
+      "loss": 0.9258,
+      "step": 5289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4945237440451976e-05,
+      "loss": 1.1221,
+      "step": 5290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4943166865351385e-05,
+      "loss": 1.0039,
+      "step": 5291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.494109592762408e-05,
+      "loss": 1.0264,
+      "step": 5292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.493902462732032e-05,
+      "loss": 1.0186,
+      "step": 5293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.493695296449037e-05,
+      "loss": 0.9824,
+      "step": 5294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4934880939184496e-05,
+      "loss": 1.0205,
+      "step": 5295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.493280855145298e-05,
+      "loss": 1.0371,
+      "step": 5296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.493073580134613e-05,
+      "loss": 1.0127,
+      "step": 5297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4928662688914216e-05,
+      "loss": 0.8105,
+      "step": 5298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4926589214207573e-05,
+      "loss": 0.9258,
+      "step": 5299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.49245153772765e-05,
+      "loss": 0.8496,
+      "step": 5300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4922441178171336e-05,
+      "loss": 0.9502,
+      "step": 5301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.49203666169424e-05,
+      "loss": 1.0332,
+      "step": 5302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.491829169364006e-05,
+      "loss": 0.959,
+      "step": 5303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.491621640831464e-05,
+      "loss": 0.9844,
+      "step": 5304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.491414076101652e-05,
+      "loss": 1.0381,
+      "step": 5305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.491206475179607e-05,
+      "loss": 0.9492,
+      "step": 5306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.490998838070366e-05,
+      "loss": 1.0068,
+      "step": 5307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.490791164778967e-05,
+      "loss": 0.8955,
+      "step": 5308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4905834553104525e-05,
+      "loss": 1.0225,
+      "step": 5309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.49037570966986e-05,
+      "loss": 0.9785,
+      "step": 5310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.490167927862233e-05,
+      "loss": 1.1016,
+      "step": 5311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.489960109892612e-05,
+      "loss": 0.7207,
+      "step": 5312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.489752255766041e-05,
+      "loss": 1.1064,
+      "step": 5313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.489544365487564e-05,
+      "loss": 0.9648,
+      "step": 5314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4893364390622256e-05,
+      "loss": 1.0322,
+      "step": 5315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4891284764950724e-05,
+      "loss": 1.0469,
+      "step": 5316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4889204777911503e-05,
+      "loss": 0.7778,
+      "step": 5317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4887124429555065e-05,
+      "loss": 0.8857,
+      "step": 5318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.488504371993191e-05,
+      "loss": 1.002,
+      "step": 5319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.488296264909251e-05,
+      "loss": 0.9258,
+      "step": 5320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4880881217087377e-05,
+      "loss": 1.0918,
+      "step": 5321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.487879942396702e-05,
+      "loss": 0.999,
+      "step": 5322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4876717269781956e-05,
+      "loss": 1.0176,
+      "step": 5323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4874634754582716e-05,
+      "loss": 1.0195,
+      "step": 5324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4872551878419836e-05,
+      "loss": 1.0498,
+      "step": 5325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.487046864134386e-05,
+      "loss": 0.8369,
+      "step": 5326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.486838504340535e-05,
+      "loss": 1.1641,
+      "step": 5327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.486630108465485e-05,
+      "loss": 0.9912,
+      "step": 5328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.486421676514295e-05,
+      "loss": 1.0264,
+      "step": 5329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.486213208492022e-05,
+      "loss": 0.918,
+      "step": 5330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.486004704403725e-05,
+      "loss": 1.0557,
+      "step": 5331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.485796164254464e-05,
+      "loss": 0.7891,
+      "step": 5332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4855875880493e-05,
+      "loss": 0.6582,
+      "step": 5333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.485378975793294e-05,
+      "loss": 0.8818,
+      "step": 5334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.485170327491509e-05,
+      "loss": 1.0654,
+      "step": 5335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.484961643149007e-05,
+      "loss": 1.0098,
+      "step": 5336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.484752922770854e-05,
+      "loss": 0.9893,
+      "step": 5337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4845441663621145e-05,
+      "loss": 1.0361,
+      "step": 5338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.484335373927853e-05,
+      "loss": 0.7954,
+      "step": 5339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.484126545473137e-05,
+      "loss": 0.9912,
+      "step": 5340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.483917681003035e-05,
+      "loss": 0.9443,
+      "step": 5341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.483708780522615e-05,
+      "loss": 1.0322,
+      "step": 5342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.483499844036946e-05,
+      "loss": 0.9609,
+      "step": 5343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.483290871551098e-05,
+      "loss": 0.8877,
+      "step": 5344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.483081863070144e-05,
+      "loss": 1.2012,
+      "step": 5345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4828728185991546e-05,
+      "loss": 0.9746,
+      "step": 5346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.482663738143202e-05,
+      "loss": 0.999,
+      "step": 5347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.482454621707361e-05,
+      "loss": 1.0967,
+      "step": 5348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.482245469296707e-05,
+      "loss": 0.8848,
+      "step": 5349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.482036280916313e-05,
+      "loss": 0.9688,
+      "step": 5350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4818270565712593e-05,
+      "loss": 1.0352,
+      "step": 5351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.481617796266619e-05,
+      "loss": 1.0215,
+      "step": 5352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4814085000074723e-05,
+      "loss": 0.9365,
+      "step": 5353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.481199167798899e-05,
+      "loss": 0.7402,
+      "step": 5354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.480989799645976e-05,
+      "loss": 0.8516,
+      "step": 5355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.480780395553788e-05,
+      "loss": 1.0938,
+      "step": 5356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.480570955527414e-05,
+      "loss": 1.0234,
+      "step": 5357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4803614795719364e-05,
+      "loss": 0.9307,
+      "step": 5358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.48015196769244e-05,
+      "loss": 1.0781,
+      "step": 5359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4799424198940084e-05,
+      "loss": 0.9487,
+      "step": 5360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.479732836181726e-05,
+      "loss": 0.9209,
+      "step": 5361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4795232165606796e-05,
+      "loss": 0.8184,
+      "step": 5362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.479313561035956e-05,
+      "loss": 1.0332,
+      "step": 5363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.479103869612643e-05,
+      "loss": 1.2852,
+      "step": 5364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.478894142295828e-05,
+      "loss": 0.9365,
+      "step": 5365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4786843790906025e-05,
+      "loss": 0.9951,
+      "step": 5366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4784745800020554e-05,
+      "loss": 1.0615,
+      "step": 5367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4782647450352785e-05,
+      "loss": 1.082,
+      "step": 5368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.478054874195363e-05,
+      "loss": 1.0615,
+      "step": 5369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4778449674874036e-05,
+      "loss": 1.0508,
+      "step": 5370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4776350249164924e-05,
+      "loss": 1.0479,
+      "step": 5371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.477425046487724e-05,
+      "loss": 0.7432,
+      "step": 5372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.477215032206196e-05,
+      "loss": 1.0303,
+      "step": 5373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.477004982077003e-05,
+      "loss": 0.8926,
+      "step": 5374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4767948961052426e-05,
+      "loss": 0.9912,
+      "step": 5375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.476584774296013e-05,
+      "loss": 0.8535,
+      "step": 5376
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.476374616654414e-05,
+      "loss": 0.9678,
+      "step": 5377
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4761644231855445e-05,
+      "loss": 0.9814,
+      "step": 5378
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.475954193894506e-05,
+      "loss": 1.0264,
+      "step": 5379
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4757439287864e-05,
+      "loss": 1.0254,
+      "step": 5380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4755336278663276e-05,
+      "loss": 1.0303,
+      "step": 5381
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4753232911393945e-05,
+      "loss": 0.9492,
+      "step": 5382
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4751129186107036e-05,
+      "loss": 0.9414,
+      "step": 5383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.474902510285361e-05,
+      "loss": 0.9385,
+      "step": 5384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.474692066168471e-05,
+      "loss": 0.9502,
+      "step": 5385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4744815862651416e-05,
+      "loss": 1.0557,
+      "step": 5386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.474271070580481e-05,
+      "loss": 0.9189,
+      "step": 5387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.474060519119597e-05,
+      "loss": 0.959,
+      "step": 5388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.473849931887599e-05,
+      "loss": 1.1816,
+      "step": 5389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4736393088895975e-05,
+      "loss": 1.1748,
+      "step": 5390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.473428650130704e-05,
+      "loss": 0.9746,
+      "step": 5391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4732179556160304e-05,
+      "loss": 0.7554,
+      "step": 5392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4730072253506896e-05,
+      "loss": 0.5815,
+      "step": 5393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4727964593397956e-05,
+      "loss": 0.875,
+      "step": 5394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.472585657588463e-05,
+      "loss": 1.0117,
+      "step": 5395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.472374820101807e-05,
+      "loss": 0.9912,
+      "step": 5396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4721639468849436e-05,
+      "loss": 1.2617,
+      "step": 5397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4719530379429914e-05,
+      "loss": 0.876,
+      "step": 5398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4717420932810675e-05,
+      "loss": 1.0312,
+      "step": 5399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4715311129042914e-05,
+      "loss": 1.1094,
+      "step": 5400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.471320096817783e-05,
+      "loss": 0.8389,
+      "step": 5401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.471109045026662e-05,
+      "loss": 0.7705,
+      "step": 5402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.470897957536052e-05,
+      "loss": 0.8389,
+      "step": 5403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.470686834351073e-05,
+      "loss": 1.0068,
+      "step": 5404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.470475675476851e-05,
+      "loss": 1.1396,
+      "step": 5405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.470264480918508e-05,
+      "loss": 0.918,
+      "step": 5406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.470053250681171e-05,
+      "loss": 0.9209,
+      "step": 5407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4698419847699635e-05,
+      "loss": 1.1045,
+      "step": 5408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.469630683190014e-05,
+      "loss": 0.9619,
+      "step": 5409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.46941934594645e-05,
+      "loss": 0.7949,
+      "step": 5410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4692079730444e-05,
+      "loss": 0.9707,
+      "step": 5411
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4689965644889936e-05,
+      "loss": 0.9561,
+      "step": 5412
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.46878512028536e-05,
+      "loss": 1.0088,
+      "step": 5413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.468573640438632e-05,
+      "loss": 1.1914,
+      "step": 5414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.46836212495394e-05,
+      "loss": 0.7383,
+      "step": 5415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4681505738364174e-05,
+      "loss": 0.9951,
+      "step": 5416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.467938987091199e-05,
+      "loss": 0.9053,
+      "step": 5417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.467727364723417e-05,
+      "loss": 1.127,
+      "step": 5418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.467515706738209e-05,
+      "loss": 0.8223,
+      "step": 5419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4673040131407114e-05,
+      "loss": 0.9297,
+      "step": 5420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.46709228393606e-05,
+      "loss": 1.1875,
+      "step": 5421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.466880519129393e-05,
+      "loss": 0.8926,
+      "step": 5422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.46666871872585e-05,
+      "loss": 0.875,
+      "step": 5423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4664568827305705e-05,
+      "loss": 0.8916,
+      "step": 5424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4662450111486956e-05,
+      "loss": 1.1094,
+      "step": 5425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.466033103985366e-05,
+      "loss": 1.0654,
+      "step": 5426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.465821161245724e-05,
+      "loss": 0.9688,
+      "step": 5427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.465609182934913e-05,
+      "loss": 0.9111,
+      "step": 5428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.465397169058078e-05,
+      "loss": 0.9785,
+      "step": 5429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4651851196203634e-05,
+      "loss": 0.9355,
+      "step": 5430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.464973034626915e-05,
+      "loss": 0.833,
+      "step": 5431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.464760914082879e-05,
+      "loss": 1.0205,
+      "step": 5432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.464548757993403e-05,
+      "loss": 1.0303,
+      "step": 5433
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.464336566363637e-05,
+      "loss": 0.8818,
+      "step": 5434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.464124339198728e-05,
+      "loss": 1.041,
+      "step": 5435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4639120765038274e-05,
+      "loss": 1.0117,
+      "step": 5436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.463699778284085e-05,
+      "loss": 0.9893,
+      "step": 5437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.463487444544655e-05,
+      "loss": 0.917,
+      "step": 5438
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.463275075290688e-05,
+      "loss": 0.9346,
+      "step": 5439
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.463062670527338e-05,
+      "loss": 0.8828,
+      "step": 5440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.46285023025976e-05,
+      "loss": 1.0469,
+      "step": 5441
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.462637754493109e-05,
+      "loss": 0.915,
+      "step": 5442
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4624252432325415e-05,
+      "loss": 1.0088,
+      "step": 5443
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4622126964832136e-05,
+      "loss": 0.9512,
+      "step": 5444
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.462000114250284e-05,
+      "loss": 0.8691,
+      "step": 5445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.461787496538912e-05,
+      "loss": 1.0332,
+      "step": 5446
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.461574843354255e-05,
+      "loss": 0.9697,
+      "step": 5447
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.461362154701475e-05,
+      "loss": 0.835,
+      "step": 5448
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4611494305857345e-05,
+      "loss": 1.0996,
+      "step": 5449
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.460936671012193e-05,
+      "loss": 0.8867,
+      "step": 5450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4607238759860165e-05,
+      "loss": 1.0176,
+      "step": 5451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.460511045512366e-05,
+      "loss": 1.1221,
+      "step": 5452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.460298179596408e-05,
+      "loss": 0.9414,
+      "step": 5453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.460085278243308e-05,
+      "loss": 0.9111,
+      "step": 5454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.459872341458232e-05,
+      "loss": 0.9023,
+      "step": 5455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4596593692463476e-05,
+      "loss": 1.0879,
+      "step": 5456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4594463616128236e-05,
+      "loss": 0.7998,
+      "step": 5457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.459233318562828e-05,
+      "loss": 0.8232,
+      "step": 5458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.459020240101531e-05,
+      "loss": 0.9541,
+      "step": 5459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.458807126234104e-05,
+      "loss": 1.0498,
+      "step": 5460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.458593976965718e-05,
+      "loss": 0.9131,
+      "step": 5461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.458380792301546e-05,
+      "loss": 0.9443,
+      "step": 5462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.458167572246761e-05,
+      "loss": 0.9766,
+      "step": 5463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.457954316806537e-05,
+      "loss": 0.9199,
+      "step": 5464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.45774102598605e-05,
+      "loss": 1.042,
+      "step": 5465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.457527699790475e-05,
+      "loss": 0.875,
+      "step": 5466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.457314338224989e-05,
+      "loss": 0.915,
+      "step": 5467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.45710094129477e-05,
+      "loss": 1.1279,
+      "step": 5468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.456887509004997e-05,
+      "loss": 1.0488,
+      "step": 5469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.456674041360848e-05,
+      "loss": 0.9004,
+      "step": 5470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.456460538367504e-05,
+      "loss": 0.833,
+      "step": 5471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.456247000030146e-05,
+      "loss": 1.1201,
+      "step": 5472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.456033426353956e-05,
+      "loss": 0.9619,
+      "step": 5473
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.455819817344117e-05,
+      "loss": 1.0566,
+      "step": 5474
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4556061730058123e-05,
+      "loss": 1.1367,
+      "step": 5475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.455392493344226e-05,
+      "loss": 0.7334,
+      "step": 5476
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.455178778364545e-05,
+      "loss": 0.8496,
+      "step": 5477
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4549650280719535e-05,
+      "loss": 0.9609,
+      "step": 5478
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.454751242471641e-05,
+      "loss": 1.0371,
+      "step": 5479
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.454537421568793e-05,
+      "loss": 0.9414,
+      "step": 5480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4543235653685995e-05,
+      "loss": 1.0049,
+      "step": 5481
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.454109673876251e-05,
+      "loss": 0.9766,
+      "step": 5482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4538957470969365e-05,
+      "loss": 0.6807,
+      "step": 5483
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.453681785035848e-05,
+      "loss": 0.8975,
+      "step": 5484
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.453467787698178e-05,
+      "loss": 1.0303,
+      "step": 5485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.453253755089119e-05,
+      "loss": 0.8389,
+      "step": 5486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.453039687213865e-05,
+      "loss": 0.8135,
+      "step": 5487
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4528255840776114e-05,
+      "loss": 0.8613,
+      "step": 5488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.452611445685553e-05,
+      "loss": 0.9512,
+      "step": 5489
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.452397272042887e-05,
+      "loss": 0.9766,
+      "step": 5490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4521830631548105e-05,
+      "loss": 0.9014,
+      "step": 5491
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4519688190265224e-05,
+      "loss": 1.0752,
+      "step": 5492
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.45175453966322e-05,
+      "loss": 0.8135,
+      "step": 5493
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4515402250701055e-05,
+      "loss": 0.9746,
+      "step": 5494
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.451325875252378e-05,
+      "loss": 0.8242,
+      "step": 5495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4511114902152404e-05,
+      "loss": 0.8677,
+      "step": 5496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.450897069963894e-05,
+      "loss": 0.918,
+      "step": 5497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.450682614503542e-05,
+      "loss": 1.0771,
+      "step": 5498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.45046812383939e-05,
+      "loss": 1.0283,
+      "step": 5499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4502535979766425e-05,
+      "loss": 0.8838,
+      "step": 5500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4500390369205045e-05,
+      "loss": 1.1699,
+      "step": 5501
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.449824440676184e-05,
+      "loss": 1.127,
+      "step": 5502
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.449609809248888e-05,
+      "loss": 0.8564,
+      "step": 5503
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.449395142643825e-05,
+      "loss": 0.7275,
+      "step": 5504
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.449180440866204e-05,
+      "loss": 1.1523,
+      "step": 5505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4489657039212366e-05,
+      "loss": 1.0127,
+      "step": 5506
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.448750931814132e-05,
+      "loss": 0.793,
+      "step": 5507
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.448536124550103e-05,
+      "loss": 1.002,
+      "step": 5508
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4483212821343625e-05,
+      "loss": 1.0283,
+      "step": 5509
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.448106404572124e-05,
+      "loss": 0.9287,
+      "step": 5510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.447891491868602e-05,
+      "loss": 0.9287,
+      "step": 5511
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4476765440290104e-05,
+      "loss": 1.0283,
+      "step": 5512
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4474615610585677e-05,
+      "loss": 0.8477,
+      "step": 5513
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4472465429624886e-05,
+      "loss": 0.8408,
+      "step": 5514
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4470314897459935e-05,
+      "loss": 0.9893,
+      "step": 5515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.446816401414299e-05,
+      "loss": 0.9873,
+      "step": 5516
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.446601277972625e-05,
+      "loss": 0.8418,
+      "step": 5517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4463861194261925e-05,
+      "loss": 0.9561,
+      "step": 5518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.446170925780223e-05,
+      "loss": 0.8584,
+      "step": 5519
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.445955697039938e-05,
+      "loss": 1.0127,
+      "step": 5520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.44574043321056e-05,
+      "loss": 0.917,
+      "step": 5521
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.445525134297314e-05,
+      "loss": 0.9209,
+      "step": 5522
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.445309800305424e-05,
+      "loss": 1.1035,
+      "step": 5523
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.445094431240116e-05,
+      "loss": 1.1182,
+      "step": 5524
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.444879027106616e-05,
+      "loss": 0.8984,
+      "step": 5525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.444663587910151e-05,
+      "loss": 1.1094,
+      "step": 5526
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.444448113655949e-05,
+      "loss": 0.9951,
+      "step": 5527
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4442326043492396e-05,
+      "loss": 1.0127,
+      "step": 5528
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.444017059995252e-05,
+      "loss": 1.0254,
+      "step": 5529
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4438014805992185e-05,
+      "loss": 0.832,
+      "step": 5530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.443585866166367e-05,
+      "loss": 0.7402,
+      "step": 5531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4433702167019326e-05,
+      "loss": 1.0752,
+      "step": 5532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4431545322111485e-05,
+      "loss": 0.9492,
+      "step": 5533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.442938812699248e-05,
+      "loss": 0.9658,
+      "step": 5534
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.442723058171465e-05,
+      "loss": 1.043,
+      "step": 5535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4425072686330374e-05,
+      "loss": 0.9482,
+      "step": 5536
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4422914440892e-05,
+      "loss": 0.9199,
+      "step": 5537
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4420755845451914e-05,
+      "loss": 1.123,
+      "step": 5538
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4418596900062494e-05,
+      "loss": 1.1816,
+      "step": 5539
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.441643760477613e-05,
+      "loss": 0.9639,
+      "step": 5540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4414277959645216e-05,
+      "loss": 0.8467,
+      "step": 5541
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4412117964722173e-05,
+      "loss": 0.833,
+      "step": 5542
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.440995762005941e-05,
+      "loss": 0.8516,
+      "step": 5543
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4407796925709354e-05,
+      "loss": 1.0137,
+      "step": 5544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.440563588172443e-05,
+      "loss": 1.0352,
+      "step": 5545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.44034744881571e-05,
+      "loss": 0.9893,
+      "step": 5546
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.44013127450598e-05,
+      "loss": 0.999,
+      "step": 5547
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4399150652484994e-05,
+      "loss": 1.0908,
+      "step": 5548
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.439698821048515e-05,
+      "loss": 0.9297,
+      "step": 5549
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4394825419112736e-05,
+      "loss": 1.0859,
+      "step": 5550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.439266227842025e-05,
+      "loss": 1.1416,
+      "step": 5551
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.439049878846016e-05,
+      "loss": 0.9736,
+      "step": 5552
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4388334949285e-05,
+      "loss": 0.8623,
+      "step": 5553
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4386170760947265e-05,
+      "loss": 0.9961,
+      "step": 5554
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.438400622349947e-05,
+      "loss": 0.9414,
+      "step": 5555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.438184133699415e-05,
+      "loss": 0.9395,
+      "step": 5556
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.437967610148384e-05,
+      "loss": 0.9326,
+      "step": 5557
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.437751051702107e-05,
+      "loss": 1.0879,
+      "step": 5558
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.437534458365841e-05,
+      "loss": 0.8672,
+      "step": 5559
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4373178301448405e-05,
+      "loss": 1.0078,
+      "step": 5560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4371011670443644e-05,
+      "loss": 1.1406,
+      "step": 5561
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.436884469069669e-05,
+      "loss": 1.123,
+      "step": 5562
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.436667736226013e-05,
+      "loss": 0.9053,
+      "step": 5563
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.436450968518656e-05,
+      "loss": 1.0996,
+      "step": 5564
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4362341659528584e-05,
+      "loss": 1.0625,
+      "step": 5565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.436017328533881e-05,
+      "loss": 0.9805,
+      "step": 5566
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.435800456266987e-05,
+      "loss": 0.8652,
+      "step": 5567
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.435583549157438e-05,
+      "loss": 0.9365,
+      "step": 5568
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.435366607210499e-05,
+      "loss": 1.0713,
+      "step": 5569
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.435149630431433e-05,
+      "loss": 0.8486,
+      "step": 5570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.434932618825506e-05,
+      "loss": 0.9697,
+      "step": 5571
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.434715572397985e-05,
+      "loss": 1.0488,
+      "step": 5572
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.434498491154136e-05,
+      "loss": 0.8672,
+      "step": 5573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.434281375099227e-05,
+      "loss": 0.9814,
+      "step": 5574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.434064224238527e-05,
+      "loss": 1.0264,
+      "step": 5575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.433847038577306e-05,
+      "loss": 0.876,
+      "step": 5576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.433629818120834e-05,
+      "loss": 1.0303,
+      "step": 5577
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.433412562874384e-05,
+      "loss": 0.8613,
+      "step": 5578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4331952728432246e-05,
+      "loss": 1.0273,
+      "step": 5579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4329779480326317e-05,
+      "loss": 1.0537,
+      "step": 5580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4327605884478776e-05,
+      "loss": 0.9775,
+      "step": 5581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.432543194094238e-05,
+      "loss": 1.1035,
+      "step": 5582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.432325764976988e-05,
+      "loss": 0.7861,
+      "step": 5583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.432108301101404e-05,
+      "loss": 0.9893,
+      "step": 5584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.431890802472764e-05,
+      "loss": 0.9014,
+      "step": 5585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4316732690963444e-05,
+      "loss": 0.9639,
+      "step": 5586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.431455700977425e-05,
+      "loss": 1.1104,
+      "step": 5587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.431238098121286e-05,
+      "loss": 1.1055,
+      "step": 5588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.431020460533207e-05,
+      "loss": 1.0762,
+      "step": 5589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4308027882184704e-05,
+      "loss": 0.9619,
+      "step": 5590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.430585081182357e-05,
+      "loss": 1.0312,
+      "step": 5591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4303673394301516e-05,
+      "loss": 0.9844,
+      "step": 5592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4301495629671374e-05,
+      "loss": 0.8057,
+      "step": 5593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.429931751798599e-05,
+      "loss": 1.0537,
+      "step": 5594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.429713905929822e-05,
+      "loss": 1.0254,
+      "step": 5595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4294960253660936e-05,
+      "loss": 1.0957,
+      "step": 5596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.429278110112701e-05,
+      "loss": 1.0156,
+      "step": 5597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.429060160174932e-05,
+      "loss": 1.0244,
+      "step": 5598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4288421755580744e-05,
+      "loss": 1.0869,
+      "step": 5599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4286241562674205e-05,
+      "loss": 1.1416,
+      "step": 5600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4284061023082596e-05,
+      "loss": 1.0127,
+      "step": 5601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4281880136858826e-05,
+      "loss": 1.1182,
+      "step": 5602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.427969890405583e-05,
+      "loss": 0.9912,
+      "step": 5603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.427751732472654e-05,
+      "loss": 1.0879,
+      "step": 5604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.42753353989239e-05,
+      "loss": 1.1221,
+      "step": 5605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.427315312670084e-05,
+      "loss": 0.8286,
+      "step": 5606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4270970508110334e-05,
+      "loss": 1.0459,
+      "step": 5607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4268787543205343e-05,
+      "loss": 1.0762,
+      "step": 5608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.426660423203884e-05,
+      "loss": 0.7812,
+      "step": 5609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.426442057466381e-05,
+      "loss": 1.0088,
+      "step": 5610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4262236571133246e-05,
+      "loss": 0.8281,
+      "step": 5611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.426005222150014e-05,
+      "loss": 1.04,
+      "step": 5612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.425786752581751e-05,
+      "loss": 1.001,
+      "step": 5613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.425568248413836e-05,
+      "loss": 0.9434,
+      "step": 5614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4253497096515716e-05,
+      "loss": 1.1113,
+      "step": 5615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.425131136300262e-05,
+      "loss": 0.959,
+      "step": 5616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4249125283652115e-05,
+      "loss": 0.8906,
+      "step": 5617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.424693885851723e-05,
+      "loss": 1.043,
+      "step": 5618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4244752087651054e-05,
+      "loss": 1.0723,
+      "step": 5619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.424256497110663e-05,
+      "loss": 0.9033,
+      "step": 5620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4240377508937045e-05,
+      "loss": 0.999,
+      "step": 5621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4238189701195365e-05,
+      "loss": 1.001,
+      "step": 5622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.42360015479347e-05,
+      "loss": 0.9453,
+      "step": 5623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.423381304920815e-05,
+      "loss": 0.9385,
+      "step": 5624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.423162420506881e-05,
+      "loss": 0.917,
+      "step": 5625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.422943501556981e-05,
+      "loss": 0.8955,
+      "step": 5626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4227245480764266e-05,
+      "loss": 0.9229,
+      "step": 5627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.422505560070532e-05,
+      "loss": 0.8525,
+      "step": 5628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.42228653754461e-05,
+      "loss": 1.0342,
+      "step": 5629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.422067480503977e-05,
+      "loss": 0.8506,
+      "step": 5630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4218483889539484e-05,
+      "loss": 1.1182,
+      "step": 5631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.421629262899841e-05,
+      "loss": 0.9424,
+      "step": 5632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.421410102346972e-05,
+      "loss": 0.9131,
+      "step": 5633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.42119090730066e-05,
+      "loss": 0.8291,
+      "step": 5634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4209716777662246e-05,
+      "loss": 0.8799,
+      "step": 5635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.420752413748985e-05,
+      "loss": 0.8496,
+      "step": 5636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.420533115254263e-05,
+      "loss": 1.0156,
+      "step": 5637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4203137822873805e-05,
+      "loss": 1.085,
+      "step": 5638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.420094414853658e-05,
+      "loss": 0.9482,
+      "step": 5639
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4198750129584215e-05,
+      "loss": 0.9209,
+      "step": 5640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.419655576606994e-05,
+      "loss": 1.0674,
+      "step": 5641
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4194361058047004e-05,
+      "loss": 1.0811,
+      "step": 5642
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4192166005568674e-05,
+      "loss": 0.9141,
+      "step": 5643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.418997060868821e-05,
+      "loss": 1.1133,
+      "step": 5644
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.418777486745889e-05,
+      "loss": 0.8589,
+      "step": 5645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4185578781934e-05,
+      "loss": 0.9062,
+      "step": 5646
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.418338235216683e-05,
+      "loss": 1.1621,
+      "step": 5647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4181185578210687e-05,
+      "loss": 1.0645,
+      "step": 5648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.417898846011887e-05,
+      "loss": 0.8809,
+      "step": 5649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.41767909979447e-05,
+      "loss": 0.9814,
+      "step": 5650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4174593191741504e-05,
+      "loss": 0.8936,
+      "step": 5651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4172395041562624e-05,
+      "loss": 0.9502,
+      "step": 5652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.417019654746139e-05,
+      "loss": 0.9297,
+      "step": 5653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.416799770949116e-05,
+      "loss": 1.1367,
+      "step": 5654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.41657985277053e-05,
+      "loss": 1.0723,
+      "step": 5655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4163599002157166e-05,
+      "loss": 0.9326,
+      "step": 5656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.416139913290013e-05,
+      "loss": 1.0635,
+      "step": 5657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4159198919987595e-05,
+      "loss": 0.9697,
+      "step": 5658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.415699836347294e-05,
+      "loss": 0.9795,
+      "step": 5659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4154797463409556e-05,
+      "loss": 0.8828,
+      "step": 5660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.415259621985088e-05,
+      "loss": 0.7783,
+      "step": 5661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4150394632850314e-05,
+      "loss": 0.8428,
+      "step": 5662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.414819270246129e-05,
+      "loss": 0.9961,
+      "step": 5663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.414599042873722e-05,
+      "loss": 1.124,
+      "step": 5664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4143787811731576e-05,
+      "loss": 0.9697,
+      "step": 5665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4141584851497795e-05,
+      "loss": 1.1152,
+      "step": 5666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.413938154808934e-05,
+      "loss": 1.0088,
+      "step": 5667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4137177901559676e-05,
+      "loss": 1.084,
+      "step": 5668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.413497391196228e-05,
+      "loss": 0.8359,
+      "step": 5669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4132769579350635e-05,
+      "loss": 0.9492,
+      "step": 5670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.413056490377824e-05,
+      "loss": 0.96,
+      "step": 5671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.412835988529859e-05,
+      "loss": 1.0059,
+      "step": 5672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.412615452396519e-05,
+      "loss": 0.959,
+      "step": 5673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.412394881983157e-05,
+      "loss": 0.9023,
+      "step": 5674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4121742772951246e-05,
+      "loss": 0.8271,
+      "step": 5675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.411953638337775e-05,
+      "loss": 0.8594,
+      "step": 5676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.411732965116464e-05,
+      "loss": 1.0547,
+      "step": 5677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4115122576365454e-05,
+      "loss": 0.9873,
+      "step": 5678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.411291515903375e-05,
+      "loss": 1.0576,
+      "step": 5679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.41107073992231e-05,
+      "loss": 1.0996,
+      "step": 5680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.410849929698708e-05,
+      "loss": 0.8828,
+      "step": 5681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4106290852379285e-05,
+      "loss": 0.7764,
+      "step": 5682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.410408206545328e-05,
+      "loss": 1.0381,
+      "step": 5683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.410187293626269e-05,
+      "loss": 0.791,
+      "step": 5684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.409966346486112e-05,
+      "loss": 0.8721,
+      "step": 5685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4097453651302175e-05,
+      "loss": 1.0967,
+      "step": 5686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.409524349563949e-05,
+      "loss": 1.0107,
+      "step": 5687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.40930329979267e-05,
+      "loss": 0.8086,
+      "step": 5688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.409082215821744e-05,
+      "loss": 1.0479,
+      "step": 5689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.408861097656538e-05,
+      "loss": 0.9258,
+      "step": 5690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.408639945302415e-05,
+      "loss": 1.0488,
+      "step": 5691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4084187587647436e-05,
+      "loss": 1.0166,
+      "step": 5692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4081975380488916e-05,
+      "loss": 1.0449,
+      "step": 5693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.407976283160226e-05,
+      "loss": 0.959,
+      "step": 5694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.407754994104117e-05,
+      "loss": 0.9199,
+      "step": 5695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.407533670885934e-05,
+      "loss": 0.9736,
+      "step": 5696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.407312313511049e-05,
+      "loss": 0.9229,
+      "step": 5697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.407090921984832e-05,
+      "loss": 0.9844,
+      "step": 5698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4068694963126576e-05,
+      "loss": 1.0918,
+      "step": 5699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.406648036499897e-05,
+      "loss": 1.2637,
+      "step": 5700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.406426542551925e-05,
+      "loss": 0.9814,
+      "step": 5701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4062050144741183e-05,
+      "loss": 0.9775,
+      "step": 5702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.405983452271851e-05,
+      "loss": 0.9639,
+      "step": 5703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.405761855950499e-05,
+      "loss": 1.2207,
+      "step": 5704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4055402255154416e-05,
+      "loss": 0.9697,
+      "step": 5705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.405318560972057e-05,
+      "loss": 0.9199,
+      "step": 5706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.405096862325723e-05,
+      "loss": 0.7502,
+      "step": 5707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.404875129581821e-05,
+      "loss": 0.8223,
+      "step": 5708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.404653362745731e-05,
+      "loss": 0.9492,
+      "step": 5709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.404431561822835e-05,
+      "loss": 1.0938,
+      "step": 5710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4042097268185155e-05,
+      "loss": 1.0293,
+      "step": 5711
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4039878577381546e-05,
+      "loss": 0.915,
+      "step": 5712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.403765954587138e-05,
+      "loss": 1.0527,
+      "step": 5713
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.40354401737085e-05,
+      "loss": 0.9365,
+      "step": 5714
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.403322046094677e-05,
+      "loss": 0.9697,
+      "step": 5715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.403100040764004e-05,
+      "loss": 0.917,
+      "step": 5716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.402878001384219e-05,
+      "loss": 0.9385,
+      "step": 5717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4026559279607126e-05,
+      "loss": 1.1426,
+      "step": 5718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4024338204988705e-05,
+      "loss": 0.8975,
+      "step": 5719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4022116790040835e-05,
+      "loss": 0.9258,
+      "step": 5720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4019895034817434e-05,
+      "loss": 0.999,
+      "step": 5721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4017672939372415e-05,
+      "loss": 1.1406,
+      "step": 5722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.401545050375969e-05,
+      "loss": 0.9004,
+      "step": 5723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.401322772803321e-05,
+      "loss": 0.9873,
+      "step": 5724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4011004612246894e-05,
+      "loss": 1.1484,
+      "step": 5725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.400878115645471e-05,
+      "loss": 1.0986,
+      "step": 5726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4006557360710596e-05,
+      "loss": 1.1152,
+      "step": 5727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.400433322506853e-05,
+      "loss": 1.0322,
+      "step": 5728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.400210874958249e-05,
+      "loss": 1.0264,
+      "step": 5729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.399988393430644e-05,
+      "loss": 0.8848,
+      "step": 5730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.399765877929438e-05,
+      "loss": 1.085,
+      "step": 5731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.399543328460031e-05,
+      "loss": 0.9561,
+      "step": 5732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.399320745027823e-05,
+      "loss": 1.0176,
+      "step": 5733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.399098127638216e-05,
+      "loss": 0.9248,
+      "step": 5734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.398875476296612e-05,
+      "loss": 1.0234,
+      "step": 5735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3986527910084146e-05,
+      "loss": 1.0811,
+      "step": 5736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.398430071779027e-05,
+      "loss": 0.9854,
+      "step": 5737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.398207318613854e-05,
+      "loss": 0.9844,
+      "step": 5738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.397984531518301e-05,
+      "loss": 0.9873,
+      "step": 5739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.397761710497776e-05,
+      "loss": 1.04,
+      "step": 5740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.397538855557684e-05,
+      "loss": 0.8438,
+      "step": 5741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3973159667034336e-05,
+      "loss": 1.1035,
+      "step": 5742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3970930439404344e-05,
+      "loss": 1.0244,
+      "step": 5743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.396870087274096e-05,
+      "loss": 0.96,
+      "step": 5744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.396647096709829e-05,
+      "loss": 1.0049,
+      "step": 5745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.396424072253043e-05,
+      "loss": 0.9707,
+      "step": 5746
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.396201013909153e-05,
+      "loss": 0.9023,
+      "step": 5747
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.395977921683569e-05,
+      "loss": 0.8467,
+      "step": 5748
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3957547955817074e-05,
+      "loss": 0.9863,
+      "step": 5749
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.395531635608982e-05,
+      "loss": 0.8623,
+      "step": 5750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.395308441770807e-05,
+      "loss": 0.8506,
+      "step": 5751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3950852140725995e-05,
+      "loss": 0.9023,
+      "step": 5752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.394861952519778e-05,
+      "loss": 1.0156,
+      "step": 5753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.394638657117757e-05,
+      "loss": 0.9863,
+      "step": 5754
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.394415327871959e-05,
+      "loss": 0.9268,
+      "step": 5755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.394191964787801e-05,
+      "loss": 0.9453,
+      "step": 5756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.393968567870704e-05,
+      "loss": 1.082,
+      "step": 5757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.39374513712609e-05,
+      "loss": 1.0557,
+      "step": 5758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.39352167255938e-05,
+      "loss": 0.9062,
+      "step": 5759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.393298174175997e-05,
+      "loss": 0.9375,
+      "step": 5760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3930746419813655e-05,
+      "loss": 0.9824,
+      "step": 5761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.392851075980909e-05,
+      "loss": 0.9111,
+      "step": 5762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.392627476180053e-05,
+      "loss": 0.9316,
+      "step": 5763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3924038425842236e-05,
+      "loss": 0.8945,
+      "step": 5764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3921801751988485e-05,
+      "loss": 1.0508,
+      "step": 5765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.391956474029354e-05,
+      "loss": 1.0488,
+      "step": 5766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.39173273908117e-05,
+      "loss": 0.8721,
+      "step": 5767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3915089703597255e-05,
+      "loss": 0.9131,
+      "step": 5768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.39128516787045e-05,
+      "loss": 1.1113,
+      "step": 5769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.391061331618775e-05,
+      "loss": 1.1777,
+      "step": 5770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.390837461610133e-05,
+      "loss": 1.0859,
+      "step": 5771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.390613557849956e-05,
+      "loss": 0.7705,
+      "step": 5772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3903896203436777e-05,
+      "loss": 0.8896,
+      "step": 5773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.390165649096732e-05,
+      "loss": 1.0879,
+      "step": 5774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.389941644114554e-05,
+      "loss": 1.085,
+      "step": 5775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.389717605402581e-05,
+      "loss": 0.9199,
+      "step": 5776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.389493532966248e-05,
+      "loss": 0.959,
+      "step": 5777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3892694268109935e-05,
+      "loss": 0.9951,
+      "step": 5778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3890452869422554e-05,
+      "loss": 1.1182,
+      "step": 5779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.388821113365474e-05,
+      "loss": 1.0938,
+      "step": 5780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.388596906086088e-05,
+      "loss": 0.7949,
+      "step": 5781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.388372665109539e-05,
+      "loss": 1.1064,
+      "step": 5782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3881483904412685e-05,
+      "loss": 1.0,
+      "step": 5783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.387924082086718e-05,
+      "loss": 0.9375,
+      "step": 5784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.387699740051333e-05,
+      "loss": 1.0166,
+      "step": 5785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.387475364340556e-05,
+      "loss": 1.1289,
+      "step": 5786
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.387250954959832e-05,
+      "loss": 1.0469,
+      "step": 5787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.387026511914608e-05,
+      "loss": 1.0547,
+      "step": 5788
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3868020352103285e-05,
+      "loss": 1.0039,
+      "step": 5789
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.386577524852443e-05,
+      "loss": 1.1309,
+      "step": 5790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.386352980846398e-05,
+      "loss": 0.7344,
+      "step": 5791
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.386128403197643e-05,
+      "loss": 0.8467,
+      "step": 5792
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.385903791911629e-05,
+      "loss": 1.0527,
+      "step": 5793
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.385679146993805e-05,
+      "loss": 0.9521,
+      "step": 5794
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.385454468449624e-05,
+      "loss": 0.8784,
+      "step": 5795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.385229756284537e-05,
+      "loss": 1.0361,
+      "step": 5796
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3850050105039986e-05,
+      "loss": 1.124,
+      "step": 5797
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3847802311134606e-05,
+      "loss": 0.9893,
+      "step": 5798
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.384555418118379e-05,
+      "loss": 0.957,
+      "step": 5799
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.38433057152421e-05,
+      "loss": 0.8896,
+      "step": 5800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.384105691336409e-05,
+      "loss": 0.8887,
+      "step": 5801
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.383880777560433e-05,
+      "loss": 0.9727,
+      "step": 5802
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.38365583020174e-05,
+      "loss": 1.0098,
+      "step": 5803
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.38343084926579e-05,
+      "loss": 0.8193,
+      "step": 5804
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.383205834758041e-05,
+      "loss": 0.9609,
+      "step": 5805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.382980786683955e-05,
+      "loss": 0.9404,
+      "step": 5806
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.382755705048992e-05,
+      "loss": 0.8467,
+      "step": 5807
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3825305898586154e-05,
+      "loss": 0.874,
+      "step": 5808
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.382305441118286e-05,
+      "loss": 0.8301,
+      "step": 5809
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.38208025883347e-05,
+      "loss": 0.9785,
+      "step": 5810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.38185504300963e-05,
+      "loss": 1.1777,
+      "step": 5811
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.381629793652231e-05,
+      "loss": 1.0029,
+      "step": 5812
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.381404510766742e-05,
+      "loss": 1.0488,
+      "step": 5813
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.381179194358627e-05,
+      "loss": 1.1172,
+      "step": 5814
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3809538444333555e-05,
+      "loss": 1.0498,
+      "step": 5815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.380728460996394e-05,
+      "loss": 1.0488,
+      "step": 5816
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3805030440532154e-05,
+      "loss": 1.1182,
+      "step": 5817
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.380277593609287e-05,
+      "loss": 0.8018,
+      "step": 5818
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3800521096700805e-05,
+      "loss": 1.0215,
+      "step": 5819
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3798265922410676e-05,
+      "loss": 1.1094,
+      "step": 5820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3796010413277214e-05,
+      "loss": 1.0244,
+      "step": 5821
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.379375456935516e-05,
+      "loss": 0.8828,
+      "step": 5822
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.379149839069924e-05,
+      "loss": 0.958,
+      "step": 5823
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.378924187736421e-05,
+      "loss": 1.0898,
+      "step": 5824
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.378698502940484e-05,
+      "loss": 0.8984,
+      "step": 5825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.378472784687588e-05,
+      "loss": 1.0254,
+      "step": 5826
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.378247032983213e-05,
+      "loss": 1.0303,
+      "step": 5827
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.378021247832835e-05,
+      "loss": 0.8623,
+      "step": 5828
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3777954292419336e-05,
+      "loss": 1.0996,
+      "step": 5829
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.377569577215989e-05,
+      "loss": 1.0859,
+      "step": 5830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.377343691760483e-05,
+      "loss": 0.7075,
+      "step": 5831
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.377117772880896e-05,
+      "loss": 0.9541,
+      "step": 5832
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.37689182058271e-05,
+      "loss": 0.9639,
+      "step": 5833
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.376665834871409e-05,
+      "loss": 1.0684,
+      "step": 5834
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.376439815752477e-05,
+      "loss": 1.0879,
+      "step": 5835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3762137632313986e-05,
+      "loss": 0.9756,
+      "step": 5836
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.375987677313659e-05,
+      "loss": 0.8984,
+      "step": 5837
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3757615580047455e-05,
+      "loss": 0.9434,
+      "step": 5838
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.375535405310145e-05,
+      "loss": 0.8955,
+      "step": 5839
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3753092192353456e-05,
+      "loss": 0.8188,
+      "step": 5840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.375082999785836e-05,
+      "loss": 1.0537,
+      "step": 5841
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3748567469671055e-05,
+      "loss": 1.0078,
+      "step": 5842
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.374630460784645e-05,
+      "loss": 1.0527,
+      "step": 5843
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.374404141243947e-05,
+      "loss": 0.9346,
+      "step": 5844
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3741777883505013e-05,
+      "loss": 0.7949,
+      "step": 5845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.373951402109803e-05,
+      "loss": 0.9355,
+      "step": 5846
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.373724982527344e-05,
+      "loss": 1.0723,
+      "step": 5847
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.37349852960862e-05,
+      "loss": 0.9648,
+      "step": 5848
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.373272043359126e-05,
+      "loss": 0.7993,
+      "step": 5849
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.373045523784358e-05,
+      "loss": 0.9072,
+      "step": 5850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3728189708898126e-05,
+      "loss": 0.8291,
+      "step": 5851
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.372592384680989e-05,
+      "loss": 0.7861,
+      "step": 5852
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.372365765163385e-05,
+      "loss": 1.0488,
+      "step": 5853
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3721391123424987e-05,
+      "loss": 1.0059,
+      "step": 5854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3719124262238323e-05,
+      "loss": 1.0049,
+      "step": 5855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3716857068128864e-05,
+      "loss": 0.9863,
+      "step": 5856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.371458954115162e-05,
+      "loss": 0.9395,
+      "step": 5857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3712321681361614e-05,
+      "loss": 1.1338,
+      "step": 5858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.371005348881389e-05,
+      "loss": 0.9248,
+      "step": 5859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.37077849635635e-05,
+      "loss": 0.917,
+      "step": 5860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.370551610566547e-05,
+      "loss": 1.0479,
+      "step": 5861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.370324691517487e-05,
+      "loss": 0.9258,
+      "step": 5862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.370097739214677e-05,
+      "loss": 0.96,
+      "step": 5863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.369870753663624e-05,
+      "loss": 0.9326,
+      "step": 5864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3696437348698364e-05,
+      "loss": 1.0996,
+      "step": 5865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.369416682838824e-05,
+      "loss": 1.0137,
+      "step": 5866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.369189597576096e-05,
+      "loss": 0.8633,
+      "step": 5867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.368962479087162e-05,
+      "loss": 0.9941,
+      "step": 5868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.368735327377536e-05,
+      "loss": 1.1895,
+      "step": 5869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.368508142452728e-05,
+      "loss": 1.0957,
+      "step": 5870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.368280924318252e-05,
+      "loss": 0.9141,
+      "step": 5871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3680536729796224e-05,
+      "loss": 1.001,
+      "step": 5872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3678263884423534e-05,
+      "loss": 0.9443,
+      "step": 5873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3675990707119605e-05,
+      "loss": 1.0596,
+      "step": 5874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3673717197939595e-05,
+      "loss": 0.9102,
+      "step": 5875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.367144335693869e-05,
+      "loss": 1.0635,
+      "step": 5876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3669169184172056e-05,
+      "loss": 1.0879,
+      "step": 5877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3666894679694885e-05,
+      "loss": 0.916,
+      "step": 5878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.366461984356238e-05,
+      "loss": 1.0977,
+      "step": 5879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.366234467582973e-05,
+      "loss": 0.8877,
+      "step": 5880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3660069176552154e-05,
+      "loss": 0.8252,
+      "step": 5881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.365779334578488e-05,
+      "loss": 0.9658,
+      "step": 5882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3655517183583126e-05,
+      "loss": 0.8311,
+      "step": 5883
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.365324069000212e-05,
+      "loss": 1.0547,
+      "step": 5884
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.365096386509713e-05,
+      "loss": 1.042,
+      "step": 5885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.364868670892338e-05,
+      "loss": 0.9102,
+      "step": 5886
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.364640922153615e-05,
+      "loss": 0.9824,
+      "step": 5887
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3644131402990694e-05,
+      "loss": 0.9824,
+      "step": 5888
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.36418532533423e-05,
+      "loss": 0.813,
+      "step": 5889
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.363957477264625e-05,
+      "loss": 1.0449,
+      "step": 5890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3637295960957825e-05,
+      "loss": 0.7988,
+      "step": 5891
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.363501681833234e-05,
+      "loss": 0.8125,
+      "step": 5892
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3632737344825095e-05,
+      "loss": 1.0723,
+      "step": 5893
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.36304575404914e-05,
+      "loss": 0.9443,
+      "step": 5894
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.362817740538659e-05,
+      "loss": 0.9502,
+      "step": 5895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.362589693956599e-05,
+      "loss": 0.9043,
+      "step": 5896
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.362361614308495e-05,
+      "loss": 0.8594,
+      "step": 5897
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.362133501599881e-05,
+      "loss": 0.8848,
+      "step": 5898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.361905355836292e-05,
+      "loss": 0.9219,
+      "step": 5899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.361677177023267e-05,
+      "loss": 1.0762,
+      "step": 5900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.36144896516634e-05,
+      "loss": 1.1602,
+      "step": 5901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.361220720271051e-05,
+      "loss": 0.8477,
+      "step": 5902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.360992442342938e-05,
+      "loss": 0.9082,
+      "step": 5903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3607641313875416e-05,
+      "loss": 0.957,
+      "step": 5904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.360535787410401e-05,
+      "loss": 1.2031,
+      "step": 5905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3603074104170584e-05,
+      "loss": 0.8975,
+      "step": 5906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.360079000413056e-05,
+      "loss": 0.9551,
+      "step": 5907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.359850557403935e-05,
+      "loss": 0.9326,
+      "step": 5908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3596220813952404e-05,
+      "loss": 1.0615,
+      "step": 5909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3593935723925165e-05,
+      "loss": 1.0107,
+      "step": 5910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.359165030401309e-05,
+      "loss": 0.9492,
+      "step": 5911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.358936455427163e-05,
+      "loss": 0.7583,
+      "step": 5912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.358707847475626e-05,
+      "loss": 1.0098,
+      "step": 5913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.358479206552244e-05,
+      "loss": 0.9863,
+      "step": 5914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.358250532662568e-05,
+      "loss": 1.1562,
+      "step": 5915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.358021825812146e-05,
+      "loss": 0.7832,
+      "step": 5916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.357793086006527e-05,
+      "loss": 0.8535,
+      "step": 5917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.357564313251264e-05,
+      "loss": 1.0342,
+      "step": 5918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.357335507551908e-05,
+      "loss": 0.7959,
+      "step": 5919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3571066689140105e-05,
+      "loss": 1.1104,
+      "step": 5920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.356877797343125e-05,
+      "loss": 1.1816,
+      "step": 5921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3566488928448065e-05,
+      "loss": 1.0967,
+      "step": 5922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.356419955424608e-05,
+      "loss": 0.7173,
+      "step": 5923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.356190985088087e-05,
+      "loss": 0.958,
+      "step": 5924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.355961981840799e-05,
+      "loss": 0.9316,
+      "step": 5925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3557329456883016e-05,
+      "loss": 0.9355,
+      "step": 5926
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3555038766361526e-05,
+      "loss": 1.0684,
+      "step": 5927
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.355274774689911e-05,
+      "loss": 0.7656,
+      "step": 5928
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.355045639855136e-05,
+      "loss": 1.1113,
+      "step": 5929
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.354816472137389e-05,
+      "loss": 0.7817,
+      "step": 5930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.35458727154223e-05,
+      "loss": 0.9062,
+      "step": 5931
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.354358038075222e-05,
+      "loss": 1.0098,
+      "step": 5932
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.354128771741927e-05,
+      "loss": 1.0537,
+      "step": 5933
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3538994725479087e-05,
+      "loss": 1.0371,
+      "step": 5934
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3536701404987326e-05,
+      "loss": 1.1191,
+      "step": 5935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3534407755999634e-05,
+      "loss": 0.9648,
+      "step": 5936
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3532113778571664e-05,
+      "loss": 1.0254,
+      "step": 5937
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.352981947275909e-05,
+      "loss": 1.1289,
+      "step": 5938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3527524838617586e-05,
+      "loss": 0.8945,
+      "step": 5939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.352522987620284e-05,
+      "loss": 0.915,
+      "step": 5940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3522934585570536e-05,
+      "loss": 0.9902,
+      "step": 5941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3520638966776385e-05,
+      "loss": 0.8945,
+      "step": 5942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.351834301987609e-05,
+      "loss": 0.9404,
+      "step": 5943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.351604674492536e-05,
+      "loss": 1.0469,
+      "step": 5944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3513750141979934e-05,
+      "loss": 0.9277,
+      "step": 5945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.351145321109552e-05,
+      "loss": 1.166,
+      "step": 5946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.350915595232789e-05,
+      "loss": 1.0264,
+      "step": 5947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.350685836573276e-05,
+      "loss": 0.9521,
+      "step": 5948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.350456045136591e-05,
+      "loss": 0.8584,
+      "step": 5949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.350226220928309e-05,
+      "loss": 1.0381,
+      "step": 5950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.349996363954007e-05,
+      "loss": 1.0283,
+      "step": 5951
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.349766474219264e-05,
+      "loss": 1.0908,
+      "step": 5952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3495365517296576e-05,
+      "loss": 0.9834,
+      "step": 5953
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.349306596490768e-05,
+      "loss": 0.9492,
+      "step": 5954
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.349076608508176e-05,
+      "loss": 1.0928,
+      "step": 5955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.348846587787462e-05,
+      "loss": 0.9326,
+      "step": 5956
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3486165343342083e-05,
+      "loss": 1.0195,
+      "step": 5957
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3483864481539966e-05,
+      "loss": 0.8867,
+      "step": 5958
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.348156329252412e-05,
+      "loss": 0.9883,
+      "step": 5959
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3479261776350375e-05,
+      "loss": 1.0176,
+      "step": 5960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3476959933074594e-05,
+      "loss": 1.0518,
+      "step": 5961
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.347465776275263e-05,
+      "loss": 1.0039,
+      "step": 5962
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.347235526544035e-05,
+      "loss": 1.1689,
+      "step": 5963
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.347005244119362e-05,
+      "loss": 0.9365,
+      "step": 5964
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.346774929006834e-05,
+      "loss": 1.1113,
+      "step": 5965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3465445812120393e-05,
+      "loss": 1.0645,
+      "step": 5966
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.346314200740567e-05,
+      "loss": 0.8896,
+      "step": 5967
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.346083787598009e-05,
+      "loss": 0.9688,
+      "step": 5968
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3458533417899556e-05,
+      "loss": 1.0781,
+      "step": 5969
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.345622863322e-05,
+      "loss": 0.8594,
+      "step": 5970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.345392352199735e-05,
+      "loss": 0.9482,
+      "step": 5971
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.345161808428754e-05,
+      "loss": 0.8291,
+      "step": 5972
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.344931232014652e-05,
+      "loss": 0.8291,
+      "step": 5973
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.344700622963024e-05,
+      "loss": 0.9346,
+      "step": 5974
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.344469981279467e-05,
+      "loss": 0.9434,
+      "step": 5975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.344239306969577e-05,
+      "loss": 1.0195,
+      "step": 5976
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.344008600038953e-05,
+      "loss": 1.042,
+      "step": 5977
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.343777860493192e-05,
+      "loss": 0.75,
+      "step": 5978
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.343547088337894e-05,
+      "loss": 0.9258,
+      "step": 5979
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.34331628357866e-05,
+      "loss": 0.9854,
+      "step": 5980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.34308544622109e-05,
+      "loss": 1.0195,
+      "step": 5981
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3428545762707866e-05,
+      "loss": 0.999,
+      "step": 5982
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3426236737333514e-05,
+      "loss": 1.0254,
+      "step": 5983
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.342392738614389e-05,
+      "loss": 0.7988,
+      "step": 5984
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.342161770919502e-05,
+      "loss": 0.9717,
+      "step": 5985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.341930770654296e-05,
+      "loss": 0.855,
+      "step": 5986
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.341699737824377e-05,
+      "loss": 0.9902,
+      "step": 5987
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3414686724353505e-05,
+      "loss": 1.0166,
+      "step": 5988
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3412375744928245e-05,
+      "loss": 1.0059,
+      "step": 5989
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.341006444002408e-05,
+      "loss": 0.9434,
+      "step": 5990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.340775280969708e-05,
+      "loss": 0.9023,
+      "step": 5991
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.340544085400335e-05,
+      "loss": 1.0146,
+      "step": 5992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3403128572999e-05,
+      "loss": 0.9824,
+      "step": 5993
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.340081596674013e-05,
+      "loss": 0.9658,
+      "step": 5994
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.339850303528287e-05,
+      "loss": 0.9609,
+      "step": 5995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3396189778683345e-05,
+      "loss": 1.0049,
+      "step": 5996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.339387619699769e-05,
+      "loss": 1.0635,
+      "step": 5997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.339156229028205e-05,
+      "loss": 0.8555,
+      "step": 5998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3389248058592576e-05,
+      "loss": 0.9883,
+      "step": 5999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.338693350198544e-05,
+      "loss": 0.9199,
+      "step": 6000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.338461862051678e-05,
+      "loss": 1.0801,
+      "step": 6001
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.338230341424279e-05,
+      "loss": 0.8721,
+      "step": 6002
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.337998788321966e-05,
+      "loss": 1.0459,
+      "step": 6003
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.337767202750357e-05,
+      "loss": 0.9258,
+      "step": 6004
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.337535584715072e-05,
+      "loss": 0.8672,
+      "step": 6005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3373039342217324e-05,
+      "loss": 1.0205,
+      "step": 6006
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.337072251275959e-05,
+      "loss": 1.0586,
+      "step": 6007
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.336840535883374e-05,
+      "loss": 1.0098,
+      "step": 6008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.336608788049601e-05,
+      "loss": 1.0938,
+      "step": 6009
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.336377007780264e-05,
+      "loss": 0.7061,
+      "step": 6010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.336145195080987e-05,
+      "loss": 0.8516,
+      "step": 6011
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.335913349957395e-05,
+      "loss": 0.999,
+      "step": 6012
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3356814724151144e-05,
+      "loss": 0.9668,
+      "step": 6013
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.335449562459773e-05,
+      "loss": 0.9453,
+      "step": 6014
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.335217620096998e-05,
+      "loss": 1.0703,
+      "step": 6015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.334985645332418e-05,
+      "loss": 1.0283,
+      "step": 6016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3347536381716635e-05,
+      "loss": 0.9219,
+      "step": 6017
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3345215986203627e-05,
+      "loss": 0.9878,
+      "step": 6018
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3342895266841476e-05,
+      "loss": 0.9951,
+      "step": 6019
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.334057422368649e-05,
+      "loss": 1.0215,
+      "step": 6020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.333825285679501e-05,
+      "loss": 0.7988,
+      "step": 6021
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.333593116622336e-05,
+      "loss": 1.0166,
+      "step": 6022
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.333360915202787e-05,
+      "loss": 1.0771,
+      "step": 6023
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.333128681426491e-05,
+      "loss": 0.8691,
+      "step": 6024
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.332896415299082e-05,
+      "loss": 0.9355,
+      "step": 6025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3326641168261976e-05,
+      "loss": 0.9736,
+      "step": 6026
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3324317860134726e-05,
+      "loss": 0.9785,
+      "step": 6027
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.332199422866548e-05,
+      "loss": 0.7041,
+      "step": 6028
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.331967027391061e-05,
+      "loss": 1.0879,
+      "step": 6029
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3317345995926516e-05,
+      "loss": 1.0713,
+      "step": 6030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.33150213947696e-05,
+      "loss": 0.8633,
+      "step": 6031
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.331269647049627e-05,
+      "loss": 1.0361,
+      "step": 6032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3310371223162945e-05,
+      "loss": 0.9043,
+      "step": 6033
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.330804565282606e-05,
+      "loss": 0.8115,
+      "step": 6034
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.330571975954205e-05,
+      "loss": 0.9971,
+      "step": 6035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.330339354336734e-05,
+      "loss": 0.8389,
+      "step": 6036
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.33010670043584e-05,
+      "loss": 0.9556,
+      "step": 6037
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3298740142571674e-05,
+      "loss": 0.9033,
+      "step": 6038
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.329641295806364e-05,
+      "loss": 0.9141,
+      "step": 6039
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.329408545089077e-05,
+      "loss": 0.9238,
+      "step": 6040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3291757621109536e-05,
+      "loss": 0.7783,
+      "step": 6041
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.328942946877644e-05,
+      "loss": 0.9043,
+      "step": 6042
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.328710099394797e-05,
+      "loss": 0.9307,
+      "step": 6043
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.328477219668064e-05,
+      "loss": 0.9658,
+      "step": 6044
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.328244307703095e-05,
+      "loss": 0.9023,
+      "step": 6045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3280113635055426e-05,
+      "loss": 0.9561,
+      "step": 6046
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3277783870810614e-05,
+      "loss": 0.9082,
+      "step": 6047
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.327545378435302e-05,
+      "loss": 0.917,
+      "step": 6048
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3273123375739216e-05,
+      "loss": 1.1367,
+      "step": 6049
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.327079264502574e-05,
+      "loss": 0.835,
+      "step": 6050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3268461592269155e-05,
+      "loss": 1.0156,
+      "step": 6051
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.326613021752602e-05,
+      "loss": 1.001,
+      "step": 6052
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3263798520852925e-05,
+      "loss": 1.0928,
+      "step": 6053
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.326146650230645e-05,
+      "loss": 1.042,
+      "step": 6054
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.325913416194318e-05,
+      "loss": 0.9355,
+      "step": 6055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.325680149981972e-05,
+      "loss": 0.9971,
+      "step": 6056
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3254468515992675e-05,
+      "loss": 0.9824,
+      "step": 6057
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.325213521051866e-05,
+      "loss": 1.0625,
+      "step": 6058
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.324980158345429e-05,
+      "loss": 1.0371,
+      "step": 6059
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.324746763485621e-05,
+      "loss": 0.9062,
+      "step": 6060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.324513336478104e-05,
+      "loss": 0.8848,
+      "step": 6061
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.324279877328544e-05,
+      "loss": 0.9062,
+      "step": 6062
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.324046386042606e-05,
+      "loss": 0.8145,
+      "step": 6063
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.323812862625956e-05,
+      "loss": 1.123,
+      "step": 6064
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.323579307084262e-05,
+      "loss": 0.8545,
+      "step": 6065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3233457194231894e-05,
+      "loss": 0.9326,
+      "step": 6066
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.323112099648409e-05,
+      "loss": 1.0762,
+      "step": 6067
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3228784477655884e-05,
+      "loss": 0.9121,
+      "step": 6068
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3226447637803984e-05,
+      "loss": 0.9775,
+      "step": 6069
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.32241104769851e-05,
+      "loss": 0.8037,
+      "step": 6070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3221772995255945e-05,
+      "loss": 1.0029,
+      "step": 6071
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.321943519267324e-05,
+      "loss": 1.0371,
+      "step": 6072
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.321709706929372e-05,
+      "loss": 1.0664,
+      "step": 6073
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.321475862517413e-05,
+      "loss": 1.0576,
+      "step": 6074
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.321241986037121e-05,
+      "loss": 1.1006,
+      "step": 6075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.321008077494171e-05,
+      "loss": 1.0449,
+      "step": 6076
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.32077413689424e-05,
+      "loss": 0.8403,
+      "step": 6077
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3205401642430053e-05,
+      "loss": 1.0762,
+      "step": 6078
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.320306159546145e-05,
+      "loss": 0.957,
+      "step": 6079
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.320072122809336e-05,
+      "loss": 1.0186,
+      "step": 6080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3198380540382594e-05,
+      "loss": 0.7407,
+      "step": 6081
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.319603953238595e-05,
+      "loss": 1.0996,
+      "step": 6082
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3193698204160224e-05,
+      "loss": 0.7788,
+      "step": 6083
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.319135655576225e-05,
+      "loss": 1.0059,
+      "step": 6084
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.318901458724885e-05,
+      "loss": 1.0088,
+      "step": 6085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.318667229867685e-05,
+      "loss": 0.9053,
+      "step": 6086
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.318432969010309e-05,
+      "loss": 0.8994,
+      "step": 6087
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.318198676158443e-05,
+      "loss": 0.8066,
+      "step": 6088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.317964351317772e-05,
+      "loss": 1.0723,
+      "step": 6089
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3177299944939814e-05,
+      "loss": 0.9941,
+      "step": 6090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.317495605692759e-05,
+      "loss": 1.1094,
+      "step": 6091
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3172611849197935e-05,
+      "loss": 0.8066,
+      "step": 6092
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3170267321807725e-05,
+      "loss": 1.0234,
+      "step": 6093
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3167922474813865e-05,
+      "loss": 0.9639,
+      "step": 6094
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.316557730827325e-05,
+      "loss": 1.1484,
+      "step": 6095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3163231822242786e-05,
+      "loss": 1.0576,
+      "step": 6096
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.31608860167794e-05,
+      "loss": 0.9814,
+      "step": 6097
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.315853989194002e-05,
+      "loss": 0.6865,
+      "step": 6098
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3156193447781576e-05,
+      "loss": 0.8506,
+      "step": 6099
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3153846684361e-05,
+      "loss": 1.1602,
+      "step": 6100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3151499601735255e-05,
+      "loss": 1.0654,
+      "step": 6101
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.314915219996129e-05,
+      "loss": 0.9385,
+      "step": 6102
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.314680447909607e-05,
+      "loss": 0.9785,
+      "step": 6103
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3144456439196565e-05,
+      "loss": 0.9385,
+      "step": 6104
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.314210808031977e-05,
+      "loss": 0.9355,
+      "step": 6105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3139759402522645e-05,
+      "loss": 1.0557,
+      "step": 6106
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.313741040586222e-05,
+      "loss": 1.0859,
+      "step": 6107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3135061090395465e-05,
+      "loss": 1.0537,
+      "step": 6108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.313271145617942e-05,
+      "loss": 1.0322,
+      "step": 6109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.313036150327108e-05,
+      "loss": 0.8379,
+      "step": 6110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.312801123172749e-05,
+      "loss": 0.7441,
+      "step": 6111
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.312566064160566e-05,
+      "loss": 0.9541,
+      "step": 6112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.312330973296266e-05,
+      "loss": 0.96,
+      "step": 6113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3120958505855525e-05,
+      "loss": 0.9287,
+      "step": 6114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.311860696034132e-05,
+      "loss": 0.7559,
+      "step": 6115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.31162550964771e-05,
+      "loss": 0.8799,
+      "step": 6116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.311390291431994e-05,
+      "loss": 0.8643,
+      "step": 6117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.311155041392692e-05,
+      "loss": 0.8115,
+      "step": 6118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3109197595355135e-05,
+      "loss": 0.8721,
+      "step": 6119
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.310684445866167e-05,
+      "loss": 0.9658,
+      "step": 6120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3104491003903646e-05,
+      "loss": 0.7529,
+      "step": 6121
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3102137231138165e-05,
+      "loss": 0.832,
+      "step": 6122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3099783140422346e-05,
+      "loss": 0.7954,
+      "step": 6123
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.309742873181331e-05,
+      "loss": 1.0566,
+      "step": 6124
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.309507400536819e-05,
+      "loss": 0.9434,
+      "step": 6125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.309271896114415e-05,
+      "loss": 0.9805,
+      "step": 6126
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3090363599198306e-05,
+      "loss": 0.9561,
+      "step": 6127
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.308800791958785e-05,
+      "loss": 1.0127,
+      "step": 6128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3085651922369935e-05,
+      "loss": 0.9951,
+      "step": 6129
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3083295607601726e-05,
+      "loss": 0.8242,
+      "step": 6130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.308093897534041e-05,
+      "loss": 1.0225,
+      "step": 6131
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3078582025643166e-05,
+      "loss": 0.9307,
+      "step": 6132
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3076224758567206e-05,
+      "loss": 1.0615,
+      "step": 6133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.307386717416973e-05,
+      "loss": 1.0615,
+      "step": 6134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.307150927250794e-05,
+      "loss": 1.1104,
+      "step": 6135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.306915105363907e-05,
+      "loss": 1.0654,
+      "step": 6136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3066792517620335e-05,
+      "loss": 1.0195,
+      "step": 6137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3064433664508974e-05,
+      "loss": 1.001,
+      "step": 6138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.306207449436223e-05,
+      "loss": 0.8701,
+      "step": 6139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.305971500723734e-05,
+      "loss": 1.0137,
+      "step": 6140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.305735520319159e-05,
+      "loss": 0.9482,
+      "step": 6141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3054995082282225e-05,
+      "loss": 0.9062,
+      "step": 6142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.305263464456652e-05,
+      "loss": 0.8984,
+      "step": 6143
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.305027389010176e-05,
+      "loss": 1.043,
+      "step": 6144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.304791281894523e-05,
+      "loss": 0.9238,
+      "step": 6145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.304555143115423e-05,
+      "loss": 1.1191,
+      "step": 6146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3043189726786055e-05,
+      "loss": 1.1123,
+      "step": 6147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.304082770589804e-05,
+      "loss": 0.8662,
+      "step": 6148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3038465368547474e-05,
+      "loss": 1.0928,
+      "step": 6149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.303610271479171e-05,
+      "loss": 0.9619,
+      "step": 6150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3033739744688066e-05,
+      "loss": 0.9248,
+      "step": 6151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.303137645829389e-05,
+      "loss": 0.8867,
+      "step": 6152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.302901285566653e-05,
+      "loss": 1.084,
+      "step": 6153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3026648936863344e-05,
+      "loss": 1.041,
+      "step": 6154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3024284701941703e-05,
+      "loss": 1.0303,
+      "step": 6155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.302192015095897e-05,
+      "loss": 1.0693,
+      "step": 6156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.301955528397253e-05,
+      "loss": 1.1152,
+      "step": 6157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3017190101039774e-05,
+      "loss": 0.9785,
+      "step": 6158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.301482460221809e-05,
+      "loss": 1.0078,
+      "step": 6159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.30124587875649e-05,
+      "loss": 0.9678,
+      "step": 6160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.30100926571376e-05,
+      "loss": 0.9805,
+      "step": 6161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.300772621099361e-05,
+      "loss": 1.0205,
+      "step": 6162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.300535944919037e-05,
+      "loss": 1.0088,
+      "step": 6163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3002992371785285e-05,
+      "loss": 0.9297,
+      "step": 6164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3000624978835824e-05,
+      "loss": 1.0781,
+      "step": 6165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2998257270399433e-05,
+      "loss": 0.8896,
+      "step": 6166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.299588924653356e-05,
+      "loss": 0.9473,
+      "step": 6167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.299352090729567e-05,
+      "loss": 0.9019,
+      "step": 6168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.299115225274325e-05,
+      "loss": 1.1016,
+      "step": 6169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.298878328293376e-05,
+      "loss": 0.9775,
+      "step": 6170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2986413997924706e-05,
+      "loss": 0.9424,
+      "step": 6171
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.298404439777358e-05,
+      "loss": 0.9336,
+      "step": 6172
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.298167448253788e-05,
+      "loss": 0.8018,
+      "step": 6173
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2979304252275105e-05,
+      "loss": 0.8896,
+      "step": 6174
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.29769337070428e-05,
+      "loss": 0.9131,
+      "step": 6175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.297456284689846e-05,
+      "loss": 1.1553,
+      "step": 6176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.297219167189966e-05,
+      "loss": 0.8447,
+      "step": 6177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2969820182103903e-05,
+      "loss": 0.96,
+      "step": 6178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.296744837756876e-05,
+      "loss": 0.9395,
+      "step": 6179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.296507625835178e-05,
+      "loss": 0.856,
+      "step": 6180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.296270382451052e-05,
+      "loss": 1.0674,
+      "step": 6181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2960331076102565e-05,
+      "loss": 1.0967,
+      "step": 6182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2957958013185495e-05,
+      "loss": 0.9365,
+      "step": 6183
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.295558463581689e-05,
+      "loss": 0.8975,
+      "step": 6184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2953210944054346e-05,
+      "loss": 1.0264,
+      "step": 6185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.295083693795546e-05,
+      "loss": 1.002,
+      "step": 6186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2948462617577855e-05,
+      "loss": 1.0469,
+      "step": 6187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.294608798297914e-05,
+      "loss": 1.1396,
+      "step": 6188
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.294371303421694e-05,
+      "loss": 1.083,
+      "step": 6189
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.29413377713489e-05,
+      "loss": 0.9492,
+      "step": 6190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.293896219443264e-05,
+      "loss": 0.8838,
+      "step": 6191
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.293658630352583e-05,
+      "loss": 0.8916,
+      "step": 6192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.293421009868611e-05,
+      "loss": 1.002,
+      "step": 6193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.293183357997115e-05,
+      "loss": 0.9844,
+      "step": 6194
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2929456747438613e-05,
+      "loss": 1.2051,
+      "step": 6195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2927079601146196e-05,
+      "loss": 0.9912,
+      "step": 6196
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2924702141151575e-05,
+      "loss": 0.9023,
+      "step": 6197
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.292232436751244e-05,
+      "loss": 0.877,
+      "step": 6198
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.291994628028649e-05,
+      "loss": 1.043,
+      "step": 6199
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.291756787953144e-05,
+      "loss": 0.8232,
+      "step": 6200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2915189165305015e-05,
+      "loss": 0.9062,
+      "step": 6201
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2912810137664926e-05,
+      "loss": 1.0244,
+      "step": 6202
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2910430796668915e-05,
+      "loss": 1.0176,
+      "step": 6203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.290805114237471e-05,
+      "loss": 0.7412,
+      "step": 6204
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.290567117484007e-05,
+      "loss": 0.9141,
+      "step": 6205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.290329089412274e-05,
+      "loss": 1.0938,
+      "step": 6206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2900910300280494e-05,
+      "loss": 0.9561,
+      "step": 6207
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.289852939337109e-05,
+      "loss": 0.9033,
+      "step": 6208
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.289614817345231e-05,
+      "loss": 0.7803,
+      "step": 6209
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.289376664058194e-05,
+      "loss": 0.9346,
+      "step": 6210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.289138479481778e-05,
+      "loss": 0.9268,
+      "step": 6211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.288900263621762e-05,
+      "loss": 0.958,
+      "step": 6212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.288662016483927e-05,
+      "loss": 0.7812,
+      "step": 6213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2884237380740545e-05,
+      "loss": 0.9541,
+      "step": 6214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.288185428397927e-05,
+      "loss": 0.9189,
+      "step": 6215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2879470874613276e-05,
+      "loss": 1.1123,
+      "step": 6216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2877087152700405e-05,
+      "loss": 0.7822,
+      "step": 6217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.287470311829849e-05,
+      "loss": 1.0127,
+      "step": 6218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.28723187714654e-05,
+      "loss": 1.0273,
+      "step": 6219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.286993411225899e-05,
+      "loss": 0.9316,
+      "step": 6220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.286754914073713e-05,
+      "loss": 1.1016,
+      "step": 6221
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.286516385695769e-05,
+      "loss": 1.0273,
+      "step": 6222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.286277826097856e-05,
+      "loss": 0.9336,
+      "step": 6223
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2860392352857635e-05,
+      "loss": 1.0576,
+      "step": 6224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.28580061326528e-05,
+      "loss": 1.002,
+      "step": 6225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.285561960042198e-05,
+      "loss": 0.8721,
+      "step": 6226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.285323275622307e-05,
+      "loss": 0.7783,
+      "step": 6227
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2850845600114004e-05,
+      "loss": 1.0869,
+      "step": 6228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2848458132152706e-05,
+      "loss": 1.0264,
+      "step": 6229
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2846070352397116e-05,
+      "loss": 0.9683,
+      "step": 6230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.284368226090518e-05,
+      "loss": 1.041,
+      "step": 6231
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2841293857734844e-05,
+      "loss": 1.0654,
+      "step": 6232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.283890514294407e-05,
+      "loss": 0.9141,
+      "step": 6233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2836516116590826e-05,
+      "loss": 0.8643,
+      "step": 6234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.283412677873309e-05,
+      "loss": 0.9053,
+      "step": 6235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2831737129428844e-05,
+      "loss": 1.001,
+      "step": 6236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.282934716873607e-05,
+      "loss": 0.8418,
+      "step": 6237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2826956896712774e-05,
+      "loss": 0.9919,
+      "step": 6238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2824566313416954e-05,
+      "loss": 0.8613,
+      "step": 6239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2822175418906626e-05,
+      "loss": 0.8691,
+      "step": 6240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.281978421323981e-05,
+      "loss": 1.0449,
+      "step": 6241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.281739269647453e-05,
+      "loss": 1.0078,
+      "step": 6242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.281500086866882e-05,
+      "loss": 0.9248,
+      "step": 6243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2812608729880735e-05,
+      "loss": 0.9155,
+      "step": 6244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2810216280168314e-05,
+      "loss": 0.79,
+      "step": 6245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.280782351958962e-05,
+      "loss": 0.9785,
+      "step": 6246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.280543044820271e-05,
+      "loss": 0.8984,
+      "step": 6247
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.280303706606566e-05,
+      "loss": 0.9873,
+      "step": 6248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.280064337323656e-05,
+      "loss": 0.8887,
+      "step": 6249
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.279824936977349e-05,
+      "loss": 0.8877,
+      "step": 6250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2795855055734543e-05,
+      "loss": 1.1172,
+      "step": 6251
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2793460431177827e-05,
+      "loss": 1.0713,
+      "step": 6252
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.279106549616145e-05,
+      "loss": 1.002,
+      "step": 6253
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2788670250743536e-05,
+      "loss": 1.0215,
+      "step": 6254
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2786274694982195e-05,
+      "loss": 0.9873,
+      "step": 6255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2783878828935576e-05,
+      "loss": 1.0518,
+      "step": 6256
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.278148265266182e-05,
+      "loss": 1.0391,
+      "step": 6257
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2779086166219056e-05,
+      "loss": 0.9961,
+      "step": 6258
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.277668936966546e-05,
+      "loss": 1.084,
+      "step": 6259
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.277429226305919e-05,
+      "loss": 0.9346,
+      "step": 6260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.27718948464584e-05,
+      "loss": 0.75,
+      "step": 6261
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2769497119921295e-05,
+      "loss": 1.0791,
+      "step": 6262
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.276709908350605e-05,
+      "loss": 0.752,
+      "step": 6263
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2764700737270855e-05,
+      "loss": 0.8945,
+      "step": 6264
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.276230208127391e-05,
+      "loss": 0.9082,
+      "step": 6265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.275990311557343e-05,
+      "loss": 0.9961,
+      "step": 6266
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.275750384022763e-05,
+      "loss": 1.1523,
+      "step": 6267
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.275510425529472e-05,
+      "loss": 1.0527,
+      "step": 6268
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.275270436083295e-05,
+      "loss": 1.0439,
+      "step": 6269
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.275030415690054e-05,
+      "loss": 1.0566,
+      "step": 6270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.274790364355576e-05,
+      "loss": 0.8789,
+      "step": 6271
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2745502820856836e-05,
+      "loss": 0.9863,
+      "step": 6272
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2743101688862044e-05,
+      "loss": 1.0049,
+      "step": 6273
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.274070024762965e-05,
+      "loss": 1.1211,
+      "step": 6274
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2738298497217934e-05,
+      "loss": 1.0557,
+      "step": 6275
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2735896437685175e-05,
+      "loss": 0.9668,
+      "step": 6276
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2733494069089656e-05,
+      "loss": 0.9785,
+      "step": 6277
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2731091391489695e-05,
+      "loss": 0.9512,
+      "step": 6278
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.272868840494359e-05,
+      "loss": 0.7847,
+      "step": 6279
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2726285109509635e-05,
+      "loss": 1.043,
+      "step": 6280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.272388150524618e-05,
+      "loss": 0.8975,
+      "step": 6281
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2721477592211534e-05,
+      "loss": 0.8604,
+      "step": 6282
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2719073370464045e-05,
+      "loss": 0.9893,
+      "step": 6283
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2716668840062046e-05,
+      "loss": 0.9209,
+      "step": 6284
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.271426400106389e-05,
+      "loss": 1.0195,
+      "step": 6285
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.271185885352795e-05,
+      "loss": 1.0029,
+      "step": 6286
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.270945339751257e-05,
+      "loss": 0.835,
+      "step": 6287
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.270704763307613e-05,
+      "loss": 1.0254,
+      "step": 6288
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.270464156027702e-05,
+      "loss": 0.9551,
+      "step": 6289
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.270223517917362e-05,
+      "loss": 0.9502,
+      "step": 6290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.269982848982434e-05,
+      "loss": 1.0371,
+      "step": 6291
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.269742149228756e-05,
+      "loss": 0.9453,
+      "step": 6292
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2695014186621703e-05,
+      "loss": 0.7451,
+      "step": 6293
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.269260657288518e-05,
+      "loss": 0.8379,
+      "step": 6294
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2690198651136434e-05,
+      "loss": 1.0137,
+      "step": 6295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.268779042143388e-05,
+      "loss": 1.082,
+      "step": 6296
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.268538188383597e-05,
+      "loss": 0.873,
+      "step": 6297
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2682973038401145e-05,
+      "loss": 0.9307,
+      "step": 6298
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2680563885187864e-05,
+      "loss": 1.0605,
+      "step": 6299
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.267815442425459e-05,
+      "loss": 0.8789,
+      "step": 6300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2675744655659795e-05,
+      "loss": 1.0244,
+      "step": 6301
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.267333457946196e-05,
+      "loss": 1.0771,
+      "step": 6302
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.267092419571956e-05,
+      "loss": 0.9463,
+      "step": 6303
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.266851350449109e-05,
+      "loss": 1.1279,
+      "step": 6304
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.266610250583506e-05,
+      "loss": 1.0469,
+      "step": 6305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.266369119980997e-05,
+      "loss": 0.6992,
+      "step": 6306
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.266127958647434e-05,
+      "loss": 1.0723,
+      "step": 6307
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.265886766588669e-05,
+      "loss": 0.9629,
+      "step": 6308
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.265645543810554e-05,
+      "loss": 0.9482,
+      "step": 6309
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.265404290318945e-05,
+      "loss": 0.8657,
+      "step": 6310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2651630061196944e-05,
+      "loss": 0.9033,
+      "step": 6311
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.264921691218658e-05,
+      "loss": 0.9141,
+      "step": 6312
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2646803456216935e-05,
+      "loss": 0.834,
+      "step": 6313
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2644389693346556e-05,
+      "loss": 0.9023,
+      "step": 6314
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2641975623634027e-05,
+      "loss": 1.1143,
+      "step": 6315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.263956124713793e-05,
+      "loss": 0.918,
+      "step": 6316
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2637146563916845e-05,
+      "loss": 1.0742,
+      "step": 6317
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.263473157402938e-05,
+      "loss": 0.9766,
+      "step": 6318
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.263231627753414e-05,
+      "loss": 0.9209,
+      "step": 6319
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.262990067448973e-05,
+      "loss": 0.9648,
+      "step": 6320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.262748476495478e-05,
+      "loss": 0.917,
+      "step": 6321
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.262506854898791e-05,
+      "loss": 0.9424,
+      "step": 6322
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2622652026647755e-05,
+      "loss": 1.0977,
+      "step": 6323
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2620235197992954e-05,
+      "loss": 0.9219,
+      "step": 6324
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.261781806308216e-05,
+      "loss": 1.0078,
+      "step": 6325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2615400621974036e-05,
+      "loss": 1.1211,
+      "step": 6326
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.261298287472723e-05,
+      "loss": 0.8506,
+      "step": 6327
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.261056482140043e-05,
+      "loss": 0.9707,
+      "step": 6328
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.260814646205231e-05,
+      "loss": 1.0088,
+      "step": 6329
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2605727796741544e-05,
+      "loss": 0.9463,
+      "step": 6330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.260330882552685e-05,
+      "loss": 1.0781,
+      "step": 6331
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.26008895484669e-05,
+      "loss": 1.0498,
+      "step": 6332
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.259846996562043e-05,
+      "loss": 1.0186,
+      "step": 6333
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2596050077046145e-05,
+      "loss": 1.1182,
+      "step": 6334
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.259362988280277e-05,
+      "loss": 0.8252,
+      "step": 6335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.259120938294903e-05,
+      "loss": 0.8052,
+      "step": 6336
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2588788577543665e-05,
+      "loss": 0.9805,
+      "step": 6337
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.258636746664542e-05,
+      "loss": 1.0117,
+      "step": 6338
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.258394605031306e-05,
+      "loss": 1.04,
+      "step": 6339
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.258152432860533e-05,
+      "loss": 1.0137,
+      "step": 6340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.257910230158102e-05,
+      "loss": 0.7549,
+      "step": 6341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.257667996929888e-05,
+      "loss": 0.9424,
+      "step": 6342
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.25742573318177e-05,
+      "loss": 1.0186,
+      "step": 6343
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.257183438919628e-05,
+      "loss": 0.8916,
+      "step": 6344
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2569411141493415e-05,
+      "loss": 0.9561,
+      "step": 6345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2566987588767905e-05,
+      "loss": 0.874,
+      "step": 6346
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.256456373107856e-05,
+      "loss": 1.04,
+      "step": 6347
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.25621395684842e-05,
+      "loss": 0.9375,
+      "step": 6348
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2559715101043665e-05,
+      "loss": 1.0801,
+      "step": 6349
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2557290328815785e-05,
+      "loss": 1.0186,
+      "step": 6350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.255486525185939e-05,
+      "loss": 1.0342,
+      "step": 6351
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.255243987023334e-05,
+      "loss": 0.8516,
+      "step": 6352
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.255001418399649e-05,
+      "loss": 0.9277,
+      "step": 6353
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.254758819320771e-05,
+      "loss": 0.9219,
+      "step": 6354
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.254516189792585e-05,
+      "loss": 0.8896,
+      "step": 6355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.254273529820981e-05,
+      "loss": 0.8936,
+      "step": 6356
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.254030839411848e-05,
+      "loss": 0.9365,
+      "step": 6357
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.253788118571074e-05,
+      "loss": 0.8447,
+      "step": 6358
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.25354536730455e-05,
+      "loss": 0.8516,
+      "step": 6359
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2533025856181666e-05,
+      "loss": 1.0938,
+      "step": 6360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.253059773517815e-05,
+      "loss": 0.73,
+      "step": 6361
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2528169310093876e-05,
+      "loss": 1.042,
+      "step": 6362
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2525740580987775e-05,
+      "loss": 1.0918,
+      "step": 6363
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.252331154791879e-05,
+      "loss": 0.9834,
+      "step": 6364
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2520882210945864e-05,
+      "loss": 0.9893,
+      "step": 6365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.251845257012795e-05,
+      "loss": 0.8906,
+      "step": 6366
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2516022625524005e-05,
+      "loss": 0.8525,
+      "step": 6367
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2513592377193e-05,
+      "loss": 0.9844,
+      "step": 6368
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2511161825193905e-05,
+      "loss": 0.8516,
+      "step": 6369
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.250873096958572e-05,
+      "loss": 0.958,
+      "step": 6370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2506299810427403e-05,
+      "loss": 1.0459,
+      "step": 6371
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2503868347777984e-05,
+      "loss": 1.0156,
+      "step": 6372
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2501436581696446e-05,
+      "loss": 0.8516,
+      "step": 6373
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.249900451224181e-05,
+      "loss": 0.9658,
+      "step": 6374
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.249657213947309e-05,
+      "loss": 1.0068,
+      "step": 6375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.249413946344932e-05,
+      "loss": 0.9912,
+      "step": 6376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.249170648422952e-05,
+      "loss": 0.9443,
+      "step": 6377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.248927320187275e-05,
+      "loss": 0.9834,
+      "step": 6378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.248683961643805e-05,
+      "loss": 1.0361,
+      "step": 6379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.248440572798447e-05,
+      "loss": 0.8369,
+      "step": 6380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.248197153657108e-05,
+      "loss": 1.0391,
+      "step": 6381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.247953704225694e-05,
+      "loss": 0.9209,
+      "step": 6382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.247710224510115e-05,
+      "loss": 0.8931,
+      "step": 6383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.247466714516278e-05,
+      "loss": 0.8135,
+      "step": 6384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.247223174250092e-05,
+      "loss": 0.9834,
+      "step": 6385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.246979603717467e-05,
+      "loss": 0.9336,
+      "step": 6386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.246736002924315e-05,
+      "loss": 1.1104,
+      "step": 6387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.246492371876547e-05,
+      "loss": 1.0957,
+      "step": 6388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2462487105800745e-05,
+      "loss": 1.0205,
+      "step": 6389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2460050190408115e-05,
+      "loss": 1.0957,
+      "step": 6390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2457612972646705e-05,
+      "loss": 0.96,
+      "step": 6391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2455175452575666e-05,
+      "loss": 0.9014,
+      "step": 6392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2452737630254146e-05,
+      "loss": 0.8691,
+      "step": 6393
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.24502995057413e-05,
+      "loss": 0.8848,
+      "step": 6394
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2447861079096314e-05,
+      "loss": 1.0801,
+      "step": 6395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.244542235037834e-05,
+      "loss": 0.9531,
+      "step": 6396
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2442983319646566e-05,
+      "loss": 0.9756,
+      "step": 6397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.244054398696018e-05,
+      "loss": 0.958,
+      "step": 6398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2438104352378386e-05,
+      "loss": 1.1016,
+      "step": 6399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.243566441596037e-05,
+      "loss": 1.0605,
+      "step": 6400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.243322417776535e-05,
+      "loss": 0.9248,
+      "step": 6401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2430783637852545e-05,
+      "loss": 0.918,
+      "step": 6402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2428342796281183e-05,
+      "loss": 0.9199,
+      "step": 6403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.242590165311049e-05,
+      "loss": 1.0254,
+      "step": 6404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2423460208399707e-05,
+      "loss": 0.7578,
+      "step": 6405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2421018462208074e-05,
+      "loss": 1.0244,
+      "step": 6406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.241857641459486e-05,
+      "loss": 1.0635,
+      "step": 6407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.241613406561931e-05,
+      "loss": 0.9268,
+      "step": 6408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.241369141534071e-05,
+      "loss": 0.9834,
+      "step": 6409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2411248463818325e-05,
+      "loss": 0.9199,
+      "step": 6410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2408805211111436e-05,
+      "loss": 0.874,
+      "step": 6411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.240636165727934e-05,
+      "loss": 1.1123,
+      "step": 6412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.240391780238134e-05,
+      "loss": 1.1426,
+      "step": 6413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.240147364647672e-05,
+      "loss": 0.873,
+      "step": 6414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.239902918962481e-05,
+      "loss": 0.8262,
+      "step": 6415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2396584431884926e-05,
+      "loss": 1.0088,
+      "step": 6416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.23941393733164e-05,
+      "loss": 1.0879,
+      "step": 6417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.239169401397856e-05,
+      "loss": 0.9707,
+      "step": 6418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.238924835393074e-05,
+      "loss": 0.9189,
+      "step": 6419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2386802393232307e-05,
+      "loss": 0.9619,
+      "step": 6420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.238435613194261e-05,
+      "loss": 1.0459,
+      "step": 6421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2381909570121004e-05,
+      "loss": 0.7217,
+      "step": 6422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.237946270782687e-05,
+      "loss": 1.0723,
+      "step": 6423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.237701554511958e-05,
+      "loss": 1.0752,
+      "step": 6424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2374568082058525e-05,
+      "loss": 1.0771,
+      "step": 6425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2372120318703095e-05,
+      "loss": 1.0674,
+      "step": 6426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.236967225511269e-05,
+      "loss": 0.9717,
+      "step": 6427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2367223891346716e-05,
+      "loss": 1.0801,
+      "step": 6428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.236477522746459e-05,
+      "loss": 1.0303,
+      "step": 6429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.236232626352573e-05,
+      "loss": 1.0371,
+      "step": 6430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.235987699958957e-05,
+      "loss": 0.959,
+      "step": 6431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.235742743571555e-05,
+      "loss": 0.999,
+      "step": 6432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.23549775719631e-05,
+      "loss": 0.9893,
+      "step": 6433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2352527408391683e-05,
+      "loss": 1.04,
+      "step": 6434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.235007694506075e-05,
+      "loss": 1.1309,
+      "step": 6435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.234762618202977e-05,
+      "loss": 1.0547,
+      "step": 6436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.234517511935821e-05,
+      "loss": 0.8994,
+      "step": 6437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.234272375710556e-05,
+      "loss": 0.9912,
+      "step": 6438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.234027209533131e-05,
+      "loss": 1.0693,
+      "step": 6439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.233782013409494e-05,
+      "loss": 0.8672,
+      "step": 6440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.233536787345596e-05,
+      "loss": 1.0449,
+      "step": 6441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.233291531347389e-05,
+      "loss": 0.938,
+      "step": 6442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2330462454208226e-05,
+      "loss": 0.96,
+      "step": 6443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.23280092957185e-05,
+      "loss": 0.8721,
+      "step": 6444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.232555583806424e-05,
+      "loss": 1.1182,
+      "step": 6445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2323102081305e-05,
+      "loss": 0.8137,
+      "step": 6446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.232064802550031e-05,
+      "loss": 1.041,
+      "step": 6447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2318193670709716e-05,
+      "loss": 0.7949,
+      "step": 6448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2315739016992795e-05,
+      "loss": 0.9111,
+      "step": 6449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.231328406440911e-05,
+      "loss": 0.9619,
+      "step": 6450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2310828813018224e-05,
+      "loss": 1.042,
+      "step": 6451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.230837326287974e-05,
+      "loss": 0.8711,
+      "step": 6452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.230591741405323e-05,
+      "loss": 1.1465,
+      "step": 6453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.230346126659829e-05,
+      "loss": 0.9795,
+      "step": 6454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.230100482057453e-05,
+      "loss": 0.9502,
+      "step": 6455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.229854807604156e-05,
+      "loss": 0.9561,
+      "step": 6456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2296091033059e-05,
+      "loss": 1.0771,
+      "step": 6457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.229363369168647e-05,
+      "loss": 1.0654,
+      "step": 6458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.22911760519836e-05,
+      "loss": 1.0127,
+      "step": 6459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2288718114010036e-05,
+      "loss": 1.0469,
+      "step": 6460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2286259877825425e-05,
+      "loss": 0.958,
+      "step": 6461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2283801343489424e-05,
+      "loss": 0.9648,
+      "step": 6462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2281342511061686e-05,
+      "loss": 0.9805,
+      "step": 6463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.227888338060188e-05,
+      "loss": 0.9697,
+      "step": 6464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.227642395216969e-05,
+      "loss": 0.7183,
+      "step": 6465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.227396422582479e-05,
+      "loss": 1.1387,
+      "step": 6466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.227150420162688e-05,
+      "loss": 0.9854,
+      "step": 6467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.226904387963565e-05,
+      "loss": 0.8271,
+      "step": 6468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.22665832599108e-05,
+      "loss": 1.1758,
+      "step": 6469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.226412234251206e-05,
+      "loss": 1.0664,
+      "step": 6470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.226166112749913e-05,
+      "loss": 0.8525,
+      "step": 6471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2259199614931745e-05,
+      "loss": 0.8848,
+      "step": 6472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2256737804869646e-05,
+      "loss": 0.7969,
+      "step": 6473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.225427569737256e-05,
+      "loss": 0.9951,
+      "step": 6474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.225181329250025e-05,
+      "loss": 0.8809,
+      "step": 6475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.224935059031245e-05,
+      "loss": 1.0439,
+      "step": 6476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.224688759086894e-05,
+      "loss": 0.9707,
+      "step": 6477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.224442429422949e-05,
+      "loss": 0.9736,
+      "step": 6478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2241960700453866e-05,
+      "loss": 1.085,
+      "step": 6479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2239496809601854e-05,
+      "loss": 0.6689,
+      "step": 6480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.223703262173326e-05,
+      "loss": 0.9102,
+      "step": 6481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2234568136907864e-05,
+      "loss": 0.8809,
+      "step": 6482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2232103355185486e-05,
+      "loss": 1.0684,
+      "step": 6483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.222963827662593e-05,
+      "loss": 0.9736,
+      "step": 6484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.222717290128902e-05,
+      "loss": 0.874,
+      "step": 6485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.222470722923458e-05,
+      "loss": 0.916,
+      "step": 6486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.222224126052245e-05,
+      "loss": 0.9561,
+      "step": 6487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2219774995212466e-05,
+      "loss": 0.917,
+      "step": 6488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.221730843336449e-05,
+      "loss": 0.9531,
+      "step": 6489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2214841575038356e-05,
+      "loss": 0.9229,
+      "step": 6490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.221237442029394e-05,
+      "loss": 0.9814,
+      "step": 6491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.220990696919112e-05,
+      "loss": 0.8965,
+      "step": 6492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2207439221789756e-05,
+      "loss": 1.0107,
+      "step": 6493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.220497117814975e-05,
+      "loss": 0.9521,
+      "step": 6494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2202502838330986e-05,
+      "loss": 1.085,
+      "step": 6495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.220003420239337e-05,
+      "loss": 1.0293,
+      "step": 6496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.219756527039679e-05,
+      "loss": 1.1016,
+      "step": 6497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2195096042401184e-05,
+      "loss": 0.8252,
+      "step": 6498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.219262651846645e-05,
+      "loss": 1.0693,
+      "step": 6499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2190156698652536e-05,
+      "loss": 0.9893,
+      "step": 6500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.218768658301937e-05,
+      "loss": 0.8955,
+      "step": 6501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2185216171626895e-05,
+      "loss": 0.9912,
+      "step": 6502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.218274546453505e-05,
+      "loss": 0.7734,
+      "step": 6503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.218027446180381e-05,
+      "loss": 0.9854,
+      "step": 6504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2177803163493124e-05,
+      "loss": 0.8301,
+      "step": 6505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2175331569662964e-05,
+      "loss": 0.7617,
+      "step": 6506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2172859680373326e-05,
+      "loss": 0.998,
+      "step": 6507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.217038749568417e-05,
+      "loss": 1.0068,
+      "step": 6508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2167915015655515e-05,
+      "loss": 1.041,
+      "step": 6509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.216544224034734e-05,
+      "loss": 0.9453,
+      "step": 6510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2162969169819656e-05,
+      "loss": 1.1191,
+      "step": 6511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.216049580413248e-05,
+      "loss": 0.8779,
+      "step": 6512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2158022143345845e-05,
+      "loss": 0.8623,
+      "step": 6513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.215554818751975e-05,
+      "loss": 1.0781,
+      "step": 6514
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.215307393671426e-05,
+      "loss": 0.8838,
+      "step": 6515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2150599390989406e-05,
+      "loss": 0.9609,
+      "step": 6516
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.214812455040524e-05,
+      "loss": 0.959,
+      "step": 6517
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.214564941502181e-05,
+      "loss": 1.0508,
+      "step": 6518
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2143173984899195e-05,
+      "loss": 0.8975,
+      "step": 6519
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.214069826009745e-05,
+      "loss": 0.9209,
+      "step": 6520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.213822224067667e-05,
+      "loss": 0.9941,
+      "step": 6521
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.213574592669693e-05,
+      "loss": 0.8164,
+      "step": 6522
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.213326931821833e-05,
+      "loss": 1.0107,
+      "step": 6523
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2130792415300964e-05,
+      "loss": 0.9346,
+      "step": 6524
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2128315218004945e-05,
+      "loss": 0.8857,
+      "step": 6525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.212583772639038e-05,
+      "loss": 1.0732,
+      "step": 6526
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.21233599405174e-05,
+      "loss": 0.8232,
+      "step": 6527
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.212088186044612e-05,
+      "loss": 0.876,
+      "step": 6528
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2118403486236684e-05,
+      "loss": 0.9521,
+      "step": 6529
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.211592481794924e-05,
+      "loss": 1.0254,
+      "step": 6530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.211344585564393e-05,
+      "loss": 1.0801,
+      "step": 6531
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2110966599380915e-05,
+      "loss": 0.9531,
+      "step": 6532
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.210848704922035e-05,
+      "loss": 0.7998,
+      "step": 6533
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.210600720522242e-05,
+      "loss": 0.874,
+      "step": 6534
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.21035270674473e-05,
+      "loss": 0.9111,
+      "step": 6535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.210104663595517e-05,
+      "loss": 0.9697,
+      "step": 6536
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.209856591080623e-05,
+      "loss": 0.959,
+      "step": 6537
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.209608489206067e-05,
+      "loss": 0.9375,
+      "step": 6538
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2093603579778706e-05,
+      "loss": 0.9756,
+      "step": 6539
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.209112197402055e-05,
+      "loss": 1.1484,
+      "step": 6540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.208864007484642e-05,
+      "loss": 0.9355,
+      "step": 6541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2086157882316545e-05,
+      "loss": 1.0527,
+      "step": 6542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2083675396491167e-05,
+      "loss": 0.8848,
+      "step": 6543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.208119261743052e-05,
+      "loss": 1.0078,
+      "step": 6544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2078709545194864e-05,
+      "loss": 0.9229,
+      "step": 6545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.207622617984445e-05,
+      "loss": 0.8359,
+      "step": 6546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.207374252143953e-05,
+      "loss": 1.0576,
+      "step": 6547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.20712585700404e-05,
+      "loss": 0.9629,
+      "step": 6548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.206877432570731e-05,
+      "loss": 0.998,
+      "step": 6549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.206628978850058e-05,
+      "loss": 0.9902,
+      "step": 6550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.206380495848046e-05,
+      "loss": 0.8652,
+      "step": 6551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2061319835707286e-05,
+      "loss": 0.9912,
+      "step": 6552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.205883442024135e-05,
+      "loss": 0.877,
+      "step": 6553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.205634871214296e-05,
+      "loss": 1.0625,
+      "step": 6554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.205386271147245e-05,
+      "loss": 0.7856,
+      "step": 6555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2051376418290135e-05,
+      "loss": 0.9346,
+      "step": 6556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.204888983265636e-05,
+      "loss": 0.9551,
+      "step": 6557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.204640295463146e-05,
+      "loss": 0.9951,
+      "step": 6558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.204391578427579e-05,
+      "loss": 0.9971,
+      "step": 6559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.204142832164971e-05,
+      "loss": 1.0371,
+      "step": 6560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.203894056681357e-05,
+      "loss": 0.9844,
+      "step": 6561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2036452519827744e-05,
+      "loss": 1.0205,
+      "step": 6562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2033964180752616e-05,
+      "loss": 0.9414,
+      "step": 6563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2031475549648575e-05,
+      "loss": 0.8711,
+      "step": 6564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2028986626576e-05,
+      "loss": 0.7197,
+      "step": 6565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.20264974115953e-05,
+      "loss": 0.9307,
+      "step": 6566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.202400790476687e-05,
+      "loss": 0.9424,
+      "step": 6567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2021518106151125e-05,
+      "loss": 1.0,
+      "step": 6568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2019028015808504e-05,
+      "loss": 1.0117,
+      "step": 6569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.201653763379941e-05,
+      "loss": 0.8135,
+      "step": 6570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.201404696018429e-05,
+      "loss": 0.9727,
+      "step": 6571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.201155599502358e-05,
+      "loss": 0.9697,
+      "step": 6572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2009064738377736e-05,
+      "loss": 0.8867,
+      "step": 6573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.200657319030721e-05,
+      "loss": 0.7354,
+      "step": 6574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.200408135087245e-05,
+      "loss": 0.9756,
+      "step": 6575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.200158922013395e-05,
+      "loss": 0.96,
+      "step": 6576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1999096798152166e-05,
+      "loss": 0.9043,
+      "step": 6577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.19966040849876e-05,
+      "loss": 0.9805,
+      "step": 6578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.199411108070073e-05,
+      "loss": 0.9727,
+      "step": 6579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1991617785352056e-05,
+      "loss": 1.0186,
+      "step": 6580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1989124199002084e-05,
+      "loss": 0.9697,
+      "step": 6581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1986630321711325e-05,
+      "loss": 1.0264,
+      "step": 6582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.19841361535403e-05,
+      "loss": 0.915,
+      "step": 6583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.198164169454954e-05,
+      "loss": 1.0957,
+      "step": 6584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.197914694479957e-05,
+      "loss": 1.0508,
+      "step": 6585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1976651904350936e-05,
+      "loss": 0.8975,
+      "step": 6586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1974156573264176e-05,
+      "loss": 0.8516,
+      "step": 6587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.197166095159985e-05,
+      "loss": 0.9844,
+      "step": 6588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.196916503941853e-05,
+      "loss": 1.084,
+      "step": 6589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1966668836780765e-05,
+      "loss": 1.0645,
+      "step": 6590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.196417234374715e-05,
+      "loss": 1.1016,
+      "step": 6591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.196167556037825e-05,
+      "loss": 0.9346,
+      "step": 6592
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.195917848673466e-05,
+      "loss": 0.9043,
+      "step": 6593
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.195668112287698e-05,
+      "loss": 1.0762,
+      "step": 6594
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.195418346886582e-05,
+      "loss": 0.8008,
+      "step": 6595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.195168552476178e-05,
+      "loss": 1.0029,
+      "step": 6596
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1949187290625475e-05,
+      "loss": 0.9424,
+      "step": 6597
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.194668876651754e-05,
+      "loss": 1.0088,
+      "step": 6598
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.19441899524986e-05,
+      "loss": 1.0996,
+      "step": 6599
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.19416908486293e-05,
+      "loss": 0.9746,
+      "step": 6600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.193919145497028e-05,
+      "loss": 0.8887,
+      "step": 6601
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.19366917715822e-05,
+      "loss": 1.1113,
+      "step": 6602
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.193419179852571e-05,
+      "loss": 0.8809,
+      "step": 6603
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1931691535861476e-05,
+      "loss": 0.9082,
+      "step": 6604
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1929190983650185e-05,
+      "loss": 0.9922,
+      "step": 6605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.192669014195252e-05,
+      "loss": 1.0049,
+      "step": 6606
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1924189010829145e-05,
+      "loss": 0.751,
+      "step": 6607
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.192168759034078e-05,
+      "loss": 1.0039,
+      "step": 6608
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.191918588054811e-05,
+      "loss": 1.0859,
+      "step": 6609
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.191668388151186e-05,
+      "loss": 0.9541,
+      "step": 6610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1914181593292744e-05,
+      "loss": 1.001,
+      "step": 6611
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1911679015951464e-05,
+      "loss": 1.0635,
+      "step": 6612
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.190917614954877e-05,
+      "loss": 1.0557,
+      "step": 6613
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.19066729941454e-05,
+      "loss": 0.9062,
+      "step": 6614
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.190416954980209e-05,
+      "loss": 1.0762,
+      "step": 6615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.190166581657959e-05,
+      "loss": 0.9678,
+      "step": 6616
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1899161794538664e-05,
+      "loss": 0.8555,
+      "step": 6617
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.189665748374008e-05,
+      "loss": 0.7627,
+      "step": 6618
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.18941528842446e-05,
+      "loss": 1.0215,
+      "step": 6619
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.189164799611301e-05,
+      "loss": 0.876,
+      "step": 6620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.18891428194061e-05,
+      "loss": 0.9717,
+      "step": 6621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1886637354184654e-05,
+      "loss": 1.0596,
+      "step": 6622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1884131600509476e-05,
+      "loss": 0.9834,
+      "step": 6623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.188162555844137e-05,
+      "loss": 0.9199,
+      "step": 6624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.187911922804117e-05,
+      "loss": 1.2578,
+      "step": 6625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.187661260936967e-05,
+      "loss": 0.9414,
+      "step": 6626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.187410570248771e-05,
+      "loss": 0.9746,
+      "step": 6627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.187159850745612e-05,
+      "loss": 0.8965,
+      "step": 6628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1869091024335756e-05,
+      "loss": 0.9453,
+      "step": 6629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.186658325318746e-05,
+      "loss": 0.9697,
+      "step": 6630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1864075194072084e-05,
+      "loss": 0.8916,
+      "step": 6631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.186156684705049e-05,
+      "loss": 0.8867,
+      "step": 6632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1859058212183555e-05,
+      "loss": 0.9648,
+      "step": 6633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1856549289532155e-05,
+      "loss": 0.9805,
+      "step": 6634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1854040079157173e-05,
+      "loss": 0.9844,
+      "step": 6635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.18515305811195e-05,
+      "loss": 1.0176,
+      "step": 6636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.184902079548003e-05,
+      "loss": 0.8154,
+      "step": 6637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.184651072229967e-05,
+      "loss": 0.8467,
+      "step": 6638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.184400036163934e-05,
+      "loss": 1.0459,
+      "step": 6639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.184148971355995e-05,
+      "loss": 0.9521,
+      "step": 6640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.183897877812243e-05,
+      "loss": 1.1211,
+      "step": 6641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1836467555387715e-05,
+      "loss": 0.9902,
+      "step": 6642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.183395604541674e-05,
+      "loss": 0.9004,
+      "step": 6643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1831444248270455e-05,
+      "loss": 0.9629,
+      "step": 6644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.182893216400981e-05,
+      "loss": 0.8936,
+      "step": 6645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.182641979269577e-05,
+      "loss": 0.8164,
+      "step": 6646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.18239071343893e-05,
+      "loss": 0.9424,
+      "step": 6647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1821394189151385e-05,
+      "loss": 1.1016,
+      "step": 6648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.181888095704299e-05,
+      "loss": 0.957,
+      "step": 6649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1816367438125114e-05,
+      "loss": 0.9697,
+      "step": 6650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.181385363245875e-05,
+      "loss": 0.9844,
+      "step": 6651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1811339540104906e-05,
+      "loss": 1.0479,
+      "step": 6652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.180882516112459e-05,
+      "loss": 0.9551,
+      "step": 6653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1806310495578806e-05,
+      "loss": 1.0684,
+      "step": 6654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1803795543528596e-05,
+      "loss": 0.9893,
+      "step": 6655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.180128030503498e-05,
+      "loss": 0.9375,
+      "step": 6656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.179876478015899e-05,
+      "loss": 0.8398,
+      "step": 6657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.179624896896169e-05,
+      "loss": 0.9268,
+      "step": 6658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.179373287150411e-05,
+      "loss": 0.9756,
+      "step": 6659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1791216487847323e-05,
+      "loss": 1.0693,
+      "step": 6660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.178869981805238e-05,
+      "loss": 1.0918,
+      "step": 6661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.178618286218037e-05,
+      "loss": 0.9414,
+      "step": 6662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.178366562029236e-05,
+      "loss": 0.8682,
+      "step": 6663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.178114809244944e-05,
+      "loss": 1.082,
+      "step": 6664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1778630278712705e-05,
+      "loss": 0.9238,
+      "step": 6665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1776112179143255e-05,
+      "loss": 0.8125,
+      "step": 6666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.177359379380219e-05,
+      "loss": 0.9932,
+      "step": 6667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.177107512275063e-05,
+      "loss": 0.9932,
+      "step": 6668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1768556166049695e-05,
+      "loss": 1.1582,
+      "step": 6669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.176603692376051e-05,
+      "loss": 0.9111,
+      "step": 6670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.176351739594421e-05,
+      "loss": 0.9736,
+      "step": 6671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1760997582661945e-05,
+      "loss": 1.0547,
+      "step": 6672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.175847748397485e-05,
+      "loss": 0.9668,
+      "step": 6673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.175595709994409e-05,
+      "loss": 0.9219,
+      "step": 6674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.175343643063082e-05,
+      "loss": 0.8711,
+      "step": 6675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.175091547609622e-05,
+      "loss": 0.96,
+      "step": 6676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.174839423640146e-05,
+      "loss": 1.0176,
+      "step": 6677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1745872711607714e-05,
+      "loss": 0.9092,
+      "step": 6678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.174335090177619e-05,
+      "loss": 0.7949,
+      "step": 6679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.174082880696807e-05,
+      "loss": 1.0371,
+      "step": 6680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.173830642724456e-05,
+      "loss": 0.8652,
+      "step": 6681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.173578376266688e-05,
+      "loss": 0.9434,
+      "step": 6682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.173326081329624e-05,
+      "loss": 0.9941,
+      "step": 6683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1730737579193864e-05,
+      "loss": 0.9316,
+      "step": 6684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1728214060421e-05,
+      "loss": 0.9072,
+      "step": 6685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1725690257038856e-05,
+      "loss": 0.792,
+      "step": 6686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.17231661691087e-05,
+      "loss": 0.8369,
+      "step": 6687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1720641796691775e-05,
+      "loss": 1.0273,
+      "step": 6688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.171811713984934e-05,
+      "loss": 0.9785,
+      "step": 6689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.171559219864267e-05,
+      "loss": 1.082,
+      "step": 6690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1713066973133026e-05,
+      "loss": 0.6887,
+      "step": 6691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.17105414633817e-05,
+      "loss": 1.1221,
+      "step": 6692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.170801566944997e-05,
+      "loss": 0.9824,
+      "step": 6693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.170548959139913e-05,
+      "loss": 1.0293,
+      "step": 6694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.170296322929048e-05,
+      "loss": 0.8916,
+      "step": 6695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.170043658318533e-05,
+      "loss": 0.6719,
+      "step": 6696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1697909653144996e-05,
+      "loss": 0.9834,
+      "step": 6697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.16953824392308e-05,
+      "loss": 0.8174,
+      "step": 6698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.169285494150407e-05,
+      "loss": 0.7095,
+      "step": 6699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1690327160026124e-05,
+      "loss": 0.9482,
+      "step": 6700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.168779909485833e-05,
+      "loss": 1.0391,
+      "step": 6701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.168527074606202e-05,
+      "loss": 1.1514,
+      "step": 6702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1682742113698556e-05,
+      "loss": 1.0352,
+      "step": 6703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.16802131978293e-05,
+      "loss": 1.0225,
+      "step": 6704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.167768399851562e-05,
+      "loss": 1.0283,
+      "step": 6705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1675154515818886e-05,
+      "loss": 0.9385,
+      "step": 6706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.167262474980049e-05,
+      "loss": 0.9297,
+      "step": 6707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.167009470052182e-05,
+      "loss": 0.7705,
+      "step": 6708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.166756436804427e-05,
+      "loss": 1.0225,
+      "step": 6709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.166503375242925e-05,
+      "loss": 0.8584,
+      "step": 6710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1662502853738164e-05,
+      "loss": 1.1094,
+      "step": 6711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1659971672032436e-05,
+      "loss": 1.0947,
+      "step": 6712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.165744020737348e-05,
+      "loss": 0.8965,
+      "step": 6713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.165490845982274e-05,
+      "loss": 0.9463,
+      "step": 6714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.165237642944164e-05,
+      "loss": 0.8945,
+      "step": 6715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.164984411629164e-05,
+      "loss": 0.9941,
+      "step": 6716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.164731152043418e-05,
+      "loss": 1.0908,
+      "step": 6717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.164477864193073e-05,
+      "loss": 0.8213,
+      "step": 6718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.164224548084273e-05,
+      "loss": 1.1328,
+      "step": 6719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1639712037231686e-05,
+      "loss": 0.8662,
+      "step": 6720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.163717831115906e-05,
+      "loss": 0.897,
+      "step": 6721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.163464430268634e-05,
+      "loss": 0.9463,
+      "step": 6722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.163211001187502e-05,
+      "loss": 0.9902,
+      "step": 6723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.162957543878659e-05,
+      "loss": 0.9678,
+      "step": 6724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.162704058348258e-05,
+      "loss": 0.9717,
+      "step": 6725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.162450544602448e-05,
+      "loss": 1.0215,
+      "step": 6726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.162197002647382e-05,
+      "loss": 0.9512,
+      "step": 6727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.161943432489213e-05,
+      "loss": 0.8291,
+      "step": 6728
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.161689834134093e-05,
+      "loss": 0.7822,
+      "step": 6729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.161436207588178e-05,
+      "loss": 0.9961,
+      "step": 6730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1611825528576216e-05,
+      "loss": 0.9033,
+      "step": 6731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.16092886994858e-05,
+      "loss": 1.0039,
+      "step": 6732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1606751588672086e-05,
+      "loss": 1.0801,
+      "step": 6733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.160421419619664e-05,
+      "loss": 0.9775,
+      "step": 6734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1601676522121057e-05,
+      "loss": 0.9727,
+      "step": 6735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.159913856650689e-05,
+      "loss": 1.0098,
+      "step": 6736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.159660032941575e-05,
+      "loss": 1.0059,
+      "step": 6737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.159406181090922e-05,
+      "loss": 0.79,
+      "step": 6738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.15915230110489e-05,
+      "loss": 1.041,
+      "step": 6739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.158898392989642e-05,
+      "loss": 0.8818,
+      "step": 6740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1586444567513376e-05,
+      "loss": 1.0791,
+      "step": 6741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.158390492396139e-05,
+      "loss": 1.0215,
+      "step": 6742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.158136499930211e-05,
+      "loss": 0.8242,
+      "step": 6743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.157882479359716e-05,
+      "loss": 0.9648,
+      "step": 6744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.157628430690817e-05,
+      "loss": 0.9922,
+      "step": 6745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.157374353929682e-05,
+      "loss": 0.9014,
+      "step": 6746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1571202490824746e-05,
+      "loss": 0.8867,
+      "step": 6747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.156866116155362e-05,
+      "loss": 0.9775,
+      "step": 6748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1566119551545105e-05,
+      "loss": 0.8594,
+      "step": 6749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1563577660860887e-05,
+      "loss": 1.0684,
+      "step": 6750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.156103548956265e-05,
+      "loss": 1.0186,
+      "step": 6751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1558493037712076e-05,
+      "loss": 1.1484,
+      "step": 6752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1555950305370865e-05,
+      "loss": 0.9727,
+      "step": 6753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.155340729260073e-05,
+      "loss": 1.0,
+      "step": 6754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.155086399946337e-05,
+      "loss": 0.9736,
+      "step": 6755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.154832042602052e-05,
+      "loss": 0.8633,
+      "step": 6756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1545776572333894e-05,
+      "loss": 1.0186,
+      "step": 6757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1543232438465223e-05,
+      "loss": 1.0928,
+      "step": 6758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1540688024476245e-05,
+      "loss": 0.9746,
+      "step": 6759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.153814333042871e-05,
+      "loss": 0.9766,
+      "step": 6760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.153559835638438e-05,
+      "loss": 0.6968,
+      "step": 6761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.153305310240499e-05,
+      "loss": 1.0703,
+      "step": 6762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1530507568552324e-05,
+      "loss": 0.9844,
+      "step": 6763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.152796175488815e-05,
+      "loss": 0.9414,
+      "step": 6764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.152541566147425e-05,
+      "loss": 1.0752,
+      "step": 6765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.15228692883724e-05,
+      "loss": 0.9209,
+      "step": 6766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.15203226356444e-05,
+      "loss": 1.0107,
+      "step": 6767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.151777570335205e-05,
+      "loss": 0.8838,
+      "step": 6768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1515228491557156e-05,
+      "loss": 0.7944,
+      "step": 6769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1512681000321536e-05,
+      "loss": 0.9053,
+      "step": 6770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1510133229707e-05,
+      "loss": 1.0527,
+      "step": 6771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.150758517977538e-05,
+      "loss": 0.8721,
+      "step": 6772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1505036850588516e-05,
+      "loss": 0.8154,
+      "step": 6773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1502488242208244e-05,
+      "loss": 0.9258,
+      "step": 6774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.14999393546964e-05,
+      "loss": 0.9854,
+      "step": 6775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1497390188114864e-05,
+      "loss": 1.0801,
+      "step": 6776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1494840742525465e-05,
+      "loss": 0.9941,
+      "step": 6777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.14922910179901e-05,
+      "loss": 0.9287,
+      "step": 6778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1489741014570617e-05,
+      "loss": 1.0273,
+      "step": 6779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1487190732328916e-05,
+      "loss": 1.0508,
+      "step": 6780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.148464017132688e-05,
+      "loss": 1.043,
+      "step": 6781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.14820893316264e-05,
+      "loss": 0.8945,
+      "step": 6782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.147953821328938e-05,
+      "loss": 0.96,
+      "step": 6783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1476986816377726e-05,
+      "loss": 0.9688,
+      "step": 6784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.147443514095336e-05,
+      "loss": 0.8896,
+      "step": 6785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.147188318707818e-05,
+      "loss": 0.8018,
+      "step": 6786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.146933095481415e-05,
+      "loss": 0.9385,
+      "step": 6787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1466778444223185e-05,
+      "loss": 0.9775,
+      "step": 6788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.146422565536723e-05,
+      "loss": 1.0156,
+      "step": 6789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.146167258830823e-05,
+      "loss": 1.1113,
+      "step": 6790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.145911924310815e-05,
+      "loss": 0.9229,
+      "step": 6791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.145656561982894e-05,
+      "loss": 0.9336,
+      "step": 6792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.145401171853258e-05,
+      "loss": 1.082,
+      "step": 6793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1451457539281033e-05,
+      "loss": 1.0029,
+      "step": 6794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1448903082136295e-05,
+      "loss": 1.1465,
+      "step": 6795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.144634834716035e-05,
+      "loss": 0.9141,
+      "step": 6796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.144379333441519e-05,
+      "loss": 0.7412,
+      "step": 6797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.144123804396282e-05,
+      "loss": 1.1719,
+      "step": 6798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1438682475865256e-05,
+      "loss": 1.1182,
+      "step": 6799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.14361266301845e-05,
+      "loss": 0.8936,
+      "step": 6800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.143357050698259e-05,
+      "loss": 0.873,
+      "step": 6801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.143101410632155e-05,
+      "loss": 0.9697,
+      "step": 6802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1428457428263404e-05,
+      "loss": 0.9492,
+      "step": 6803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.142590047287022e-05,
+      "loss": 0.8853,
+      "step": 6804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.142334324020403e-05,
+      "loss": 1.0361,
+      "step": 6805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1420785730326886e-05,
+      "loss": 0.9844,
+      "step": 6806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.141822794330087e-05,
+      "loss": 0.916,
+      "step": 6807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.141566987918803e-05,
+      "loss": 0.8076,
+      "step": 6808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.141311153805046e-05,
+      "loss": 0.8975,
+      "step": 6809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.141055291995025e-05,
+      "loss": 1.0508,
+      "step": 6810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.140799402494946e-05,
+      "loss": 1.0146,
+      "step": 6811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.140543485311021e-05,
+      "loss": 0.9795,
+      "step": 6812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1402875404494596e-05,
+      "loss": 0.7305,
+      "step": 6813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.140031567916474e-05,
+      "loss": 0.9023,
+      "step": 6814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.139775567718274e-05,
+      "loss": 0.9287,
+      "step": 6815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.139519539861074e-05,
+      "loss": 1.1309,
+      "step": 6816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.139263484351084e-05,
+      "loss": 0.8955,
+      "step": 6817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1390074011945216e-05,
+      "loss": 1.0127,
+      "step": 6818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.138751290397599e-05,
+      "loss": 1.0225,
+      "step": 6819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.138495151966531e-05,
+      "loss": 0.8477,
+      "step": 6820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.138238985907534e-05,
+      "loss": 0.8809,
+      "step": 6821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.137982792226824e-05,
+      "loss": 0.915,
+      "step": 6822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.137726570930619e-05,
+      "loss": 1.04,
+      "step": 6823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.137470322025135e-05,
+      "loss": 0.8213,
+      "step": 6824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.137214045516593e-05,
+      "loss": 0.9238,
+      "step": 6825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.136957741411209e-05,
+      "loss": 0.8887,
+      "step": 6826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1367014097152055e-05,
+      "loss": 0.96,
+      "step": 6827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1364450504348005e-05,
+      "loss": 0.9795,
+      "step": 6828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.136188663576218e-05,
+      "loss": 1.0264,
+      "step": 6829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1359322491456766e-05,
+      "loss": 0.7075,
+      "step": 6830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1356758071494e-05,
+      "loss": 1.0264,
+      "step": 6831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1354193375936125e-05,
+      "loss": 0.8818,
+      "step": 6832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.135162840484536e-05,
+      "loss": 0.7949,
+      "step": 6833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.134906315828396e-05,
+      "loss": 0.9336,
+      "step": 6834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1346497636314176e-05,
+      "loss": 0.9941,
+      "step": 6835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1343931838998265e-05,
+      "loss": 0.8799,
+      "step": 6836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.134136576639849e-05,
+      "loss": 0.9502,
+      "step": 6837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1338799418577116e-05,
+      "loss": 1.0195,
+      "step": 6838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1336232795596425e-05,
+      "loss": 0.9756,
+      "step": 6839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1333665897518705e-05,
+      "loss": 0.9082,
+      "step": 6840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.133109872440625e-05,
+      "loss": 0.8066,
+      "step": 6841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1328531276321346e-05,
+      "loss": 0.6963,
+      "step": 6842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.132596355332631e-05,
+      "loss": 0.7979,
+      "step": 6843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.132339555548344e-05,
+      "loss": 0.9766,
+      "step": 6844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1320827282855066e-05,
+      "loss": 0.6836,
+      "step": 6845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.131825873550351e-05,
+      "loss": 1.0283,
+      "step": 6846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.131568991349109e-05,
+      "loss": 0.9619,
+      "step": 6847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.131312081688016e-05,
+      "loss": 0.8594,
+      "step": 6848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.131055144573306e-05,
+      "loss": 0.9639,
+      "step": 6849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.130798180011214e-05,
+      "loss": 0.9727,
+      "step": 6850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.130541188007975e-05,
+      "loss": 0.7969,
+      "step": 6851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.130284168569827e-05,
+      "loss": 0.8008,
+      "step": 6852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.130027121703006e-05,
+      "loss": 1.0449,
+      "step": 6853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.129770047413749e-05,
+      "loss": 1.0566,
+      "step": 6854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.129512945708297e-05,
+      "loss": 0.9854,
+      "step": 6855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.129255816592887e-05,
+      "loss": 0.835,
+      "step": 6856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1289986600737595e-05,
+      "loss": 0.8799,
+      "step": 6857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1287414761571544e-05,
+      "loss": 1.002,
+      "step": 6858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.128484264849314e-05,
+      "loss": 0.8936,
+      "step": 6859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1282270261564785e-05,
+      "loss": 1.043,
+      "step": 6860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1279697600848916e-05,
+      "loss": 1.0908,
+      "step": 6861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.127712466640795e-05,
+      "loss": 0.9639,
+      "step": 6862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.127455145830434e-05,
+      "loss": 1.0186,
+      "step": 6863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1271977976600526e-05,
+      "loss": 0.9678,
+      "step": 6864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.126940422135895e-05,
+      "loss": 0.8662,
+      "step": 6865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1266830192642084e-05,
+      "loss": 0.9072,
+      "step": 6866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.126425589051237e-05,
+      "loss": 1.1318,
+      "step": 6867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.126168131503231e-05,
+      "loss": 0.9087,
+      "step": 6868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1259106466264346e-05,
+      "loss": 1.1191,
+      "step": 6869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.125653134427099e-05,
+      "loss": 0.8525,
+      "step": 6870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.125395594911472e-05,
+      "loss": 0.8857,
+      "step": 6871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.125138028085804e-05,
+      "loss": 0.8877,
+      "step": 6872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.124880433956344e-05,
+      "loss": 0.9404,
+      "step": 6873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.124622812529345e-05,
+      "loss": 0.8955,
+      "step": 6874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1243651638110576e-05,
+      "loss": 0.8623,
+      "step": 6875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.124107487807734e-05,
+      "loss": 0.9268,
+      "step": 6876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.123849784525628e-05,
+      "loss": 1.0518,
+      "step": 6877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.123592053970992e-05,
+      "loss": 1.0957,
+      "step": 6878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.123334296150082e-05,
+      "loss": 0.9434,
+      "step": 6879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1230765110691516e-05,
+      "loss": 0.9414,
+      "step": 6880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.122818698734458e-05,
+      "loss": 1.0322,
+      "step": 6881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1225608591522555e-05,
+      "loss": 1.0049,
+      "step": 6882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.122302992328803e-05,
+      "loss": 0.9385,
+      "step": 6883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.122045098270358e-05,
+      "loss": 0.8398,
+      "step": 6884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1217871769831775e-05,
+      "loss": 0.9014,
+      "step": 6885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.121529228473521e-05,
+      "loss": 0.8779,
+      "step": 6886
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.121271252747649e-05,
+      "loss": 1.0293,
+      "step": 6887
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.121013249811821e-05,
+      "loss": 0.7148,
+      "step": 6888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1207552196722986e-05,
+      "loss": 0.8691,
+      "step": 6889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1204971623353424e-05,
+      "loss": 0.9932,
+      "step": 6890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.120239077807216e-05,
+      "loss": 0.9307,
+      "step": 6891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.119980966094181e-05,
+      "loss": 0.8848,
+      "step": 6892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.119722827202503e-05,
+      "loss": 1.0264,
+      "step": 6893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.119464661138444e-05,
+      "loss": 1.0723,
+      "step": 6894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.119206467908269e-05,
+      "loss": 1.1143,
+      "step": 6895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1189482475182455e-05,
+      "loss": 0.9414,
+      "step": 6896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.118689999974639e-05,
+      "loss": 0.9512,
+      "step": 6897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.118431725283715e-05,
+      "loss": 0.9404,
+      "step": 6898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1181734234517436e-05,
+      "loss": 0.959,
+      "step": 6899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.117915094484991e-05,
+      "loss": 1.0186,
+      "step": 6900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1176567383897265e-05,
+      "loss": 0.7649,
+      "step": 6901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1173983551722205e-05,
+      "loss": 0.7646,
+      "step": 6902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.117139944838742e-05,
+      "loss": 0.9375,
+      "step": 6903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.116881507395563e-05,
+      "loss": 0.9736,
+      "step": 6904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.116623042848954e-05,
+      "loss": 0.9375,
+      "step": 6905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.116364551205188e-05,
+      "loss": 0.958,
+      "step": 6906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.116106032470537e-05,
+      "loss": 0.958,
+      "step": 6907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.115847486651276e-05,
+      "loss": 0.9297,
+      "step": 6908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1155889137536766e-05,
+      "loss": 1.0645,
+      "step": 6909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.115330313784016e-05,
+      "loss": 0.8872,
+      "step": 6910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.115071686748568e-05,
+      "loss": 0.833,
+      "step": 6911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.11481303265361e-05,
+      "loss": 0.9365,
+      "step": 6912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.114554351505418e-05,
+      "loss": 0.8799,
+      "step": 6913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.11429564331027e-05,
+      "loss": 0.9336,
+      "step": 6914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.114036908074444e-05,
+      "loss": 0.9814,
+      "step": 6915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.113778145804218e-05,
+      "loss": 0.7852,
+      "step": 6916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.113519356505872e-05,
+      "loss": 1.0312,
+      "step": 6917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.113260540185686e-05,
+      "loss": 1.0312,
+      "step": 6918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.11300169684994e-05,
+      "loss": 1.1152,
+      "step": 6919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.112742826504917e-05,
+      "loss": 0.7998,
+      "step": 6920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.112483929156898e-05,
+      "loss": 0.8428,
+      "step": 6921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.112225004812166e-05,
+      "loss": 1.0762,
+      "step": 6922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.111966053477003e-05,
+      "loss": 1.0078,
+      "step": 6923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.111707075157695e-05,
+      "loss": 0.9756,
+      "step": 6924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.111448069860525e-05,
+      "loss": 1.0938,
+      "step": 6925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.111189037591779e-05,
+      "loss": 0.8857,
+      "step": 6926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1109299783577434e-05,
+      "loss": 0.9932,
+      "step": 6927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.110670892164704e-05,
+      "loss": 1.083,
+      "step": 6928
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1104117790189494e-05,
+      "loss": 0.9824,
+      "step": 6929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.110152638926766e-05,
+      "loss": 0.7852,
+      "step": 6930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.109893471894442e-05,
+      "loss": 0.9365,
+      "step": 6931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.109634277928268e-05,
+      "loss": 0.959,
+      "step": 6932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.109375057034534e-05,
+      "loss": 0.769,
+      "step": 6933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.109115809219529e-05,
+      "loss": 0.9863,
+      "step": 6934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.108856534489546e-05,
+      "loss": 0.917,
+      "step": 6935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1085972328508756e-05,
+      "loss": 0.9814,
+      "step": 6936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1083379043098104e-05,
+      "loss": 0.9834,
+      "step": 6937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.108078548872644e-05,
+      "loss": 0.9873,
+      "step": 6938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.107819166545669e-05,
+      "loss": 1.042,
+      "step": 6939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.107559757335182e-05,
+      "loss": 0.8223,
+      "step": 6940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1073003212474766e-05,
+      "loss": 1.1895,
+      "step": 6941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.107040858288849e-05,
+      "loss": 0.9463,
+      "step": 6942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1067813684655955e-05,
+      "loss": 0.8799,
+      "step": 6943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.106521851784013e-05,
+      "loss": 1.1367,
+      "step": 6944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.106262308250399e-05,
+      "loss": 1.0059,
+      "step": 6945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.106002737871053e-05,
+      "loss": 1.0352,
+      "step": 6946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.105743140652272e-05,
+      "loss": 1.0645,
+      "step": 6947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.105483516600358e-05,
+      "loss": 0.8447,
+      "step": 6948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.105223865721609e-05,
+      "loss": 0.9736,
+      "step": 6949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.104964188022328e-05,
+      "loss": 0.9209,
+      "step": 6950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.104704483508815e-05,
+      "loss": 0.9717,
+      "step": 6951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.104444752187374e-05,
+      "loss": 0.9443,
+      "step": 6952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.104184994064306e-05,
+      "loss": 1.0742,
+      "step": 6953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.103925209145916e-05,
+      "loss": 0.9502,
+      "step": 6954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1036653974385075e-05,
+      "loss": 0.8164,
+      "step": 6955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.103405558948386e-05,
+      "loss": 0.8896,
+      "step": 6956
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1031456936818564e-05,
+      "loss": 0.958,
+      "step": 6957
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.102885801645225e-05,
+      "loss": 0.8027,
+      "step": 6958
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1026258828447984e-05,
+      "loss": 1.0762,
+      "step": 6959
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1023659372868844e-05,
+      "loss": 0.9121,
+      "step": 6960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1021059649777915e-05,
+      "loss": 0.7695,
+      "step": 6961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.101845965923827e-05,
+      "loss": 0.8369,
+      "step": 6962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1015859401313026e-05,
+      "loss": 1.0703,
+      "step": 6963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.101325887606526e-05,
+      "loss": 0.9668,
+      "step": 6964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.1010658083558096e-05,
+      "loss": 0.8662,
+      "step": 6965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.100805702385464e-05,
+      "loss": 0.8848,
+      "step": 6966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.1005455697018015e-05,
+      "loss": 0.8789,
+      "step": 6967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.100285410311134e-05,
+      "loss": 1.1309,
+      "step": 6968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.1000252242197766e-05,
+      "loss": 0.9922,
+      "step": 6969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0997650114340415e-05,
+      "loss": 0.9619,
+      "step": 6970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.099504771960243e-05,
+      "loss": 0.9648,
+      "step": 6971
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.099244505804698e-05,
+      "loss": 0.7827,
+      "step": 6972
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.098984212973722e-05,
+      "loss": 1.0752,
+      "step": 6973
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.09872389347363e-05,
+      "loss": 0.8677,
+      "step": 6974
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.098463547310742e-05,
+      "loss": 0.9775,
+      "step": 6975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.098203174491373e-05,
+      "loss": 0.9473,
+      "step": 6976
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.097942775021842e-05,
+      "loss": 0.9697,
+      "step": 6977
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.09768234890847e-05,
+      "loss": 0.9521,
+      "step": 6978
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.097421896157575e-05,
+      "loss": 1.0146,
+      "step": 6979
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0971614167754785e-05,
+      "loss": 0.917,
+      "step": 6980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0969009107685005e-05,
+      "loss": 0.8643,
+      "step": 6981
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.096640378142963e-05,
+      "loss": 0.9697,
+      "step": 6982
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.096379818905189e-05,
+      "loss": 1.0664,
+      "step": 6983
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.096119233061502e-05,
+      "loss": 1.0273,
+      "step": 6984
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.095858620618223e-05,
+      "loss": 0.9854,
+      "step": 6985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.09559798158168e-05,
+      "loss": 1.1182,
+      "step": 6986
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.095337315958195e-05,
+      "loss": 0.9395,
+      "step": 6987
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0950766237540946e-05,
+      "loss": 1.0488,
+      "step": 6988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.094815904975705e-05,
+      "loss": 1.0273,
+      "step": 6989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.094555159629353e-05,
+      "loss": 0.8574,
+      "step": 6990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.094294387721367e-05,
+      "loss": 0.9111,
+      "step": 6991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.094033589258074e-05,
+      "loss": 0.9404,
+      "step": 6992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.093772764245803e-05,
+      "loss": 1.0391,
+      "step": 6993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.093511912690884e-05,
+      "loss": 0.9229,
+      "step": 6994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.093251034599646e-05,
+      "loss": 0.9932,
+      "step": 6995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.092990129978422e-05,
+      "loss": 1.0,
+      "step": 6996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.092729198833541e-05,
+      "loss": 0.8545,
+      "step": 6997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.092468241171336e-05,
+      "loss": 0.8384,
+      "step": 6998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0922072569981396e-05,
+      "loss": 0.9883,
+      "step": 6999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.091946246320285e-05,
+      "loss": 1.0645,
+      "step": 7000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0916852091441067e-05,
+      "loss": 0.9307,
+      "step": 7001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.091424145475939e-05,
+      "loss": 0.833,
+      "step": 7002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.091163055322117e-05,
+      "loss": 0.8174,
+      "step": 7003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0909019386889756e-05,
+      "loss": 1.0312,
+      "step": 7004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0906407955828534e-05,
+      "loss": 0.9834,
+      "step": 7005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.090379626010086e-05,
+      "loss": 1.0137,
+      "step": 7006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.090118429977012e-05,
+      "loss": 0.8779,
+      "step": 7007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.08985720748997e-05,
+      "loss": 1.0107,
+      "step": 7008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.089595958555299e-05,
+      "loss": 0.8667,
+      "step": 7009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.089334683179337e-05,
+      "loss": 0.7646,
+      "step": 7010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.089073381368427e-05,
+      "loss": 0.9336,
+      "step": 7011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.088812053128909e-05,
+      "loss": 1.0576,
+      "step": 7012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.088550698467124e-05,
+      "loss": 1.0654,
+      "step": 7013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.088289317389415e-05,
+      "loss": 0.9766,
+      "step": 7014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.088027909902125e-05,
+      "loss": 0.8809,
+      "step": 7015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0877664760115975e-05,
+      "loss": 0.9326,
+      "step": 7016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.087505015724176e-05,
+      "loss": 0.9707,
+      "step": 7017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0872435290462074e-05,
+      "loss": 0.8955,
+      "step": 7018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0869820159840344e-05,
+      "loss": 0.9717,
+      "step": 7019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0867204765440045e-05,
+      "loss": 0.9199,
+      "step": 7020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.086458910732465e-05,
+      "loss": 1.001,
+      "step": 7021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0861973185557634e-05,
+      "loss": 0.9551,
+      "step": 7022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.085935700020247e-05,
+      "loss": 0.9443,
+      "step": 7023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.085674055132264e-05,
+      "loss": 0.9199,
+      "step": 7024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.085412383898165e-05,
+      "loss": 1.1426,
+      "step": 7025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0851506863243e-05,
+      "loss": 0.9766,
+      "step": 7026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.084888962417019e-05,
+      "loss": 0.7959,
+      "step": 7027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.084627212182673e-05,
+      "loss": 0.9756,
+      "step": 7028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0843654356276144e-05,
+      "loss": 0.9639,
+      "step": 7029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.084103632758196e-05,
+      "loss": 0.7461,
+      "step": 7030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.083841803580771e-05,
+      "loss": 0.9541,
+      "step": 7031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.083579948101692e-05,
+      "loss": 1.0156,
+      "step": 7032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.083318066327315e-05,
+      "loss": 1.0449,
+      "step": 7033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.083056158263994e-05,
+      "loss": 0.8662,
+      "step": 7034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.082794223918086e-05,
+      "loss": 0.9756,
+      "step": 7035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0825322632959456e-05,
+      "loss": 1.0195,
+      "step": 7036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0822702764039315e-05,
+      "loss": 1.0352,
+      "step": 7037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0820082632484005e-05,
+      "loss": 1.082,
+      "step": 7038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.081746223835711e-05,
+      "loss": 0.9873,
+      "step": 7039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.081484158172221e-05,
+      "loss": 1.0469,
+      "step": 7040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0812220662642925e-05,
+      "loss": 0.8906,
+      "step": 7041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.080959948118283e-05,
+      "loss": 0.8115,
+      "step": 7042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.080697803740555e-05,
+      "loss": 0.8721,
+      "step": 7043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0804356331374704e-05,
+      "loss": 0.8232,
+      "step": 7044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.080173436315389e-05,
+      "loss": 1.1221,
+      "step": 7045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0799112132806764e-05,
+      "loss": 1.083,
+      "step": 7046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0796489640396934e-05,
+      "loss": 0.9209,
+      "step": 7047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.079386688598805e-05,
+      "loss": 0.8701,
+      "step": 7048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.079124386964377e-05,
+      "loss": 0.9492,
+      "step": 7049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0788620591427725e-05,
+      "loss": 0.8154,
+      "step": 7050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.07859970514036e-05,
+      "loss": 0.8486,
+      "step": 7051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0783373249635036e-05,
+      "loss": 0.832,
+      "step": 7052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0780749186185725e-05,
+      "loss": 0.9023,
+      "step": 7053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.077812486111934e-05,
+      "loss": 0.8535,
+      "step": 7054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.077550027449955e-05,
+      "loss": 0.96,
+      "step": 7055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.077287542639006e-05,
+      "loss": 0.8584,
+      "step": 7056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.077025031685457e-05,
+      "loss": 0.9707,
+      "step": 7057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.076762494595677e-05,
+      "loss": 0.9307,
+      "step": 7058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0764999313760396e-05,
+      "loss": 0.8711,
+      "step": 7059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.076237342032913e-05,
+      "loss": 0.9863,
+      "step": 7060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.075974726572672e-05,
+      "loss": 1.0693,
+      "step": 7061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.075712085001689e-05,
+      "loss": 1.0371,
+      "step": 7062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.075449417326336e-05,
+      "loss": 1.043,
+      "step": 7063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.07518672355299e-05,
+      "loss": 0.8276,
+      "step": 7064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.074924003688024e-05,
+      "loss": 1.1318,
+      "step": 7065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0746612577378126e-05,
+      "loss": 0.9961,
+      "step": 7066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0743984857087335e-05,
+      "loss": 0.8613,
+      "step": 7067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0741356876071625e-05,
+      "loss": 0.8721,
+      "step": 7068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.073872863439478e-05,
+      "loss": 1.1035,
+      "step": 7069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.073610013212057e-05,
+      "loss": 0.918,
+      "step": 7070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.073347136931278e-05,
+      "loss": 1.0098,
+      "step": 7071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.073084234603521e-05,
+      "loss": 0.8574,
+      "step": 7072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.072821306235166e-05,
+      "loss": 1.0527,
+      "step": 7073
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.072558351832592e-05,
+      "loss": 0.8701,
+      "step": 7074
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0722953714021814e-05,
+      "loss": 1.0107,
+      "step": 7075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.072032364950315e-05,
+      "loss": 1.0078,
+      "step": 7076
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.071769332483377e-05,
+      "loss": 1.0117,
+      "step": 7077
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.071506274007748e-05,
+      "loss": 0.8877,
+      "step": 7078
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0712431895298144e-05,
+      "loss": 0.8105,
+      "step": 7079
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.070980079055958e-05,
+      "loss": 0.8462,
+      "step": 7080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.070716942592565e-05,
+      "loss": 0.9443,
+      "step": 7081
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.070453780146021e-05,
+      "loss": 1.0068,
+      "step": 7082
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0701905917227115e-05,
+      "loss": 0.9434,
+      "step": 7083
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.069927377329023e-05,
+      "loss": 0.9424,
+      "step": 7084
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.069664136971344e-05,
+      "loss": 0.895,
+      "step": 7085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.069400870656063e-05,
+      "loss": 0.791,
+      "step": 7086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0691375783895674e-05,
+      "loss": 1.0068,
+      "step": 7087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.068874260178246e-05,
+      "loss": 1.1133,
+      "step": 7088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0686109160284914e-05,
+      "loss": 1.1602,
+      "step": 7089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.068347545946691e-05,
+      "loss": 0.8779,
+      "step": 7090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.068084149939239e-05,
+      "loss": 1.0352,
+      "step": 7091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.067820728012525e-05,
+      "loss": 1.0166,
+      "step": 7092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0675572801729415e-05,
+      "loss": 0.8994,
+      "step": 7093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.067293806426884e-05,
+      "loss": 0.7529,
+      "step": 7094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.067030306780744e-05,
+      "loss": 1.0068,
+      "step": 7095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.066766781240916e-05,
+      "loss": 0.915,
+      "step": 7096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.066503229813796e-05,
+      "loss": 0.9951,
+      "step": 7097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.066239652505779e-05,
+      "loss": 0.9844,
+      "step": 7098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0659760493232616e-05,
+      "loss": 1.002,
+      "step": 7099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.065712420272639e-05,
+      "loss": 1.0078,
+      "step": 7100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0654487653603115e-05,
+      "loss": 0.9355,
+      "step": 7101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0651850845926756e-05,
+      "loss": 0.8936,
+      "step": 7102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.06492137797613e-05,
+      "loss": 0.9219,
+      "step": 7103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0646576455170745e-05,
+      "loss": 0.8887,
+      "step": 7104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.064393887221909e-05,
+      "loss": 1.041,
+      "step": 7105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.064130103097034e-05,
+      "loss": 0.9277,
+      "step": 7106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.063866293148851e-05,
+      "loss": 0.8457,
+      "step": 7107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.063602457383762e-05,
+      "loss": 0.9268,
+      "step": 7108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0633385958081694e-05,
+      "loss": 1.0684,
+      "step": 7109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.063074708428476e-05,
+      "loss": 0.9307,
+      "step": 7110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0628107952510855e-05,
+      "loss": 1.1719,
+      "step": 7111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.062546856282403e-05,
+      "loss": 1.0391,
+      "step": 7112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.062282891528833e-05,
+      "loss": 0.8799,
+      "step": 7113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0620189009967815e-05,
+      "loss": 0.9199,
+      "step": 7114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.061754884692654e-05,
+      "loss": 0.9385,
+      "step": 7115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0614908426228584e-05,
+      "loss": 0.9053,
+      "step": 7116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0612267747938015e-05,
+      "loss": 1.0068,
+      "step": 7117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0609626812118924e-05,
+      "loss": 1.0957,
+      "step": 7118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0606985618835386e-05,
+      "loss": 1.1143,
+      "step": 7119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0604344168151505e-05,
+      "loss": 0.8262,
+      "step": 7120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.060170246013138e-05,
+      "loss": 0.959,
+      "step": 7121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.05990604948391e-05,
+      "loss": 0.9629,
+      "step": 7122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0596418272338806e-05,
+      "loss": 0.9541,
+      "step": 7123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0593775792694594e-05,
+      "loss": 0.957,
+      "step": 7124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.059113305597061e-05,
+      "loss": 0.9854,
+      "step": 7125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.058849006223097e-05,
+      "loss": 1.1426,
+      "step": 7126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0585846811539814e-05,
+      "loss": 0.7529,
+      "step": 7127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.058320330396129e-05,
+      "loss": 0.9814,
+      "step": 7128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.058055953955954e-05,
+      "loss": 0.9277,
+      "step": 7129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.057791551839873e-05,
+      "loss": 1.0361,
+      "step": 7130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0575271240543024e-05,
+      "loss": 1.0391,
+      "step": 7131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.057262670605658e-05,
+      "loss": 0.7759,
+      "step": 7132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0569981915003575e-05,
+      "loss": 0.8584,
+      "step": 7133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0567336867448196e-05,
+      "loss": 0.9199,
+      "step": 7134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0564691563454635e-05,
+      "loss": 1.002,
+      "step": 7135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.056204600308708e-05,
+      "loss": 0.9375,
+      "step": 7136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.055940018640972e-05,
+      "loss": 0.9404,
+      "step": 7137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.055675411348678e-05,
+      "loss": 1.2246,
+      "step": 7138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0554107784382466e-05,
+      "loss": 1.1172,
+      "step": 7139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0551461199160994e-05,
+      "loss": 0.9619,
+      "step": 7140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.054881435788658e-05,
+      "loss": 0.9912,
+      "step": 7141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.054616726062347e-05,
+      "loss": 1.0635,
+      "step": 7142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.05435199074359e-05,
+      "loss": 0.8799,
+      "step": 7143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0540872298388106e-05,
+      "loss": 1.0371,
+      "step": 7144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.053822443354434e-05,
+      "loss": 0.9854,
+      "step": 7145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0535576312968856e-05,
+      "loss": 0.9736,
+      "step": 7146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.053292793672592e-05,
+      "loss": 0.8154,
+      "step": 7147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.05302793048798e-05,
+      "loss": 0.8984,
+      "step": 7148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.052763041749478e-05,
+      "loss": 0.8408,
+      "step": 7149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.052498127463511e-05,
+      "loss": 0.9951,
+      "step": 7150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.052233187636511e-05,
+      "loss": 1.0068,
+      "step": 7151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.051968222274906e-05,
+      "loss": 0.959,
+      "step": 7152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0517032313851254e-05,
+      "loss": 0.9082,
+      "step": 7153
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.051438214973601e-05,
+      "loss": 1.1523,
+      "step": 7154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0511731730467633e-05,
+      "loss": 0.9775,
+      "step": 7155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0509081056110438e-05,
+      "loss": 1.0303,
+      "step": 7156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0506430126728755e-05,
+      "loss": 0.9639,
+      "step": 7157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0503778942386906e-05,
+      "loss": 1.0361,
+      "step": 7158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.050112750314924e-05,
+      "loss": 0.7725,
+      "step": 7159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0498475809080082e-05,
+      "loss": 1.04,
+      "step": 7160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0495823860243798e-05,
+      "loss": 0.7275,
+      "step": 7161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0493171656704734e-05,
+      "loss": 0.9375,
+      "step": 7162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0490519198527254e-05,
+      "loss": 1.0127,
+      "step": 7163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0487866485775727e-05,
+      "loss": 1.0566,
+      "step": 7164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.048521351851452e-05,
+      "loss": 1.0801,
+      "step": 7165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0482560296808018e-05,
+      "loss": 0.999,
+      "step": 7166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0479906820720603e-05,
+      "loss": 0.8584,
+      "step": 7167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.047725309031668e-05,
+      "loss": 1.0566,
+      "step": 7168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0474599105660626e-05,
+      "loss": 0.9746,
+      "step": 7169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.047194486681686e-05,
+      "loss": 0.875,
+      "step": 7170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0469290373849784e-05,
+      "loss": 1.0518,
+      "step": 7171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0466635626823832e-05,
+      "loss": 1.0264,
+      "step": 7172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0463980625803402e-05,
+      "loss": 0.8574,
+      "step": 7173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0461325370852943e-05,
+      "loss": 0.9893,
+      "step": 7174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.045866986203688e-05,
+      "loss": 1.0605,
+      "step": 7175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0456014099419658e-05,
+      "loss": 0.6675,
+      "step": 7176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0453358083065727e-05,
+      "loss": 1.0879,
+      "step": 7177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0450701813039532e-05,
+      "loss": 1.0322,
+      "step": 7178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0448045289405537e-05,
+      "loss": 1.1133,
+      "step": 7179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.044538851222822e-05,
+      "loss": 1.0381,
+      "step": 7180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.044273148157203e-05,
+      "loss": 1.0332,
+      "step": 7181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0440074197501465e-05,
+      "loss": 0.9062,
+      "step": 7182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0437416660081003e-05,
+      "loss": 0.9717,
+      "step": 7183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0434758869375132e-05,
+      "loss": 1.0508,
+      "step": 7184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0432100825448355e-05,
+      "loss": 0.9834,
+      "step": 7185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0429442528365168e-05,
+      "loss": 0.709,
+      "step": 7186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.042678397819007e-05,
+      "loss": 0.9443,
+      "step": 7187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0424125174987602e-05,
+      "loss": 1.1045,
+      "step": 7188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.042146611882227e-05,
+      "loss": 0.8789,
+      "step": 7189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.041880680975861e-05,
+      "loss": 0.9316,
+      "step": 7190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.041614724786114e-05,
+      "loss": 1.0176,
+      "step": 7191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0413487433194415e-05,
+      "loss": 0.772,
+      "step": 7192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.041082736582297e-05,
+      "loss": 0.9971,
+      "step": 7193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0408167045811368e-05,
+      "loss": 0.9678,
+      "step": 7194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0405506473224157e-05,
+      "loss": 1.0732,
+      "step": 7195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0402845648125904e-05,
+      "loss": 1.002,
+      "step": 7196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0400184570581187e-05,
+      "loss": 1.0117,
+      "step": 7197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0397523240654574e-05,
+      "loss": 0.917,
+      "step": 7198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0394861658410642e-05,
+      "loss": 1.0869,
+      "step": 7199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0392199823913997e-05,
+      "loss": 0.8574,
+      "step": 7200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.038953773722922e-05,
+      "loss": 0.8633,
+      "step": 7201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.038687539842092e-05,
+      "loss": 0.9844,
+      "step": 7202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.03842128075537e-05,
+      "loss": 0.8828,
+      "step": 7203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.038154996469217e-05,
+      "loss": 0.9619,
+      "step": 7204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.037888686990096e-05,
+      "loss": 0.9658,
+      "step": 7205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0376223523244685e-05,
+      "loss": 1.0752,
+      "step": 7206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0373559924787978e-05,
+      "loss": 1.0381,
+      "step": 7207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0370896074595484e-05,
+      "loss": 0.9004,
+      "step": 7208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.036823197273184e-05,
+      "loss": 0.9775,
+      "step": 7209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0365567619261694e-05,
+      "loss": 0.9395,
+      "step": 7210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.036290301424971e-05,
+      "loss": 0.9648,
+      "step": 7211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0360238157760552e-05,
+      "loss": 0.9482,
+      "step": 7212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.035757304985887e-05,
+      "loss": 0.8408,
+      "step": 7213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0354907690609363e-05,
+      "loss": 1.1289,
+      "step": 7214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0352242080076684e-05,
+      "loss": 1.0381,
+      "step": 7215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.034957621832555e-05,
+      "loss": 0.9336,
+      "step": 7216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.034691010542063e-05,
+      "loss": 1.1377,
+      "step": 7217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0344243741426632e-05,
+      "loss": 0.9941,
+      "step": 7218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.034157712640826e-05,
+      "loss": 0.876,
+      "step": 7219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0338910260430223e-05,
+      "loss": 0.8691,
+      "step": 7220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0336243143557245e-05,
+      "loss": 0.9258,
+      "step": 7221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0333575775854033e-05,
+      "loss": 0.9893,
+      "step": 7222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.033090815738534e-05,
+      "loss": 1.0234,
+      "step": 7223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0328240288215877e-05,
+      "loss": 1.0801,
+      "step": 7224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0325572168410407e-05,
+      "loss": 1.0244,
+      "step": 7225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0322903798033655e-05,
+      "loss": 1.0195,
+      "step": 7226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0320235177150396e-05,
+      "loss": 0.9219,
+      "step": 7227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0317566305825375e-05,
+      "loss": 0.8052,
+      "step": 7228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0314897184123365e-05,
+      "loss": 0.9238,
+      "step": 7229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0312227812109137e-05,
+      "loss": 0.8262,
+      "step": 7230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0309558189847467e-05,
+      "loss": 0.7617,
+      "step": 7231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0306888317403134e-05,
+      "loss": 1.042,
+      "step": 7232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.030421819484094e-05,
+      "loss": 0.8779,
+      "step": 7233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0301547822225666e-05,
+      "loss": 0.8711,
+      "step": 7234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0298877199622124e-05,
+      "loss": 1.1006,
+      "step": 7235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0296206327095125e-05,
+      "loss": 0.8945,
+      "step": 7236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0293535204709476e-05,
+      "loss": 0.9619,
+      "step": 7237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.029086383253e-05,
+      "loss": 0.9521,
+      "step": 7238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0288192210621523e-05,
+      "loss": 0.9922,
+      "step": 7239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0285520339048876e-05,
+      "loss": 1.1523,
+      "step": 7240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.02828482178769e-05,
+      "loss": 0.9072,
+      "step": 7241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0280175847170438e-05,
+      "loss": 0.8477,
+      "step": 7242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.027750322699434e-05,
+      "loss": 0.9004,
+      "step": 7243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0274830357413466e-05,
+      "loss": 1.083,
+      "step": 7244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0272157238492677e-05,
+      "loss": 1.0596,
+      "step": 7245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.026948387029684e-05,
+      "loss": 0.8086,
+      "step": 7246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0266810252890835e-05,
+      "loss": 0.8965,
+      "step": 7247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.026413638633953e-05,
+      "loss": 1.0752,
+      "step": 7248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0261462270707828e-05,
+      "loss": 0.9277,
+      "step": 7249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0258787906060614e-05,
+      "loss": 0.9336,
+      "step": 7250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0256113292462786e-05,
+      "loss": 1.1367,
+      "step": 7251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0253438429979245e-05,
+      "loss": 1.1035,
+      "step": 7252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0250763318674917e-05,
+      "loss": 1.085,
+      "step": 7253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0248087958614705e-05,
+      "loss": 0.7549,
+      "step": 7254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.024541234986354e-05,
+      "loss": 0.9961,
+      "step": 7255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.024273649248634e-05,
+      "loss": 1.0703,
+      "step": 7256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.024006038654806e-05,
+      "loss": 1.1162,
+      "step": 7257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.023738403211362e-05,
+      "loss": 0.8311,
+      "step": 7258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.023470742924798e-05,
+      "loss": 1.0527,
+      "step": 7259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0232030578016086e-05,
+      "loss": 0.8516,
+      "step": 7260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.022935347848291e-05,
+      "loss": 0.9141,
+      "step": 7261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.02266761307134e-05,
+      "loss": 0.8613,
+      "step": 7262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0223998534772535e-05,
+      "loss": 1.043,
+      "step": 7263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0221320690725302e-05,
+      "loss": 0.8564,
+      "step": 7264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0218642598636667e-05,
+      "loss": 0.9766,
+      "step": 7265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0215964258571632e-05,
+      "loss": 0.8467,
+      "step": 7266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.021328567059519e-05,
+      "loss": 1.0986,
+      "step": 7267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0210606834772337e-05,
+      "loss": 0.9893,
+      "step": 7268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0207927751168092e-05,
+      "loss": 0.8672,
+      "step": 7269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0205248419847447e-05,
+      "loss": 0.8525,
+      "step": 7270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0202568840875445e-05,
+      "loss": 1.0742,
+      "step": 7271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.01998890143171e-05,
+      "loss": 0.9287,
+      "step": 7272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.019720894023745e-05,
+      "loss": 0.8989,
+      "step": 7273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0194528618701522e-05,
+      "loss": 0.8701,
+      "step": 7274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.019184804977437e-05,
+      "loss": 1.001,
+      "step": 7275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0189167233521036e-05,
+      "loss": 0.8701,
+      "step": 7276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.018648617000658e-05,
+      "loss": 1.1172,
+      "step": 7277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0183804859296058e-05,
+      "loss": 0.9062,
+      "step": 7278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0181123301454548e-05,
+      "loss": 1.125,
+      "step": 7279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0178441496547113e-05,
+      "loss": 1.0645,
+      "step": 7280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.017575944463884e-05,
+      "loss": 0.9395,
+      "step": 7281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.017307714579481e-05,
+      "loss": 1.0449,
+      "step": 7282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.017039460008011e-05,
+      "loss": 0.9443,
+      "step": 7283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.016771180755985e-05,
+      "loss": 1.0693,
+      "step": 7284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.016502876829912e-05,
+      "loss": 0.8916,
+      "step": 7285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0162345482363045e-05,
+      "loss": 0.8604,
+      "step": 7286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0159661949816725e-05,
+      "loss": 1.0029,
+      "step": 7287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0156978170725287e-05,
+      "loss": 0.96,
+      "step": 7288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0154294145153865e-05,
+      "loss": 1.0488,
+      "step": 7289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0151609873167582e-05,
+      "loss": 0.9004,
+      "step": 7290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0148925354831585e-05,
+      "loss": 1.0195,
+      "step": 7291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0146240590211015e-05,
+      "loss": 1.1523,
+      "step": 7292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0143555579371028e-05,
+      "loss": 1.0166,
+      "step": 7293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.014087032237677e-05,
+      "loss": 0.9648,
+      "step": 7294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0138184819293426e-05,
+      "loss": 1.0557,
+      "step": 7295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0135499070186136e-05,
+      "loss": 1.0176,
+      "step": 7296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0132813075120106e-05,
+      "loss": 0.8516,
+      "step": 7297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0130126834160492e-05,
+      "loss": 0.8281,
+      "step": 7298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0127440347372493e-05,
+      "loss": 0.9629,
+      "step": 7299
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.01247536148213e-05,
+      "loss": 0.9014,
+      "step": 7300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.012206663657212e-05,
+      "loss": 1.0723,
+      "step": 7301
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0119379412690143e-05,
+      "loss": 1.1055,
+      "step": 7302
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0116691943240587e-05,
+      "loss": 1.0439,
+      "step": 7303
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0114004228288672e-05,
+      "loss": 0.9639,
+      "step": 7304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0111316267899623e-05,
+      "loss": 0.9932,
+      "step": 7305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.010862806213866e-05,
+      "loss": 1.0283,
+      "step": 7306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0105939611071018e-05,
+      "loss": 0.9482,
+      "step": 7307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.010325091476195e-05,
+      "loss": 0.7554,
+      "step": 7308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.010056197327669e-05,
+      "loss": 0.8672,
+      "step": 7309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.00978727866805e-05,
+      "loss": 0.9629,
+      "step": 7310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.009518335503863e-05,
+      "loss": 1.0176,
+      "step": 7311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0092493678416353e-05,
+      "loss": 1.1074,
+      "step": 7312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0089803756878934e-05,
+      "loss": 0.8311,
+      "step": 7313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.008711359049165e-05,
+      "loss": 1.0684,
+      "step": 7314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0084423179319784e-05,
+      "loss": 0.9307,
+      "step": 7315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0081732523428622e-05,
+      "loss": 0.8486,
+      "step": 7316
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0079041622883465e-05,
+      "loss": 0.7363,
+      "step": 7317
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0076350477749607e-05,
+      "loss": 0.8711,
+      "step": 7318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0073659088092356e-05,
+      "loss": 0.9375,
+      "step": 7319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0070967453977026e-05,
+      "loss": 0.9854,
+      "step": 7320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0068275575468936e-05,
+      "loss": 0.9238,
+      "step": 7321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.00655834526334e-05,
+      "loss": 0.9365,
+      "step": 7322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0062891085535767e-05,
+      "loss": 0.7803,
+      "step": 7323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.006019847424135e-05,
+      "loss": 0.9463,
+      "step": 7324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0057505618815503e-05,
+      "loss": 1.083,
+      "step": 7325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0054812519323583e-05,
+      "loss": 0.9512,
+      "step": 7326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0052119175830926e-05,
+      "loss": 1.0801,
+      "step": 7327
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0049425588402898e-05,
+      "loss": 1.0557,
+      "step": 7328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.004673175710487e-05,
+      "loss": 0.8066,
+      "step": 7329
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.00440376820022e-05,
+      "loss": 0.9365,
+      "step": 7330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.004134336316028e-05,
+      "loss": 1.1816,
+      "step": 7331
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.003864880064449e-05,
+      "loss": 1.0479,
+      "step": 7332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0035953994520208e-05,
+      "loss": 0.8662,
+      "step": 7333
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0033258944852846e-05,
+      "loss": 1.0547,
+      "step": 7334
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0030563651707792e-05,
+      "loss": 0.9404,
+      "step": 7335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0027868115150457e-05,
+      "loss": 1.084,
+      "step": 7336
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0025172335246247e-05,
+      "loss": 0.7344,
+      "step": 7337
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.00224763120606e-05,
+      "loss": 0.7266,
+      "step": 7338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.001978004565892e-05,
+      "loss": 1.0361,
+      "step": 7339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.001708353610665e-05,
+      "loss": 1.0039,
+      "step": 7340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0014386783469213e-05,
+      "loss": 0.9248,
+      "step": 7341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0011689787812072e-05,
+      "loss": 0.8232,
+      "step": 7342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0008992549200654e-05,
+      "loss": 0.9131,
+      "step": 7343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.000629506770043e-05,
+      "loss": 0.6841,
+      "step": 7344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0003597343376847e-05,
+      "loss": 1.0078,
+      "step": 7345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.000089937629538e-05,
+      "loss": 0.7744,
+      "step": 7346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9998201166521486e-05,
+      "loss": 1.0303,
+      "step": 7347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9995502714120667e-05,
+      "loss": 0.915,
+      "step": 7348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9992804019158384e-05,
+      "loss": 0.9414,
+      "step": 7349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.999010508170014e-05,
+      "loss": 0.8916,
+      "step": 7350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9987405901811422e-05,
+      "loss": 0.9111,
+      "step": 7351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.998470647955774e-05,
+      "loss": 0.8662,
+      "step": 7352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9982006815004587e-05,
+      "loss": 0.8398,
+      "step": 7353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9979306908217487e-05,
+      "loss": 0.8115,
+      "step": 7354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.997660675926196e-05,
+      "loss": 0.9268,
+      "step": 7355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9973906368203526e-05,
+      "loss": 0.8916,
+      "step": 7356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9971205735107713e-05,
+      "loss": 0.8018,
+      "step": 7357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9968504860040067e-05,
+      "loss": 0.9326,
+      "step": 7358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.996580374306612e-05,
+      "loss": 1.1592,
+      "step": 7359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9963102384251428e-05,
+      "loss": 0.9395,
+      "step": 7360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9960400783661542e-05,
+      "loss": 1.0889,
+      "step": 7361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.995769894136202e-05,
+      "loss": 0.9365,
+      "step": 7362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9954996857418428e-05,
+      "loss": 0.958,
+      "step": 7363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9952294531896336e-05,
+      "loss": 0.9609,
+      "step": 7364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.994959196486133e-05,
+      "loss": 0.9932,
+      "step": 7365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9946889156378985e-05,
+      "loss": 1.0029,
+      "step": 7366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9944186106514893e-05,
+      "loss": 0.9141,
+      "step": 7367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9941482815334652e-05,
+      "loss": 0.9922,
+      "step": 7368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9938779282903856e-05,
+      "loss": 0.9531,
+      "step": 7369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9936075509288122e-05,
+      "loss": 0.8662,
+      "step": 7370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.993337149455305e-05,
+      "loss": 0.8682,
+      "step": 7371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9930667238764267e-05,
+      "loss": 0.9932,
+      "step": 7372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9927962741987392e-05,
+      "loss": 0.9922,
+      "step": 7373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.992525800428806e-05,
+      "loss": 1.1191,
+      "step": 7374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9922553025731907e-05,
+      "loss": 0.9658,
+      "step": 7375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9919847806384578e-05,
+      "loss": 1.0566,
+      "step": 7376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9917142346311708e-05,
+      "loss": 0.9248,
+      "step": 7377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9914436645578962e-05,
+      "loss": 0.8574,
+      "step": 7378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.991173070425199e-05,
+      "loss": 0.9971,
+      "step": 7379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.990902452239647e-05,
+      "loss": 1.0488,
+      "step": 7380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9906318100078065e-05,
+      "loss": 0.9834,
+      "step": 7381
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9903611437362453e-05,
+      "loss": 1.0801,
+      "step": 7382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9900904534315317e-05,
+      "loss": 0.9336,
+      "step": 7383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9898197391002344e-05,
+      "loss": 1.0215,
+      "step": 7384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9895490007489233e-05,
+      "loss": 0.8838,
+      "step": 7385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9892782383841678e-05,
+      "loss": 0.958,
+      "step": 7386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9890074520125388e-05,
+      "loss": 0.9053,
+      "step": 7387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9887366416406077e-05,
+      "loss": 0.8828,
+      "step": 7388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9884658072749458e-05,
+      "loss": 0.7197,
+      "step": 7389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.988194948922126e-05,
+      "loss": 1.1152,
+      "step": 7390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.987924066588721e-05,
+      "loss": 0.9883,
+      "step": 7391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.987653160281304e-05,
+      "loss": 1.0098,
+      "step": 7392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.98738223000645e-05,
+      "loss": 1.0352,
+      "step": 7393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9871112757707325e-05,
+      "loss": 1.0723,
+      "step": 7394
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.986840297580727e-05,
+      "loss": 1.0283,
+      "step": 7395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.98656929544301e-05,
+      "loss": 0.7705,
+      "step": 7396
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9862982693641575e-05,
+      "loss": 0.9971,
+      "step": 7397
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.986027219350747e-05,
+      "loss": 0.9736,
+      "step": 7398
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9857561454093553e-05,
+      "loss": 0.877,
+      "step": 7399
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.985485047546562e-05,
+      "loss": 0.9736,
+      "step": 7400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9852139257689434e-05,
+      "loss": 1.1016,
+      "step": 7401
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9849427800830813e-05,
+      "loss": 0.998,
+      "step": 7402
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9846716104955534e-05,
+      "loss": 0.9873,
+      "step": 7403
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9844004170129422e-05,
+      "loss": 0.8203,
+      "step": 7404
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9841291996418277e-05,
+      "loss": 1.084,
+      "step": 7405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9838579583887917e-05,
+      "loss": 1.0088,
+      "step": 7406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9835866932604163e-05,
+      "loss": 0.9131,
+      "step": 7407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.983315404263285e-05,
+      "loss": 1.0703,
+      "step": 7408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9830440914039804e-05,
+      "loss": 1.1016,
+      "step": 7409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9827727546890864e-05,
+      "loss": 1.1484,
+      "step": 7410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.982501394125188e-05,
+      "loss": 0.9004,
+      "step": 7411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.98223000971887e-05,
+      "loss": 0.9199,
+      "step": 7412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.981958601476719e-05,
+      "loss": 1.123,
+      "step": 7413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9816871694053203e-05,
+      "loss": 0.9814,
+      "step": 7414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9814157135112606e-05,
+      "loss": 0.9805,
+      "step": 7415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.981144233801128e-05,
+      "loss": 0.9902,
+      "step": 7416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9808727302815107e-05,
+      "loss": 0.959,
+      "step": 7417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9806012029589966e-05,
+      "loss": 0.9072,
+      "step": 7418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.980329651840175e-05,
+      "loss": 0.9551,
+      "step": 7419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9800580769316362e-05,
+      "loss": 1.0371,
+      "step": 7420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9797864782399695e-05,
+      "loss": 0.9307,
+      "step": 7421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9795148557717667e-05,
+      "loss": 0.8711,
+      "step": 7422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9792432095336195e-05,
+      "loss": 0.8052,
+      "step": 7423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9789715395321184e-05,
+      "loss": 0.8574,
+      "step": 7424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.978699845773858e-05,
+      "loss": 0.9346,
+      "step": 7425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9784281282654302e-05,
+      "loss": 0.8428,
+      "step": 7426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.97815638701343e-05,
+      "loss": 0.9678,
+      "step": 7427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9778846220244497e-05,
+      "loss": 0.7783,
+      "step": 7428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9776128333050862e-05,
+      "loss": 1.0537,
+      "step": 7429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.977341020861934e-05,
+      "loss": 0.915,
+      "step": 7430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9770691847015896e-05,
+      "loss": 0.8447,
+      "step": 7431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9767973248306492e-05,
+      "loss": 0.875,
+      "step": 7432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.976525441255711e-05,
+      "loss": 0.8066,
+      "step": 7433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.976253533983372e-05,
+      "loss": 1.0029,
+      "step": 7434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9759816030202304e-05,
+      "loss": 0.9043,
+      "step": 7435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9757096483728857e-05,
+      "loss": 0.9619,
+      "step": 7436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9754376700479373e-05,
+      "loss": 0.9189,
+      "step": 7437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975165668051985e-05,
+      "loss": 1.0801,
+      "step": 7438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9748936423916305e-05,
+      "loss": 0.9375,
+      "step": 7439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.974621593073474e-05,
+      "loss": 0.8174,
+      "step": 7440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9743495201041183e-05,
+      "loss": 0.9746,
+      "step": 7441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9740774234901644e-05,
+      "loss": 0.957,
+      "step": 7442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9738053032382165e-05,
+      "loss": 0.9619,
+      "step": 7443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9735331593548778e-05,
+      "loss": 0.8037,
+      "step": 7444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9732609918467524e-05,
+      "loss": 1.0957,
+      "step": 7445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9729888007204444e-05,
+      "loss": 0.8877,
+      "step": 7446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.97271658598256e-05,
+      "loss": 0.7778,
+      "step": 7447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9724443476397046e-05,
+      "loss": 0.8594,
+      "step": 7448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9721720856984854e-05,
+      "loss": 0.915,
+      "step": 7449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9718998001655078e-05,
+      "loss": 1.0332,
+      "step": 7450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9716274910473812e-05,
+      "loss": 1.1484,
+      "step": 7451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9713551583507122e-05,
+      "loss": 0.7319,
+      "step": 7452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9710828020821103e-05,
+      "loss": 0.9775,
+      "step": 7453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.970810422248185e-05,
+      "loss": 0.8066,
+      "step": 7454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9705380188555454e-05,
+      "loss": 1.0576,
+      "step": 7455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9702655919108024e-05,
+      "loss": 0.9561,
+      "step": 7456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9699931414205672e-05,
+      "loss": 0.7822,
+      "step": 7457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9697206673914504e-05,
+      "loss": 0.7427,
+      "step": 7458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9694481698300656e-05,
+      "loss": 1.0986,
+      "step": 7459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9691756487430245e-05,
+      "loss": 1.0508,
+      "step": 7460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9689031041369406e-05,
+      "loss": 0.96,
+      "step": 7461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.968630536018428e-05,
+      "loss": 0.9092,
+      "step": 7462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9683579443941007e-05,
+      "loss": 0.9971,
+      "step": 7463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.968085329270574e-05,
+      "loss": 0.9746,
+      "step": 7464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9678126906544636e-05,
+      "loss": 0.9053,
+      "step": 7465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9675400285523846e-05,
+      "loss": 0.8984,
+      "step": 7466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9672673429709556e-05,
+      "loss": 0.812,
+      "step": 7467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9669946339167924e-05,
+      "loss": 0.9453,
+      "step": 7468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9667219013965135e-05,
+      "loss": 0.9893,
+      "step": 7469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9664491454167366e-05,
+      "loss": 0.9971,
+      "step": 7470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9661763659840814e-05,
+      "loss": 1.0762,
+      "step": 7471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9659035631051673e-05,
+      "loss": 1.0791,
+      "step": 7472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9656307367866143e-05,
+      "loss": 1.0332,
+      "step": 7473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.965357887035043e-05,
+      "loss": 1.0146,
+      "step": 7474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.965085013857075e-05,
+      "loss": 0.9053,
+      "step": 7475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9648121172593316e-05,
+      "loss": 1.0801,
+      "step": 7476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9645391972484356e-05,
+      "loss": 1.1543,
+      "step": 7477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.96426625383101e-05,
+      "loss": 0.8232,
+      "step": 7478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9639932870136782e-05,
+      "loss": 1.0312,
+      "step": 7479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9637202968030645e-05,
+      "loss": 0.9414,
+      "step": 7480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9634472832057932e-05,
+      "loss": 0.9165,
+      "step": 7481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.96317424622849e-05,
+      "loss": 1.0869,
+      "step": 7482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.96290118587778e-05,
+      "loss": 0.9453,
+      "step": 7483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9626281021602904e-05,
+      "loss": 1.0488,
+      "step": 7484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9623549950826477e-05,
+      "loss": 1.0234,
+      "step": 7485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9620818646514792e-05,
+      "loss": 1.0117,
+      "step": 7486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9618087108734132e-05,
+      "loss": 0.9346,
+      "step": 7487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9615355337550786e-05,
+      "loss": 1.0078,
+      "step": 7488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9612623333031045e-05,
+      "loss": 0.8379,
+      "step": 7489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9609891095241202e-05,
+      "loss": 0.9814,
+      "step": 7490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9607158624247566e-05,
+      "loss": 0.9512,
+      "step": 7491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9604425920116443e-05,
+      "loss": 0.8633,
+      "step": 7492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9601692982914155e-05,
+      "loss": 0.9854,
+      "step": 7493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9598959812707005e-05,
+      "loss": 1.0537,
+      "step": 7494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9596226409561336e-05,
+      "loss": 0.9453,
+      "step": 7495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.959349277354347e-05,
+      "loss": 1.083,
+      "step": 7496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.959075890471975e-05,
+      "loss": 1.0068,
+      "step": 7497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.958802480315652e-05,
+      "loss": 0.9258,
+      "step": 7498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.958529046892012e-05,
+      "loss": 0.873,
+      "step": 7499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9582555902076913e-05,
+      "loss": 0.9121,
+      "step": 7500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.957982110269325e-05,
+      "loss": 0.9883,
+      "step": 7501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.957708607083551e-05,
+      "loss": 0.9268,
+      "step": 7502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.957435080657005e-05,
+      "loss": 0.9805,
+      "step": 7503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9571615309963255e-05,
+      "loss": 0.8252,
+      "step": 7504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9568879581081505e-05,
+      "loss": 1.0166,
+      "step": 7505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9566143619991196e-05,
+      "loss": 0.8076,
+      "step": 7506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9563407426758704e-05,
+      "loss": 0.9277,
+      "step": 7507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9560671001450446e-05,
+      "loss": 0.9541,
+      "step": 7508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9557934344132814e-05,
+      "loss": 0.9238,
+      "step": 7509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.955519745487223e-05,
+      "loss": 1.0508,
+      "step": 7510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9552460333735097e-05,
+      "loss": 0.8789,
+      "step": 7511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9549722980787852e-05,
+      "loss": 0.686,
+      "step": 7512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.954698539609691e-05,
+      "loss": 0.8887,
+      "step": 7513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.954424757972871e-05,
+      "loss": 0.8975,
+      "step": 7514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9541509531749693e-05,
+      "loss": 0.9619,
+      "step": 7515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9538771252226303e-05,
+      "loss": 1.0488,
+      "step": 7516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9536032741224977e-05,
+      "loss": 1.0693,
+      "step": 7517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9533293998812193e-05,
+      "loss": 1.0596,
+      "step": 7518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.953055502505439e-05,
+      "loss": 0.8545,
+      "step": 7519
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.952781582001804e-05,
+      "loss": 0.8877,
+      "step": 7520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.952507638376963e-05,
+      "loss": 0.8662,
+      "step": 7521
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9522336716375634e-05,
+      "loss": 0.9258,
+      "step": 7522
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9519596817902515e-05,
+      "loss": 0.9365,
+      "step": 7523
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.951685668841679e-05,
+      "loss": 1.04,
+      "step": 7524
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9514116327984933e-05,
+      "loss": 0.8477,
+      "step": 7525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.951137573667346e-05,
+      "loss": 0.9785,
+      "step": 7526
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9508634914548863e-05,
+      "loss": 0.79,
+      "step": 7527
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9505893861677668e-05,
+      "loss": 0.9688,
+      "step": 7528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.950315257812638e-05,
+      "loss": 1.0029,
+      "step": 7529
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9500411063961528e-05,
+      "loss": 0.917,
+      "step": 7530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9497669319249634e-05,
+      "loss": 0.916,
+      "step": 7531
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9494927344057243e-05,
+      "loss": 1.0146,
+      "step": 7532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9492185138450887e-05,
+      "loss": 0.9541,
+      "step": 7533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9489442702497116e-05,
+      "loss": 0.9209,
+      "step": 7534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9486700036262473e-05,
+      "loss": 1.0801,
+      "step": 7535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9483957139813525e-05,
+      "loss": 0.9648,
+      "step": 7536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9481214013216824e-05,
+      "loss": 0.9316,
+      "step": 7537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9478470656538948e-05,
+      "loss": 1.0791,
+      "step": 7538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.947572706984646e-05,
+      "loss": 0.9443,
+      "step": 7539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.947298325320594e-05,
+      "loss": 0.9805,
+      "step": 7540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9470239206683982e-05,
+      "loss": 1.0605,
+      "step": 7541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9467494930347168e-05,
+      "loss": 0.9941,
+      "step": 7542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.946475042426209e-05,
+      "loss": 0.9502,
+      "step": 7543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.946200568849536e-05,
+      "loss": 0.8936,
+      "step": 7544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9459260723113577e-05,
+      "loss": 0.8467,
+      "step": 7545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.945651552818335e-05,
+      "loss": 1.0967,
+      "step": 7546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9453770103771308e-05,
+      "loss": 0.9072,
+      "step": 7547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9451024449944066e-05,
+      "loss": 1.0029,
+      "step": 7548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9448278566768255e-05,
+      "loss": 0.9102,
+      "step": 7549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.944553245431051e-05,
+      "loss": 0.8242,
+      "step": 7550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9442786112637473e-05,
+      "loss": 0.9746,
+      "step": 7551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9440039541815785e-05,
+      "loss": 0.9863,
+      "step": 7552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9437292741912098e-05,
+      "loss": 0.9805,
+      "step": 7553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9434545712993074e-05,
+      "loss": 0.6943,
+      "step": 7554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9431798455125364e-05,
+      "loss": 0.9277,
+      "step": 7555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.942905096837565e-05,
+      "loss": 0.9512,
+      "step": 7556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9426303252810595e-05,
+      "loss": 1.0645,
+      "step": 7557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9423555308496885e-05,
+      "loss": 0.8672,
+      "step": 7558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.94208071355012e-05,
+      "loss": 0.8232,
+      "step": 7559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.941805873389023e-05,
+      "loss": 1.0518,
+      "step": 7560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9415310103730674e-05,
+      "loss": 0.8281,
+      "step": 7561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.941256124508923e-05,
+      "loss": 0.9023,
+      "step": 7562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9409812158032608e-05,
+      "loss": 0.9678,
+      "step": 7563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.940706284262752e-05,
+      "loss": 1.0332,
+      "step": 7564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9404313298940673e-05,
+      "loss": 0.8877,
+      "step": 7565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9401563527038806e-05,
+      "loss": 1.002,
+      "step": 7566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.939881352698864e-05,
+      "loss": 0.9131,
+      "step": 7567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9396063298856914e-05,
+      "loss": 0.7969,
+      "step": 7568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.939331284271036e-05,
+      "loss": 1.0205,
+      "step": 7569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.939056215861573e-05,
+      "loss": 1.0176,
+      "step": 7570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9387811246639772e-05,
+      "loss": 0.9971,
+      "step": 7571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9385060106849247e-05,
+      "loss": 0.96,
+      "step": 7572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.938230873931091e-05,
+      "loss": 0.9072,
+      "step": 7573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9379557144091534e-05,
+      "loss": 0.9004,
+      "step": 7574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.937680532125789e-05,
+      "loss": 1.083,
+      "step": 7575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9374053270876754e-05,
+      "loss": 0.8779,
+      "step": 7576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9371300993014916e-05,
+      "loss": 0.8711,
+      "step": 7577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9368548487739165e-05,
+      "loss": 0.8799,
+      "step": 7578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.936579575511629e-05,
+      "loss": 0.8545,
+      "step": 7579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9363042795213103e-05,
+      "loss": 0.9707,
+      "step": 7580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9360289608096395e-05,
+      "loss": 1.0938,
+      "step": 7581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9357536193832987e-05,
+      "loss": 0.9453,
+      "step": 7582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9354782552489694e-05,
+      "loss": 0.9766,
+      "step": 7583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9352028684133346e-05,
+      "loss": 0.873,
+      "step": 7584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.934927458883076e-05,
+      "loss": 0.8838,
+      "step": 7585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9346520266648776e-05,
+      "loss": 1.0215,
+      "step": 7586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9343765717654234e-05,
+      "loss": 0.8789,
+      "step": 7587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9341010941913976e-05,
+      "loss": 0.9653,
+      "step": 7588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9338255939494854e-05,
+      "loss": 0.9834,
+      "step": 7589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9335500710463725e-05,
+      "loss": 1.0273,
+      "step": 7590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9332745254887448e-05,
+      "loss": 0.8623,
+      "step": 7591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9329989572832884e-05,
+      "loss": 0.9697,
+      "step": 7592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.932723366436692e-05,
+      "loss": 0.7891,
+      "step": 7593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9324477529556417e-05,
+      "loss": 0.9668,
+      "step": 7594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9321721168468275e-05,
+      "loss": 0.9277,
+      "step": 7595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9318964581169366e-05,
+      "loss": 0.7651,
+      "step": 7596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9316207767726602e-05,
+      "loss": 0.9502,
+      "step": 7597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.931345072820687e-05,
+      "loss": 0.9316,
+      "step": 7598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9310693462677075e-05,
+      "loss": 1.0889,
+      "step": 7599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9307935971204132e-05,
+      "loss": 0.8535,
+      "step": 7600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9305178253854963e-05,
+      "loss": 0.916,
+      "step": 7601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9302420310696473e-05,
+      "loss": 0.916,
+      "step": 7602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.929966214179561e-05,
+      "loss": 0.9854,
+      "step": 7603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.929690374721928e-05,
+      "loss": 0.9717,
+      "step": 7604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9294145127034455e-05,
+      "loss": 0.9287,
+      "step": 7605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9291386281308055e-05,
+      "loss": 0.9463,
+      "step": 7606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9288627210107034e-05,
+      "loss": 1.0576,
+      "step": 7607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.928586791349835e-05,
+      "loss": 0.9805,
+      "step": 7608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.928310839154896e-05,
+      "loss": 0.8721,
+      "step": 7609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9280348644325826e-05,
+      "loss": 0.9883,
+      "step": 7610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9277588671895925e-05,
+      "loss": 0.9287,
+      "step": 7611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9274828474326232e-05,
+      "loss": 0.9316,
+      "step": 7612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.927206805168373e-05,
+      "loss": 1.0986,
+      "step": 7613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.92693074040354e-05,
+      "loss": 0.8975,
+      "step": 7614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9266546531448242e-05,
+      "loss": 0.9414,
+      "step": 7615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9263785433989248e-05,
+      "loss": 0.9326,
+      "step": 7616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9261024111725437e-05,
+      "loss": 0.8491,
+      "step": 7617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9258262564723795e-05,
+      "loss": 0.9551,
+      "step": 7618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.925550079305135e-05,
+      "loss": 1.041,
+      "step": 7619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.925273879677512e-05,
+      "loss": 0.8252,
+      "step": 7620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9249976575962137e-05,
+      "loss": 0.9707,
+      "step": 7621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9247214130679418e-05,
+      "loss": 0.9639,
+      "step": 7622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9244451460994012e-05,
+      "loss": 0.9375,
+      "step": 7623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9241688566972954e-05,
+      "loss": 0.7637,
+      "step": 7624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9238925448683294e-05,
+      "loss": 0.8477,
+      "step": 7625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.923616210619208e-05,
+      "loss": 1.1006,
+      "step": 7626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.923339853956638e-05,
+      "loss": 0.9531,
+      "step": 7627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9230634748873247e-05,
+      "loss": 0.9648,
+      "step": 7628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9227870734179762e-05,
+      "loss": 0.9033,
+      "step": 7629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9225106495552986e-05,
+      "loss": 0.8701,
+      "step": 7630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9222342033060012e-05,
+      "loss": 0.9639,
+      "step": 7631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.921957734676791e-05,
+      "loss": 0.9209,
+      "step": 7632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9216812436743788e-05,
+      "loss": 1.043,
+      "step": 7633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9214047303054728e-05,
+      "loss": 1.0088,
+      "step": 7634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9211281945767838e-05,
+      "loss": 0.8389,
+      "step": 7635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9208516364950224e-05,
+      "loss": 0.9648,
+      "step": 7636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9205750560669e-05,
+      "loss": 0.9219,
+      "step": 7637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9202984532991285e-05,
+      "loss": 1.0566,
+      "step": 7638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9200218281984205e-05,
+      "loss": 0.9336,
+      "step": 7639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9197451807714876e-05,
+      "loss": 0.8281,
+      "step": 7640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9194685110250445e-05,
+      "loss": 1.0957,
+      "step": 7641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9191918189658043e-05,
+      "loss": 0.9814,
+      "step": 7642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.918915104600482e-05,
+      "loss": 0.9814,
+      "step": 7643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.918638367935793e-05,
+      "loss": 1.001,
+      "step": 7644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9183616089784513e-05,
+      "loss": 1.0449,
+      "step": 7645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9180848277351748e-05,
+      "loss": 0.9404,
+      "step": 7646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.91780802421268e-05,
+      "loss": 0.9717,
+      "step": 7647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.917531198417683e-05,
+      "loss": 0.959,
+      "step": 7648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9172543503569025e-05,
+      "loss": 1.1348,
+      "step": 7649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.916977480037056e-05,
+      "loss": 0.7627,
+      "step": 7650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9167005874648638e-05,
+      "loss": 0.8545,
+      "step": 7651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.916423672647043e-05,
+      "loss": 0.8623,
+      "step": 7652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9161467355903158e-05,
+      "loss": 0.9229,
+      "step": 7653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9158697763014015e-05,
+      "loss": 0.8721,
+      "step": 7654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.915592794787021e-05,
+      "loss": 0.834,
+      "step": 7655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9153157910538958e-05,
+      "loss": 0.876,
+      "step": 7656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9150387651087486e-05,
+      "loss": 1.0078,
+      "step": 7657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9147617169583015e-05,
+      "loss": 0.8428,
+      "step": 7658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9144846466092773e-05,
+      "loss": 0.8838,
+      "step": 7659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9142075540684005e-05,
+      "loss": 0.9941,
+      "step": 7660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9139304393423953e-05,
+      "loss": 1.0049,
+      "step": 7661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9136533024379853e-05,
+      "loss": 1.0547,
+      "step": 7662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9133761433618973e-05,
+      "loss": 1.0088,
+      "step": 7663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.913098962120856e-05,
+      "loss": 0.9814,
+      "step": 7664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9128217587215887e-05,
+      "loss": 0.7012,
+      "step": 7665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.912544533170821e-05,
+      "loss": 1.0117,
+      "step": 7666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9122672854752817e-05,
+      "loss": 1.0361,
+      "step": 7667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9119900156416982e-05,
+      "loss": 1.0371,
+      "step": 7668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9117127236767992e-05,
+      "loss": 0.9326,
+      "step": 7669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9114354095873136e-05,
+      "loss": 0.9893,
+      "step": 7670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9111580733799712e-05,
+      "loss": 0.959,
+      "step": 7671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.910880715061502e-05,
+      "loss": 1.1055,
+      "step": 7672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.910603334638636e-05,
+      "loss": 1.1094,
+      "step": 7673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9103259321181057e-05,
+      "loss": 1.0498,
+      "step": 7674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.910048507506642e-05,
+      "loss": 0.9199,
+      "step": 7675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9097710608109776e-05,
+      "loss": 0.9893,
+      "step": 7676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9094935920378446e-05,
+      "loss": 0.9482,
+      "step": 7677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9092161011939775e-05,
+      "loss": 0.731,
+      "step": 7678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.908938588286109e-05,
+      "loss": 0.9951,
+      "step": 7679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9086610533209745e-05,
+      "loss": 0.9023,
+      "step": 7680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.908383496305308e-05,
+      "loss": 1.0029,
+      "step": 7681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9081059172458456e-05,
+      "loss": 1.001,
+      "step": 7682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9078283161493234e-05,
+      "loss": 0.8076,
+      "step": 7683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.907550693022478e-05,
+      "loss": 0.9688,
+      "step": 7684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9072730478720463e-05,
+      "loss": 0.8076,
+      "step": 7685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9069953807047655e-05,
+      "loss": 1.1279,
+      "step": 7686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9067176915273743e-05,
+      "loss": 0.9746,
+      "step": 7687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.906439980346611e-05,
+      "loss": 1.0547,
+      "step": 7688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9061622471692153e-05,
+      "loss": 0.8525,
+      "step": 7689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.905884492001927e-05,
+      "loss": 0.9932,
+      "step": 7690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9056067148514865e-05,
+      "loss": 0.9648,
+      "step": 7691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.905328915724634e-05,
+      "loss": 1.0146,
+      "step": 7692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.905051094628111e-05,
+      "loss": 0.8652,
+      "step": 7693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9047732515686598e-05,
+      "loss": 1.082,
+      "step": 7694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9044953865530224e-05,
+      "loss": 1.0264,
+      "step": 7695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9042174995879425e-05,
+      "loss": 1.0498,
+      "step": 7696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9039395906801626e-05,
+      "loss": 0.8594,
+      "step": 7697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9036616598364277e-05,
+      "loss": 1.0205,
+      "step": 7698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.903383707063481e-05,
+      "loss": 1.0557,
+      "step": 7699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9031057323680694e-05,
+      "loss": 0.875,
+      "step": 7700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9028277357569373e-05,
+      "loss": 0.9365,
+      "step": 7701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.902549717236831e-05,
+      "loss": 0.915,
+      "step": 7702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9022716768144977e-05,
+      "loss": 1.0215,
+      "step": 7703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.901993614496684e-05,
+      "loss": 0.8281,
+      "step": 7704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.901715530290138e-05,
+      "loss": 0.96,
+      "step": 7705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9014374242016083e-05,
+      "loss": 0.9824,
+      "step": 7706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9011592962378428e-05,
+      "loss": 1.0156,
+      "step": 7707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.900881146405592e-05,
+      "loss": 1.0127,
+      "step": 7708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9006029747116045e-05,
+      "loss": 0.9893,
+      "step": 7709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9003247811626322e-05,
+      "loss": 0.9531,
+      "step": 7710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.900046565765424e-05,
+      "loss": 0.9551,
+      "step": 7711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.899768328526734e-05,
+      "loss": 0.9902,
+      "step": 7712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8994900694533118e-05,
+      "loss": 0.9541,
+      "step": 7713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8992117885519112e-05,
+      "loss": 0.8906,
+      "step": 7714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.898933485829285e-05,
+      "loss": 0.8838,
+      "step": 7715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8986551612921867e-05,
+      "loss": 1.0596,
+      "step": 7716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8983768149473706e-05,
+      "loss": 1.1348,
+      "step": 7717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.898098446801591e-05,
+      "loss": 1.0469,
+      "step": 7718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8978200568616036e-05,
+      "loss": 0.6841,
+      "step": 7719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.897541645134164e-05,
+      "loss": 0.9375,
+      "step": 7720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8972632116260278e-05,
+      "loss": 0.915,
+      "step": 7721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8969847563439524e-05,
+      "loss": 0.8496,
+      "step": 7722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8967062792946948e-05,
+      "loss": 0.9443,
+      "step": 7723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.896427780485013e-05,
+      "loss": 0.9336,
+      "step": 7724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8961492599216652e-05,
+      "loss": 0.9258,
+      "step": 7725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8958707176114105e-05,
+      "loss": 0.917,
+      "step": 7726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.895592153561008e-05,
+      "loss": 1.0088,
+      "step": 7727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8953135677772178e-05,
+      "loss": 0.9717,
+      "step": 7728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8950349602668e-05,
+      "loss": 0.9795,
+      "step": 7729
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8947563310365163e-05,
+      "loss": 1.0332,
+      "step": 7730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8944776800931277e-05,
+      "loss": 1.042,
+      "step": 7731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8941990074433967e-05,
+      "loss": 1.0127,
+      "step": 7732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.893920313094085e-05,
+      "loss": 0.9873,
+      "step": 7733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8936415970519567e-05,
+      "loss": 1.0596,
+      "step": 7734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.893362859323775e-05,
+      "loss": 1.0088,
+      "step": 7735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8930840999163033e-05,
+      "loss": 0.791,
+      "step": 7736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8928053188363075e-05,
+      "loss": 0.9141,
+      "step": 7737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8925265160905526e-05,
+      "loss": 0.8896,
+      "step": 7738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8922476916858034e-05,
+      "loss": 1.085,
+      "step": 7739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8919688456288267e-05,
+      "loss": 1.0186,
+      "step": 7740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8916899779263892e-05,
+      "loss": 0.9404,
+      "step": 7741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.891411088585259e-05,
+      "loss": 1.0059,
+      "step": 7742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8911321776122022e-05,
+      "loss": 1.0713,
+      "step": 7743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8908532450139892e-05,
+      "loss": 0.9004,
+      "step": 7744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8905742907973873e-05,
+      "loss": 0.8799,
+      "step": 7745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8902953149691666e-05,
+      "loss": 0.998,
+      "step": 7746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.890016317536097e-05,
+      "loss": 0.6851,
+      "step": 7747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8897372985049487e-05,
+      "loss": 1.0713,
+      "step": 7748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.889458257882493e-05,
+      "loss": 0.9639,
+      "step": 7749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.889179195675501e-05,
+      "loss": 1.1162,
+      "step": 7750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8889001118907444e-05,
+      "loss": 1.0898,
+      "step": 7751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.888621006534997e-05,
+      "loss": 0.8545,
+      "step": 7752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8883418796150304e-05,
+      "loss": 0.8281,
+      "step": 7753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8880627311376192e-05,
+      "loss": 0.96,
+      "step": 7754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8877835611095375e-05,
+      "loss": 1.1191,
+      "step": 7755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8875043695375593e-05,
+      "loss": 1.0342,
+      "step": 7756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8872251564284605e-05,
+      "loss": 0.8828,
+      "step": 7757
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.886945921789016e-05,
+      "loss": 0.8701,
+      "step": 7758
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.886666665626003e-05,
+      "loss": 0.9414,
+      "step": 7759
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.886387387946197e-05,
+      "loss": 0.9414,
+      "step": 7760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.886108088756376e-05,
+      "loss": 0.8457,
+      "step": 7761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.885828768063317e-05,
+      "loss": 0.999,
+      "step": 7762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8855494258738003e-05,
+      "loss": 1.0107,
+      "step": 7763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8852700621946023e-05,
+      "loss": 0.9492,
+      "step": 7764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8849906770325043e-05,
+      "loss": 1.001,
+      "step": 7765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8847112703942845e-05,
+      "loss": 1.0742,
+      "step": 7766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8844318422867243e-05,
+      "loss": 0.9795,
+      "step": 7767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8841523927166036e-05,
+      "loss": 0.917,
+      "step": 7768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8838729216907057e-05,
+      "loss": 1.0586,
+      "step": 7769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8835934292158105e-05,
+      "loss": 0.9541,
+      "step": 7770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8833139152987012e-05,
+      "loss": 0.8965,
+      "step": 7771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8830343799461617e-05,
+      "loss": 0.9961,
+      "step": 7772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.882754823164974e-05,
+      "loss": 0.9736,
+      "step": 7773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8824752449619228e-05,
+      "loss": 0.8828,
+      "step": 7774
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8821956453437932e-05,
+      "loss": 0.8574,
+      "step": 7775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8819160243173697e-05,
+      "loss": 0.9512,
+      "step": 7776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.881636381889438e-05,
+      "loss": 0.8896,
+      "step": 7777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8813567180667833e-05,
+      "loss": 0.8906,
+      "step": 7778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.881077032856194e-05,
+      "loss": 0.8438,
+      "step": 7779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8807973262644556e-05,
+      "loss": 1.0527,
+      "step": 7780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.880517598298357e-05,
+      "loss": 0.6279,
+      "step": 7781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.880237848964685e-05,
+      "loss": 0.9023,
+      "step": 7782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.87995807827023e-05,
+      "loss": 1.0215,
+      "step": 7783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8796782862217794e-05,
+      "loss": 0.9883,
+      "step": 7784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8793984728261247e-05,
+      "loss": 0.9512,
+      "step": 7785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.879118638090055e-05,
+      "loss": 1.0723,
+      "step": 7786
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.878838782020361e-05,
+      "loss": 1.0732,
+      "step": 7787
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8785589046238347e-05,
+      "loss": 0.7393,
+      "step": 7788
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.878279005907268e-05,
+      "loss": 0.8271,
+      "step": 7789
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8779990858774516e-05,
+      "loss": 0.627,
+      "step": 7790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.87771914454118e-05,
+      "loss": 0.9775,
+      "step": 7791
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.877439181905246e-05,
+      "loss": 0.9116,
+      "step": 7792
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8771591979764443e-05,
+      "loss": 0.8994,
+      "step": 7793
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.876879192761567e-05,
+      "loss": 1.0596,
+      "step": 7794
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8765991662674116e-05,
+      "loss": 0.8652,
+      "step": 7795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.876319118500772e-05,
+      "loss": 1.0078,
+      "step": 7796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8760390494684452e-05,
+      "loss": 1.0078,
+      "step": 7797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.875758959177226e-05,
+      "loss": 0.9072,
+      "step": 7798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8754788476339134e-05,
+      "loss": 0.9785,
+      "step": 7799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.875198714845303e-05,
+      "loss": 0.9424,
+      "step": 7800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.874918560818194e-05,
+      "loss": 0.9883,
+      "step": 7801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8746383855593848e-05,
+      "loss": 0.9834,
+      "step": 7802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8743581890756736e-05,
+      "loss": 1.0156,
+      "step": 7803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8740779713738608e-05,
+      "loss": 1.0586,
+      "step": 7804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8737977324607466e-05,
+      "loss": 0.9902,
+      "step": 7805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.873517472343131e-05,
+      "loss": 1.0596,
+      "step": 7806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8732371910278153e-05,
+      "loss": 0.9492,
+      "step": 7807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8729568885216003e-05,
+      "loss": 0.9697,
+      "step": 7808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8726765648312897e-05,
+      "loss": 0.9658,
+      "step": 7809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.872396219963685e-05,
+      "loss": 0.8809,
+      "step": 7810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.87211585392559e-05,
+      "loss": 0.8955,
+      "step": 7811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.871835466723807e-05,
+      "loss": 0.9766,
+      "step": 7812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8715550583651427e-05,
+      "loss": 0.9092,
+      "step": 7813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8712746288563992e-05,
+      "loss": 0.9795,
+      "step": 7814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8709941782043836e-05,
+      "loss": 0.709,
+      "step": 7815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8707137064159e-05,
+      "loss": 0.876,
+      "step": 7816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.870433213497756e-05,
+      "loss": 1.0527,
+      "step": 7817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8701526994567578e-05,
+      "loss": 0.9229,
+      "step": 7818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8698721642997124e-05,
+      "loss": 0.8389,
+      "step": 7819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8695916080334276e-05,
+      "loss": 0.9316,
+      "step": 7820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8693110306647122e-05,
+      "loss": 0.957,
+      "step": 7821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8690304322003742e-05,
+      "loss": 1.0742,
+      "step": 7822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8687498126472238e-05,
+      "loss": 1.0586,
+      "step": 7823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.86846917201207e-05,
+      "loss": 1.002,
+      "step": 7824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8681885103017238e-05,
+      "loss": 1.0625,
+      "step": 7825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.867907827522995e-05,
+      "loss": 0.9658,
+      "step": 7826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.867627123682697e-05,
+      "loss": 1.0801,
+      "step": 7827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.867346398787639e-05,
+      "loss": 1.0322,
+      "step": 7828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8670656528446352e-05,
+      "loss": 0.855,
+      "step": 7829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8667848858604977e-05,
+      "loss": 0.8535,
+      "step": 7830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8665040978420406e-05,
+      "loss": 0.9648,
+      "step": 7831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.866223288796077e-05,
+      "loss": 0.8896,
+      "step": 7832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.865942458729422e-05,
+      "loss": 0.9414,
+      "step": 7833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8656616076488894e-05,
+      "loss": 0.8027,
+      "step": 7834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8653807355612962e-05,
+      "loss": 0.8115,
+      "step": 7835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8650998424734573e-05,
+      "loss": 0.8281,
+      "step": 7836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8648189283921898e-05,
+      "loss": 0.9561,
+      "step": 7837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.864537993324309e-05,
+      "loss": 0.9551,
+      "step": 7838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8642570372766346e-05,
+      "loss": 0.9844,
+      "step": 7839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8639760602559835e-05,
+      "loss": 0.9092,
+      "step": 7840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8636950622691738e-05,
+      "loss": 0.8887,
+      "step": 7841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8634140433230254e-05,
+      "loss": 1.0586,
+      "step": 7842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8631330034243574e-05,
+      "loss": 0.8662,
+      "step": 7843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.86285194257999e-05,
+      "loss": 0.8584,
+      "step": 7844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.862570860796743e-05,
+      "loss": 1.0654,
+      "step": 7845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8622897580814383e-05,
+      "loss": 1.0029,
+      "step": 7846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.862008634440897e-05,
+      "loss": 0.9961,
+      "step": 7847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.861727489881941e-05,
+      "loss": 0.8994,
+      "step": 7848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.861446324411393e-05,
+      "loss": 1.0566,
+      "step": 7849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8611651380360767e-05,
+      "loss": 0.7451,
+      "step": 7850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.860883930762815e-05,
+      "loss": 0.8662,
+      "step": 7851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.860602702598432e-05,
+      "loss": 0.8613,
+      "step": 7852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8603214535497518e-05,
+      "loss": 0.9277,
+      "step": 7853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8600401836236002e-05,
+      "loss": 0.9365,
+      "step": 7854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.859758892826803e-05,
+      "loss": 0.8848,
+      "step": 7855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8594775811661855e-05,
+      "loss": 1.0049,
+      "step": 7856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8591962486485753e-05,
+      "loss": 0.9639,
+      "step": 7857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8589148952807987e-05,
+      "loss": 0.8086,
+      "step": 7858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8586335210696834e-05,
+      "loss": 0.9121,
+      "step": 7859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.858352126022058e-05,
+      "loss": 1.0439,
+      "step": 7860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.85807071014475e-05,
+      "loss": 1.0889,
+      "step": 7861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.85778927344459e-05,
+      "loss": 1.0146,
+      "step": 7862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8575078159284075e-05,
+      "loss": 1.0391,
+      "step": 7863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.857226337603032e-05,
+      "loss": 1.1338,
+      "step": 7864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8569448384752935e-05,
+      "loss": 0.9697,
+      "step": 7865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.856663318552024e-05,
+      "loss": 1.0684,
+      "step": 7866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8563817778400557e-05,
+      "loss": 0.8789,
+      "step": 7867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8561002163462202e-05,
+      "loss": 0.9004,
+      "step": 7868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8558186340773497e-05,
+      "loss": 0.8965,
+      "step": 7869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8555370310402785e-05,
+      "loss": 0.9658,
+      "step": 7870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.855255407241839e-05,
+      "loss": 1.0225,
+      "step": 7871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8549737626888664e-05,
+      "loss": 0.8525,
+      "step": 7872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8546920973881946e-05,
+      "loss": 0.7568,
+      "step": 7873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8544104113466593e-05,
+      "loss": 0.9404,
+      "step": 7874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.854128704571096e-05,
+      "loss": 0.9629,
+      "step": 7875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.853846977068341e-05,
+      "loss": 1.0977,
+      "step": 7876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8535652288452307e-05,
+      "loss": 0.9629,
+      "step": 7877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8532834599086033e-05,
+      "loss": 0.8711,
+      "step": 7878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.853001670265295e-05,
+      "loss": 1.0352,
+      "step": 7879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.852719859922145e-05,
+      "loss": 0.9951,
+      "step": 7880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8524380288859915e-05,
+      "loss": 1.0508,
+      "step": 7881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.852156177163674e-05,
+      "loss": 0.9385,
+      "step": 7882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8518743047620324e-05,
+      "loss": 1.0283,
+      "step": 7883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8515924116879073e-05,
+      "loss": 0.8965,
+      "step": 7884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.851310497948138e-05,
+      "loss": 1.0312,
+      "step": 7885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8510285635495667e-05,
+      "loss": 0.9512,
+      "step": 7886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8507466084990345e-05,
+      "loss": 0.8457,
+      "step": 7887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.850464632803385e-05,
+      "loss": 0.9502,
+      "step": 7888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.850182636469459e-05,
+      "loss": 0.9336,
+      "step": 7889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8499006195041014e-05,
+      "loss": 0.7803,
+      "step": 7890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8496185819141547e-05,
+      "loss": 1.1367,
+      "step": 7891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8493365237064642e-05,
+      "loss": 1.0469,
+      "step": 7892
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8490544448878742e-05,
+      "loss": 0.8721,
+      "step": 7893
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8487723454652295e-05,
+      "loss": 1.0469,
+      "step": 7894
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8484902254453754e-05,
+      "loss": 1.0225,
+      "step": 7895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.84820808483516e-05,
+      "loss": 1.0205,
+      "step": 7896
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.847925923641428e-05,
+      "loss": 0.959,
+      "step": 7897
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8476437418710282e-05,
+      "loss": 0.9072,
+      "step": 7898
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8473615395308073e-05,
+      "loss": 0.8965,
+      "step": 7899
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.847079316627614e-05,
+      "loss": 0.9316,
+      "step": 7900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8467970731682964e-05,
+      "loss": 1.0918,
+      "step": 7901
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.846514809159705e-05,
+      "loss": 0.8857,
+      "step": 7902
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8462325246086878e-05,
+      "loss": 1.0537,
+      "step": 7903
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8459502195220965e-05,
+      "loss": 1.0498,
+      "step": 7904
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8456678939067805e-05,
+      "loss": 0.9189,
+      "step": 7905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.845385547769593e-05,
+      "loss": 1.0107,
+      "step": 7906
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8451031811173835e-05,
+      "loss": 1.1006,
+      "step": 7907
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8448207939570052e-05,
+      "loss": 0.9404,
+      "step": 7908
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8445383862953105e-05,
+      "loss": 0.9668,
+      "step": 7909
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8442559581391535e-05,
+      "loss": 1.0566,
+      "step": 7910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.843973509495387e-05,
+      "loss": 0.9697,
+      "step": 7911
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.843691040370866e-05,
+      "loss": 0.9121,
+      "step": 7912
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.843408550772444e-05,
+      "loss": 0.9082,
+      "step": 7913
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.843126040706977e-05,
+      "loss": 0.9131,
+      "step": 7914
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8428435101813204e-05,
+      "loss": 0.8467,
+      "step": 7915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.842560959202331e-05,
+      "loss": 0.9844,
+      "step": 7916
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8422783877768644e-05,
+      "loss": 0.9482,
+      "step": 7917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8419957959117786e-05,
+      "loss": 1.043,
+      "step": 7918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8417131836139308e-05,
+      "loss": 1.0576,
+      "step": 7919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8414305508901803e-05,
+      "loss": 0.8721,
+      "step": 7920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.841147897747384e-05,
+      "loss": 0.793,
+      "step": 7921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8408652241924028e-05,
+      "loss": 0.8047,
+      "step": 7922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8405825302320943e-05,
+      "loss": 1.0527,
+      "step": 7923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.840299815873321e-05,
+      "loss": 1.0684,
+      "step": 7924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.840017081122942e-05,
+      "loss": 0.9941,
+      "step": 7925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8397343259878183e-05,
+      "loss": 0.9541,
+      "step": 7926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8394515504748124e-05,
+      "loss": 0.9521,
+      "step": 7927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8391687545907865e-05,
+      "loss": 1.001,
+      "step": 7928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8388859383426028e-05,
+      "loss": 0.8081,
+      "step": 7929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8386031017371237e-05,
+      "loss": 0.7744,
+      "step": 7930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8383202447812145e-05,
+      "loss": 0.8252,
+      "step": 7931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8380373674817376e-05,
+      "loss": 1.0176,
+      "step": 7932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.837754469845559e-05,
+      "loss": 1.0537,
+      "step": 7933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.837471551879543e-05,
+      "loss": 0.9424,
+      "step": 7934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8371886135905552e-05,
+      "loss": 0.9033,
+      "step": 7935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8369056549854618e-05,
+      "loss": 0.876,
+      "step": 7936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8366226760711292e-05,
+      "loss": 0.9043,
+      "step": 7937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8363396768544255e-05,
+      "loss": 0.8696,
+      "step": 7938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8360566573422167e-05,
+      "loss": 0.9414,
+      "step": 7939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8357736175413716e-05,
+      "loss": 0.9102,
+      "step": 7940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.835490557458759e-05,
+      "loss": 0.8516,
+      "step": 7941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8352074771012477e-05,
+      "loss": 1.1172,
+      "step": 7942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.834924376475707e-05,
+      "loss": 0.8896,
+      "step": 7943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.834641255589007e-05,
+      "loss": 1.0244,
+      "step": 7944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8343581144480184e-05,
+      "loss": 0.9268,
+      "step": 7945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.834074953059612e-05,
+      "loss": 0.9648,
+      "step": 7946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8337917714306596e-05,
+      "loss": 0.8994,
+      "step": 7947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.833508569568033e-05,
+      "loss": 0.835,
+      "step": 7948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8332253474786048e-05,
+      "loss": 0.7539,
+      "step": 7949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8329421051692477e-05,
+      "loss": 0.9629,
+      "step": 7950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8326588426468354e-05,
+      "loss": 0.8486,
+      "step": 7951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8323755599182418e-05,
+      "loss": 1.1416,
+      "step": 7952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.832092256990341e-05,
+      "loss": 0.8203,
+      "step": 7953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8318089338700082e-05,
+      "loss": 0.9746,
+      "step": 7954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8315255905641197e-05,
+      "loss": 0.9375,
+      "step": 7955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8312422270795495e-05,
+      "loss": 0.8486,
+      "step": 7956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8309588434231757e-05,
+      "loss": 0.8633,
+      "step": 7957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.830675439601874e-05,
+      "loss": 0.9609,
+      "step": 7958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.830392015622523e-05,
+      "loss": 0.833,
+      "step": 7959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.830108571491999e-05,
+      "loss": 0.8477,
+      "step": 7960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8298251072171817e-05,
+      "loss": 0.7441,
+      "step": 7961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8295416228049493e-05,
+      "loss": 0.8877,
+      "step": 7962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.829258118262182e-05,
+      "loss": 0.8325,
+      "step": 7963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8289745935957577e-05,
+      "loss": 0.9346,
+      "step": 7964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.828691048812559e-05,
+      "loss": 0.8916,
+      "step": 7965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.828407483919465e-05,
+      "loss": 0.8506,
+      "step": 7966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.828123898923358e-05,
+      "loss": 0.9062,
+      "step": 7967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8278402938311192e-05,
+      "loss": 0.8369,
+      "step": 7968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8275566686496315e-05,
+      "loss": 0.9492,
+      "step": 7969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8272730233857767e-05,
+      "loss": 0.7402,
+      "step": 7970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8269893580464387e-05,
+      "loss": 1.0527,
+      "step": 7971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8267056726385012e-05,
+      "loss": 1.0693,
+      "step": 7972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8264219671688482e-05,
+      "loss": 0.8984,
+      "step": 7973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.826138241644364e-05,
+      "loss": 0.8574,
+      "step": 7974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.825854496071935e-05,
+      "loss": 0.8594,
+      "step": 7975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.825570730458446e-05,
+      "loss": 0.8496,
+      "step": 7976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.825286944810783e-05,
+      "loss": 0.7773,
+      "step": 7977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8250031391358326e-05,
+      "loss": 0.8145,
+      "step": 7978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8247193134404827e-05,
+      "loss": 0.9453,
+      "step": 7979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8244354677316202e-05,
+      "loss": 1.0225,
+      "step": 7980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.824151602016134e-05,
+      "loss": 0.916,
+      "step": 7981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8238677163009117e-05,
+      "loss": 1.041,
+      "step": 7982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.823583810592843e-05,
+      "loss": 0.875,
+      "step": 7983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.823299884898817e-05,
+      "loss": 1.0186,
+      "step": 7984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.823015939225725e-05,
+      "loss": 1.0938,
+      "step": 7985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8227319735804557e-05,
+      "loss": 0.916,
+      "step": 7986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8224479879699013e-05,
+      "loss": 0.9766,
+      "step": 7987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8221639824009525e-05,
+      "loss": 1.0186,
+      "step": 7988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8218799568805025e-05,
+      "loss": 1.1113,
+      "step": 7989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8215959114154424e-05,
+      "loss": 1.0166,
+      "step": 7990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.821311846012666e-05,
+      "loss": 0.9814,
+      "step": 7991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8210277606790667e-05,
+      "loss": 1.126,
+      "step": 7992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.820743655421538e-05,
+      "loss": 0.7434,
+      "step": 7993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.820459530246974e-05,
+      "loss": 0.8447,
+      "step": 7994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.820175385162271e-05,
+      "loss": 1.0703,
+      "step": 7995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8198912201743226e-05,
+      "loss": 1.1035,
+      "step": 7996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8196070352900262e-05,
+      "loss": 0.8213,
+      "step": 7997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.819322830516277e-05,
+      "loss": 1.1719,
+      "step": 7998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8190386058599727e-05,
+      "loss": 0.8955,
+      "step": 7999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8187543613280092e-05,
+      "loss": 0.8496,
+      "step": 8000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8184700969272864e-05,
+      "loss": 0.9805,
+      "step": 8001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8181858126647005e-05,
+      "loss": 0.9951,
+      "step": 8002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8179015085471516e-05,
+      "loss": 0.9434,
+      "step": 8003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8176171845815382e-05,
+      "loss": 0.8184,
+      "step": 8004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8173328407747607e-05,
+      "loss": 1.083,
+      "step": 8005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8170484771337184e-05,
+      "loss": 0.7529,
+      "step": 8006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8167640936653123e-05,
+      "loss": 0.9736,
+      "step": 8007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8164796903764437e-05,
+      "loss": 0.7246,
+      "step": 8008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8161952672740148e-05,
+      "loss": 0.8379,
+      "step": 8009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8159108243649266e-05,
+      "loss": 0.9141,
+      "step": 8010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8156263616560817e-05,
+      "loss": 0.8633,
+      "step": 8011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8153418791543842e-05,
+      "loss": 0.8838,
+      "step": 8012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8150573768667374e-05,
+      "loss": 0.9434,
+      "step": 8013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.814772854800045e-05,
+      "loss": 0.8486,
+      "step": 8014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.814488312961211e-05,
+      "loss": 1.0098,
+      "step": 8015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8142037513571418e-05,
+      "loss": 0.8799,
+      "step": 8016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8139191699947416e-05,
+      "loss": 0.8301,
+      "step": 8017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8136345688809168e-05,
+      "loss": 0.959,
+      "step": 8018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8133499480225738e-05,
+      "loss": 0.875,
+      "step": 8019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8130653074266188e-05,
+      "loss": 0.8896,
+      "step": 8020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8127806470999614e-05,
+      "loss": 0.9688,
+      "step": 8021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8124959670495066e-05,
+      "loss": 0.9092,
+      "step": 8022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.812211267282165e-05,
+      "loss": 1.084,
+      "step": 8023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8119265478048446e-05,
+      "loss": 0.8506,
+      "step": 8024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8116418086244546e-05,
+      "loss": 0.9082,
+      "step": 8025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8113570497479048e-05,
+      "loss": 0.8594,
+      "step": 8026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8110722711821056e-05,
+      "loss": 0.8984,
+      "step": 8027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8107874729339672e-05,
+      "loss": 0.9717,
+      "step": 8028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8105026550104017e-05,
+      "loss": 0.9805,
+      "step": 8029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.81021781741832e-05,
+      "loss": 0.8066,
+      "step": 8030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.809932960164635e-05,
+      "loss": 0.8682,
+      "step": 8031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8096480832562592e-05,
+      "loss": 0.9561,
+      "step": 8032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8093631867001047e-05,
+      "loss": 0.9268,
+      "step": 8033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8090782705030865e-05,
+      "loss": 0.8838,
+      "step": 8034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.808793334672118e-05,
+      "loss": 1.0127,
+      "step": 8035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.808508379214114e-05,
+      "loss": 0.918,
+      "step": 8036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8082234041359887e-05,
+      "loss": 0.8252,
+      "step": 8037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8079384094446593e-05,
+      "loss": 0.9492,
+      "step": 8038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.80765339514704e-05,
+      "loss": 0.9346,
+      "step": 8039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8073683612500483e-05,
+      "loss": 0.8516,
+      "step": 8040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8070833077606008e-05,
+      "loss": 1.082,
+      "step": 8041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8067982346856154e-05,
+      "loss": 0.8848,
+      "step": 8042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8065131420320088e-05,
+      "loss": 0.9531,
+      "step": 8043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8062280298067004e-05,
+      "loss": 0.8291,
+      "step": 8044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8059428980166087e-05,
+      "loss": 0.9512,
+      "step": 8045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8056577466686533e-05,
+      "loss": 0.9795,
+      "step": 8046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8053725757697535e-05,
+      "loss": 0.6614,
+      "step": 8047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.80508738532683e-05,
+      "loss": 0.958,
+      "step": 8048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.804802175346803e-05,
+      "loss": 0.8887,
+      "step": 8049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8045169458365946e-05,
+      "loss": 0.9199,
+      "step": 8050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8042316968031253e-05,
+      "loss": 0.8799,
+      "step": 8051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8039464282533186e-05,
+      "loss": 0.9541,
+      "step": 8052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8036611401940957e-05,
+      "loss": 1.1064,
+      "step": 8053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.803375832632381e-05,
+      "loss": 0.9668,
+      "step": 8054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8030905055750973e-05,
+      "loss": 0.9922,
+      "step": 8055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.802805159029169e-05,
+      "loss": 1.0703,
+      "step": 8056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8025197930015195e-05,
+      "loss": 0.9268,
+      "step": 8057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.802234407499076e-05,
+      "loss": 0.9102,
+      "step": 8058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8019490025287617e-05,
+      "loss": 0.9912,
+      "step": 8059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8016635780975044e-05,
+      "loss": 0.8721,
+      "step": 8060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8013781342122287e-05,
+      "loss": 0.9893,
+      "step": 8061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.801092670879863e-05,
+      "loss": 0.8213,
+      "step": 8062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8008071881073342e-05,
+      "loss": 1.0605,
+      "step": 8063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8005216859015703e-05,
+      "loss": 1.0684,
+      "step": 8064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8002361642694985e-05,
+      "loss": 0.9336,
+      "step": 8065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7999506232180487e-05,
+      "loss": 0.9375,
+      "step": 8066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7996650627541502e-05,
+      "loss": 0.9443,
+      "step": 8067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7993794828847326e-05,
+      "loss": 0.9727,
+      "step": 8068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.799093883616725e-05,
+      "loss": 0.8438,
+      "step": 8069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7988082649570595e-05,
+      "loss": 0.9463,
+      "step": 8070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7985226269126662e-05,
+      "loss": 0.8291,
+      "step": 8071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7982369694904776e-05,
+      "loss": 1.0986,
+      "step": 8072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.797951292697425e-05,
+      "loss": 0.9932,
+      "step": 8073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7976655965404415e-05,
+      "loss": 0.8555,
+      "step": 8074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7973798810264593e-05,
+      "loss": 0.9971,
+      "step": 8075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.797094146162413e-05,
+      "loss": 0.8555,
+      "step": 8076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7968083919552355e-05,
+      "loss": 0.9424,
+      "step": 8077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.796522618411862e-05,
+      "loss": 0.9844,
+      "step": 8078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7962368255392266e-05,
+      "loss": 0.9385,
+      "step": 8079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7959510133442654e-05,
+      "loss": 1.1006,
+      "step": 8080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7956651818339136e-05,
+      "loss": 1.0752,
+      "step": 8081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7953793310151086e-05,
+      "loss": 0.9053,
+      "step": 8082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7950934608947852e-05,
+      "loss": 0.9971,
+      "step": 8083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7948075714798828e-05,
+      "loss": 0.9521,
+      "step": 8084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.794521662777338e-05,
+      "loss": 0.9131,
+      "step": 8085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7942357347940885e-05,
+      "loss": 0.9463,
+      "step": 8086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7939497875370737e-05,
+      "loss": 0.9209,
+      "step": 8087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7936638210132323e-05,
+      "loss": 0.9297,
+      "step": 8088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.793377835229504e-05,
+      "loss": 0.9434,
+      "step": 8089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7930918301928292e-05,
+      "loss": 0.7881,
+      "step": 8090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7928058059101483e-05,
+      "loss": 1.0283,
+      "step": 8091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7925197623884015e-05,
+      "loss": 0.9473,
+      "step": 8092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.792233699634531e-05,
+      "loss": 1.0039,
+      "step": 8093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7919476176554785e-05,
+      "loss": 0.9629,
+      "step": 8094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7916615164581863e-05,
+      "loss": 0.9082,
+      "step": 8095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7913753960495977e-05,
+      "loss": 0.9688,
+      "step": 8096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.791089256436655e-05,
+      "loss": 1.0381,
+      "step": 8097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7908030976263022e-05,
+      "loss": 0.8887,
+      "step": 8098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.790516919625485e-05,
+      "loss": 0.957,
+      "step": 8099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7902307224411464e-05,
+      "loss": 1.0527,
+      "step": 8100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7899445060802324e-05,
+      "loss": 0.9277,
+      "step": 8101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.789658270549688e-05,
+      "loss": 0.9824,
+      "step": 8102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.78937201585646e-05,
+      "loss": 0.8145,
+      "step": 8103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7890857420074945e-05,
+      "loss": 0.8335,
+      "step": 8104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.788799449009738e-05,
+      "loss": 1.0078,
+      "step": 8105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.78851313687014e-05,
+      "loss": 0.7578,
+      "step": 8106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.788226805595646e-05,
+      "loss": 0.8486,
+      "step": 8107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7879404551932063e-05,
+      "loss": 1.0078,
+      "step": 8108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7876540856697684e-05,
+      "loss": 0.8955,
+      "step": 8109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7873676970322826e-05,
+      "loss": 1.1602,
+      "step": 8110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7870812892876983e-05,
+      "loss": 1.0918,
+      "step": 8111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7867948624429656e-05,
+      "loss": 0.9648,
+      "step": 8112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7865084165050355e-05,
+      "loss": 0.7866,
+      "step": 8113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7862219514808598e-05,
+      "loss": 0.9355,
+      "step": 8114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.785935467377389e-05,
+      "loss": 0.9077,
+      "step": 8115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7856489642015758e-05,
+      "loss": 0.9814,
+      "step": 8116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7853624419603726e-05,
+      "loss": 0.9004,
+      "step": 8117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7850759006607323e-05,
+      "loss": 1.0156,
+      "step": 8118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7847893403096097e-05,
+      "loss": 0.8965,
+      "step": 8119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.784502760913957e-05,
+      "loss": 0.957,
+      "step": 8120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7842161624807297e-05,
+      "loss": 0.9727,
+      "step": 8121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7839295450168822e-05,
+      "loss": 1.002,
+      "step": 8122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7836429085293706e-05,
+      "loss": 0.8398,
+      "step": 8123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.78335625302515e-05,
+      "loss": 0.9033,
+      "step": 8124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7830695785111773e-05,
+      "loss": 0.7061,
+      "step": 8125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.782782884994408e-05,
+      "loss": 1.042,
+      "step": 8126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7824961724818008e-05,
+      "loss": 0.9365,
+      "step": 8127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.782209440980312e-05,
+      "loss": 0.8438,
+      "step": 8128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7819226904969016e-05,
+      "loss": 0.9082,
+      "step": 8129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7816359210385262e-05,
+      "loss": 1.0176,
+      "step": 8130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.781349132612146e-05,
+      "loss": 0.9482,
+      "step": 8131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.78106232522472e-05,
+      "loss": 0.9307,
+      "step": 8132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7807754988832086e-05,
+      "loss": 1.0918,
+      "step": 8133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7804886535945715e-05,
+      "loss": 0.9365,
+      "step": 8134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7802017893657712e-05,
+      "loss": 0.9844,
+      "step": 8135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7799149062037672e-05,
+      "loss": 0.8867,
+      "step": 8136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7796280041155222e-05,
+      "loss": 0.8623,
+      "step": 8137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.779341083107998e-05,
+      "loss": 0.8057,
+      "step": 8138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7790541431881584e-05,
+      "loss": 1.0186,
+      "step": 8139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7787671843629652e-05,
+      "loss": 0.7627,
+      "step": 8140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.778480206639383e-05,
+      "loss": 0.7637,
+      "step": 8141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7781932100243757e-05,
+      "loss": 0.9961,
+      "step": 8142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.777906194524908e-05,
+      "loss": 1.0996,
+      "step": 8143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.777619160147944e-05,
+      "loss": 0.7852,
+      "step": 8144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.77733210690045e-05,
+      "loss": 0.8506,
+      "step": 8145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7770450347893922e-05,
+      "loss": 0.8818,
+      "step": 8146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7767579438217365e-05,
+      "loss": 1.0137,
+      "step": 8147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7764708340044494e-05,
+      "loss": 0.9463,
+      "step": 8148
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7761837053444994e-05,
+      "loss": 0.6169,
+      "step": 8149
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7758965578488535e-05,
+      "loss": 0.9463,
+      "step": 8150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.77560939152448e-05,
+      "loss": 0.9727,
+      "step": 8151
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7753222063783465e-05,
+      "loss": 0.9258,
+      "step": 8152
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7750350024174247e-05,
+      "loss": 1.0391,
+      "step": 8153
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7747477796486814e-05,
+      "loss": 0.8633,
+      "step": 8154
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.774460538079089e-05,
+      "loss": 0.8037,
+      "step": 8155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.774173277715616e-05,
+      "loss": 0.9941,
+      "step": 8156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.773885998565235e-05,
+      "loss": 0.7656,
+      "step": 8157
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7735987006349167e-05,
+      "loss": 1.0146,
+      "step": 8158
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.773311383931633e-05,
+      "loss": 0.7383,
+      "step": 8159
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7730240484623555e-05,
+      "loss": 0.8359,
+      "step": 8160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.772736694234059e-05,
+      "loss": 0.9492,
+      "step": 8161
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7724493212537144e-05,
+      "loss": 0.8936,
+      "step": 8162
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7721619295282968e-05,
+      "loss": 0.9004,
+      "step": 8163
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7718745190647803e-05,
+      "loss": 1.1025,
+      "step": 8164
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.771587089870139e-05,
+      "loss": 0.9932,
+      "step": 8165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7712996419513477e-05,
+      "loss": 1.0293,
+      "step": 8166
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.771012175315383e-05,
+      "loss": 0.9844,
+      "step": 8167
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7707246899692203e-05,
+      "loss": 1.1426,
+      "step": 8168
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7704371859198356e-05,
+      "loss": 0.8428,
+      "step": 8169
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7701496631742063e-05,
+      "loss": 0.8428,
+      "step": 8170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7698621217393103e-05,
+      "loss": 0.9014,
+      "step": 8171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7695745616221236e-05,
+      "loss": 1.1035,
+      "step": 8172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.769286982829626e-05,
+      "loss": 0.9482,
+      "step": 8173
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7689993853687958e-05,
+      "loss": 0.9727,
+      "step": 8174
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7687117692466122e-05,
+      "loss": 0.7236,
+      "step": 8175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.768424134470054e-05,
+      "loss": 1.1133,
+      "step": 8176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.768136481046103e-05,
+      "loss": 0.9404,
+      "step": 8177
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7678488089817378e-05,
+      "loss": 0.8721,
+      "step": 8178
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.76756111828394e-05,
+      "loss": 0.915,
+      "step": 8179
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7672734089596916e-05,
+      "loss": 1.0078,
+      "step": 8180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.766985681015974e-05,
+      "loss": 1.0391,
+      "step": 8181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.766697934459769e-05,
+      "loss": 0.7769,
+      "step": 8182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7664101692980606e-05,
+      "loss": 0.8916,
+      "step": 8183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7661223855378314e-05,
+      "loss": 0.8545,
+      "step": 8184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7658345831860645e-05,
+      "loss": 0.8223,
+      "step": 8185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7655467622497448e-05,
+      "loss": 1.0615,
+      "step": 8186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.765258922735856e-05,
+      "loss": 0.8828,
+      "step": 8187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.764971064651384e-05,
+      "loss": 1.0469,
+      "step": 8188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7646831880033147e-05,
+      "loss": 1.1494,
+      "step": 8189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7643952927986322e-05,
+      "loss": 0.8374,
+      "step": 8190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.764107379044325e-05,
+      "loss": 0.8477,
+      "step": 8191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7638194467473776e-05,
+      "loss": 1.0283,
+      "step": 8192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.76353149591478e-05,
+      "loss": 0.8828,
+      "step": 8193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7632435265535175e-05,
+      "loss": 1.1338,
+      "step": 8194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7629555386705798e-05,
+      "loss": 0.8682,
+      "step": 8195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7626675322729544e-05,
+      "loss": 0.9912,
+      "step": 8196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.762379507367631e-05,
+      "loss": 0.874,
+      "step": 8197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.762091463961599e-05,
+      "loss": 0.9717,
+      "step": 8198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7618034020618488e-05,
+      "loss": 0.7979,
+      "step": 8199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7615153216753703e-05,
+      "loss": 0.9258,
+      "step": 8200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7612272228091543e-05,
+      "loss": 0.9287,
+      "step": 8201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7609391054701925e-05,
+      "loss": 0.9629,
+      "step": 8202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.760650969665476e-05,
+      "loss": 0.8809,
+      "step": 8203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7603628154019976e-05,
+      "loss": 1.1084,
+      "step": 8204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7600746426867496e-05,
+      "loss": 0.9727,
+      "step": 8205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7597864515267262e-05,
+      "loss": 0.9639,
+      "step": 8206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7594982419289196e-05,
+      "loss": 0.8989,
+      "step": 8207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7592100139003243e-05,
+      "loss": 0.8818,
+      "step": 8208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7589217674479342e-05,
+      "loss": 1.2207,
+      "step": 8209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7586335025787457e-05,
+      "loss": 0.8174,
+      "step": 8210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.758345219299752e-05,
+      "loss": 0.9668,
+      "step": 8211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.758056917617951e-05,
+      "loss": 0.8838,
+      "step": 8212
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.757768597540338e-05,
+      "loss": 0.875,
+      "step": 8213
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7574802590739095e-05,
+      "loss": 1.0059,
+      "step": 8214
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7571919022256622e-05,
+      "loss": 1.0283,
+      "step": 8215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7569035270025947e-05,
+      "loss": 1.0566,
+      "step": 8216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.756615133411704e-05,
+      "loss": 0.96,
+      "step": 8217
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7563267214599904e-05,
+      "loss": 0.8506,
+      "step": 8218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7560382911544503e-05,
+      "loss": 1.0439,
+      "step": 8219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.755749842502085e-05,
+      "loss": 0.9766,
+      "step": 8220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.755461375509893e-05,
+      "loss": 0.9814,
+      "step": 8221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7551728901848758e-05,
+      "loss": 1.0381,
+      "step": 8222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.754884386534033e-05,
+      "loss": 1.0557,
+      "step": 8223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7545958645643666e-05,
+      "loss": 0.9775,
+      "step": 8224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7543073242828774e-05,
+      "loss": 1.0938,
+      "step": 8225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.754018765696568e-05,
+      "loss": 1.0566,
+      "step": 8226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7537301888124407e-05,
+      "loss": 0.8877,
+      "step": 8227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.753441593637498e-05,
+      "loss": 1.0283,
+      "step": 8228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7531529801787434e-05,
+      "loss": 0.9648,
+      "step": 8229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7528643484431814e-05,
+      "loss": 0.8555,
+      "step": 8230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7525756984378157e-05,
+      "loss": 1.0664,
+      "step": 8231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.752287030169651e-05,
+      "loss": 0.8457,
+      "step": 8232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7519983436456922e-05,
+      "loss": 0.9033,
+      "step": 8233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7517096388729458e-05,
+      "loss": 0.9434,
+      "step": 8234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7514209158584164e-05,
+      "loss": 1.0674,
+      "step": 8235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.751132174609112e-05,
+      "loss": 0.876,
+      "step": 8236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.750843415132038e-05,
+      "loss": 1.125,
+      "step": 8237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7505546374342032e-05,
+      "loss": 0.8193,
+      "step": 8238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.750265841522614e-05,
+      "loss": 0.8828,
+      "step": 8239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7499770274042797e-05,
+      "loss": 1.0029,
+      "step": 8240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7496881950862082e-05,
+      "loss": 1.0576,
+      "step": 8241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7493993445754096e-05,
+      "loss": 0.9678,
+      "step": 8242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.749110475878892e-05,
+      "loss": 0.8779,
+      "step": 8243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7488215890036668e-05,
+      "loss": 1.0771,
+      "step": 8244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7485326839567433e-05,
+      "loss": 0.9521,
+      "step": 8245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.748243760745134e-05,
+      "loss": 0.8418,
+      "step": 8246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.747954819375848e-05,
+      "loss": 1.0029,
+      "step": 8247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7476658598558982e-05,
+      "loss": 0.7197,
+      "step": 8248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.747376882192297e-05,
+      "loss": 0.9326,
+      "step": 8249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7470878863920573e-05,
+      "loss": 0.8848,
+      "step": 8250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7467988724621913e-05,
+      "loss": 0.958,
+      "step": 8251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7465098404097132e-05,
+      "loss": 0.958,
+      "step": 8252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.746220790241636e-05,
+      "loss": 0.8789,
+      "step": 8253
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7459317219649758e-05,
+      "loss": 1.0088,
+      "step": 8254
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7456426355867453e-05,
+      "loss": 0.8398,
+      "step": 8255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.745353531113962e-05,
+      "loss": 0.9824,
+      "step": 8256
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7450644085536397e-05,
+      "loss": 0.9902,
+      "step": 8257
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7447752679127958e-05,
+      "loss": 0.8369,
+      "step": 8258
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7444861091984463e-05,
+      "loss": 0.9531,
+      "step": 8259
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7441969324176085e-05,
+      "loss": 0.7832,
+      "step": 8260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.743907737577299e-05,
+      "loss": 0.9443,
+      "step": 8261
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7436185246845376e-05,
+      "loss": 1.1104,
+      "step": 8262
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7433292937463406e-05,
+      "loss": 1.0596,
+      "step": 8263
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7430400447697283e-05,
+      "loss": 0.9639,
+      "step": 8264
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7427507777617188e-05,
+      "loss": 1.0469,
+      "step": 8265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7424614927293327e-05,
+      "loss": 1.0801,
+      "step": 8266
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7421721896795895e-05,
+      "loss": 0.6406,
+      "step": 8267
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.74188286861951e-05,
+      "loss": 0.9258,
+      "step": 8268
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.741593529556115e-05,
+      "loss": 0.9385,
+      "step": 8269
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7413041724964256e-05,
+      "loss": 0.9141,
+      "step": 8270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7410147974474647e-05,
+      "loss": 0.9336,
+      "step": 8271
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7407254044162538e-05,
+      "loss": 0.8809,
+      "step": 8272
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7404359934098152e-05,
+      "loss": 1.0371,
+      "step": 8273
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7401465644351732e-05,
+      "loss": 1.0029,
+      "step": 8274
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7398571174993503e-05,
+      "loss": 0.8604,
+      "step": 8275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7395676526093712e-05,
+      "loss": 0.9375,
+      "step": 8276
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7392781697722602e-05,
+      "loss": 1.0918,
+      "step": 8277
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7389886689950428e-05,
+      "loss": 0.8916,
+      "step": 8278
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7386991502847427e-05,
+      "loss": 1.1289,
+      "step": 8279
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7384096136483872e-05,
+      "loss": 0.8486,
+      "step": 8280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7381200590930013e-05,
+      "loss": 1.0977,
+      "step": 8281
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7378304866256132e-05,
+      "loss": 0.876,
+      "step": 8282
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7375408962532484e-05,
+      "loss": 0.8428,
+      "step": 8283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7372512879829357e-05,
+      "loss": 1.126,
+      "step": 8284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7369616618217015e-05,
+      "loss": 0.9629,
+      "step": 8285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.736672017776576e-05,
+      "loss": 0.6663,
+      "step": 8286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.736382355854586e-05,
+      "loss": 1.0088,
+      "step": 8287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7360926760627637e-05,
+      "loss": 0.8291,
+      "step": 8288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.735802978408135e-05,
+      "loss": 1.0898,
+      "step": 8289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7355132628977328e-05,
+      "loss": 0.8291,
+      "step": 8290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.735223529538587e-05,
+      "loss": 0.96,
+      "step": 8291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7349337783377276e-05,
+      "loss": 1.0332,
+      "step": 8292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7346440093021876e-05,
+      "loss": 0.8193,
+      "step": 8293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7343542224389973e-05,
+      "loss": 0.8477,
+      "step": 8294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7340644177551903e-05,
+      "loss": 0.9082,
+      "step": 8295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7337745952577983e-05,
+      "loss": 0.9561,
+      "step": 8296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.733484754953855e-05,
+      "loss": 0.9121,
+      "step": 8297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7331948968503933e-05,
+      "loss": 0.9307,
+      "step": 8298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7329050209544488e-05,
+      "loss": 0.8428,
+      "step": 8299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7326151272730537e-05,
+      "loss": 1.1113,
+      "step": 8300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7323252158132447e-05,
+      "loss": 1.0889,
+      "step": 8301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7320352865820557e-05,
+      "loss": 0.9297,
+      "step": 8302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.731745339586524e-05,
+      "loss": 0.9189,
+      "step": 8303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7314553748336844e-05,
+      "loss": 0.9941,
+      "step": 8304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7311653923305743e-05,
+      "loss": 0.9775,
+      "step": 8305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.73087539208423e-05,
+      "loss": 1.085,
+      "step": 8306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7305853741016898e-05,
+      "loss": 0.9219,
+      "step": 8307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.730295338389991e-05,
+      "loss": 0.917,
+      "step": 8308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.730005284956172e-05,
+      "loss": 0.999,
+      "step": 8309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7297152138072712e-05,
+      "loss": 0.9316,
+      "step": 8310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7294251249503285e-05,
+      "loss": 0.8379,
+      "step": 8311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7291350183923827e-05,
+      "loss": 1.0752,
+      "step": 8312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7288448941404753e-05,
+      "loss": 0.8193,
+      "step": 8313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.728554752201645e-05,
+      "loss": 1.0498,
+      "step": 8314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7282645925829338e-05,
+      "loss": 0.9531,
+      "step": 8315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7279744152913824e-05,
+      "loss": 1.0449,
+      "step": 8316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7276842203340333e-05,
+      "loss": 0.9512,
+      "step": 8317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7273940077179283e-05,
+      "loss": 0.8389,
+      "step": 8318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7271037774501094e-05,
+      "loss": 0.875,
+      "step": 8319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.72681352953762e-05,
+      "loss": 0.8623,
+      "step": 8320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7265232639875043e-05,
+      "loss": 0.9805,
+      "step": 8321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7262329808068054e-05,
+      "loss": 1.0918,
+      "step": 8322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.725942680002568e-05,
+      "loss": 0.999,
+      "step": 8323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7256523615818366e-05,
+      "loss": 1.0479,
+      "step": 8324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7253620255516566e-05,
+      "loss": 0.876,
+      "step": 8325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.725071671919073e-05,
+      "loss": 0.9424,
+      "step": 8326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7247813006911327e-05,
+      "loss": 0.8145,
+      "step": 8327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.724490911874881e-05,
+      "loss": 0.9707,
+      "step": 8328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7242005054773665e-05,
+      "loss": 0.8994,
+      "step": 8329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.723910081505635e-05,
+      "loss": 0.8145,
+      "step": 8330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.723619639966735e-05,
+      "loss": 0.9922,
+      "step": 8331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.723329180867714e-05,
+      "loss": 0.9541,
+      "step": 8332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7230387042156214e-05,
+      "loss": 1.0596,
+      "step": 8333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7227482100175057e-05,
+      "loss": 0.9258,
+      "step": 8334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7224576982804168e-05,
+      "loss": 0.7446,
+      "step": 8335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7221671690114037e-05,
+      "loss": 1.0459,
+      "step": 8336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7218766222175178e-05,
+      "loss": 0.8975,
+      "step": 8337
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7215860579058088e-05,
+      "loss": 0.8154,
+      "step": 8338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.721295476083328e-05,
+      "loss": 0.9287,
+      "step": 8339
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.721004876757128e-05,
+      "loss": 1.0811,
+      "step": 8340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7207142599342594e-05,
+      "loss": 0.7842,
+      "step": 8341
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7204236256217754e-05,
+      "loss": 1.0176,
+      "step": 8342
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7201329738267292e-05,
+      "loss": 1.0195,
+      "step": 8343
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.719842304556173e-05,
+      "loss": 0.9229,
+      "step": 8344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7195516178171613e-05,
+      "loss": 1.0439,
+      "step": 8345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7192609136167477e-05,
+      "loss": 1.0908,
+      "step": 8346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7189701919619874e-05,
+      "loss": 0.7949,
+      "step": 8347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7186794528599346e-05,
+      "loss": 0.6699,
+      "step": 8348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7183886963176456e-05,
+      "loss": 1.0615,
+      "step": 8349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7180979223421747e-05,
+      "loss": 1.084,
+      "step": 8350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.71780713094058e-05,
+      "loss": 0.9688,
+      "step": 8351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7175163221199167e-05,
+      "loss": 1.0215,
+      "step": 8352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7172254958872425e-05,
+      "loss": 0.8965,
+      "step": 8353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7169346522496143e-05,
+      "loss": 1.0234,
+      "step": 8354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7166437912140913e-05,
+      "loss": 0.8555,
+      "step": 8355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7163529127877307e-05,
+      "loss": 0.916,
+      "step": 8356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.716062016977592e-05,
+      "loss": 0.9268,
+      "step": 8357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7157711037907333e-05,
+      "loss": 0.8203,
+      "step": 8358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.715480173234216e-05,
+      "loss": 0.9443,
+      "step": 8359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.715189225315098e-05,
+      "loss": 0.876,
+      "step": 8360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.714898260040441e-05,
+      "loss": 0.7935,
+      "step": 8361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.714607277417306e-05,
+      "loss": 1.0342,
+      "step": 8362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.714316277452754e-05,
+      "loss": 0.8711,
+      "step": 8363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7140252601538466e-05,
+      "loss": 0.7734,
+      "step": 8364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7137342255276462e-05,
+      "loss": 0.959,
+      "step": 8365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7134431735812145e-05,
+      "loss": 0.9521,
+      "step": 8366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7131521043216156e-05,
+      "loss": 0.96,
+      "step": 8367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7128610177559123e-05,
+      "loss": 1.0088,
+      "step": 8368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7125699138911686e-05,
+      "loss": 0.7109,
+      "step": 8369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7122787927344486e-05,
+      "loss": 0.8994,
+      "step": 8370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7119876542928172e-05,
+      "loss": 0.7617,
+      "step": 8371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7116964985733393e-05,
+      "loss": 0.9639,
+      "step": 8372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7114053255830804e-05,
+      "loss": 0.9844,
+      "step": 8373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7111141353291063e-05,
+      "loss": 0.7402,
+      "step": 8374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.710822927818483e-05,
+      "loss": 0.9551,
+      "step": 8375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7105317030582785e-05,
+      "loss": 0.8271,
+      "step": 8376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.710240461055559e-05,
+      "loss": 0.9893,
+      "step": 8377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7099492018173924e-05,
+      "loss": 1.041,
+      "step": 8378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.709657925350846e-05,
+      "loss": 1.127,
+      "step": 8379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.709366631662989e-05,
+      "loss": 0.8975,
+      "step": 8380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.70907532076089e-05,
+      "loss": 0.9043,
+      "step": 8381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7087839926516183e-05,
+      "loss": 0.9873,
+      "step": 8382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.708492647342244e-05,
+      "loss": 1.0391,
+      "step": 8383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7082012848398363e-05,
+      "loss": 0.8848,
+      "step": 8384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.707909905151466e-05,
+      "loss": 0.8604,
+      "step": 8385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7076185082842044e-05,
+      "loss": 0.7676,
+      "step": 8386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7073270942451224e-05,
+      "loss": 0.9268,
+      "step": 8387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7070356630412925e-05,
+      "loss": 1.0557,
+      "step": 8388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7067442146797863e-05,
+      "loss": 0.9121,
+      "step": 8389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7064527491676763e-05,
+      "loss": 1.0361,
+      "step": 8390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7061612665120355e-05,
+      "loss": 0.9111,
+      "step": 8391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.705869766719938e-05,
+      "loss": 0.9424,
+      "step": 8392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7055782497984568e-05,
+      "loss": 1.0547,
+      "step": 8393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7052867157546664e-05,
+      "loss": 0.9316,
+      "step": 8394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7049951645956412e-05,
+      "loss": 1.0156,
+      "step": 8395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7047035963284573e-05,
+      "loss": 0.9092,
+      "step": 8396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7044120109601895e-05,
+      "loss": 1.0762,
+      "step": 8397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.704120408497914e-05,
+      "loss": 0.959,
+      "step": 8398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.703828788948706e-05,
+      "loss": 0.9893,
+      "step": 8399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7035371523196445e-05,
+      "loss": 0.9893,
+      "step": 8400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7032454986178045e-05,
+      "loss": 1.0312,
+      "step": 8401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.702953827850265e-05,
+      "loss": 0.874,
+      "step": 8402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.702662140024103e-05,
+      "loss": 0.8457,
+      "step": 8403
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7023704351463976e-05,
+      "loss": 0.7354,
+      "step": 8404
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.702078713224227e-05,
+      "loss": 1.0205,
+      "step": 8405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7017869742646707e-05,
+      "loss": 0.9365,
+      "step": 8406
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7014952182748083e-05,
+      "loss": 0.9873,
+      "step": 8407
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7012034452617207e-05,
+      "loss": 1.0195,
+      "step": 8408
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7009116552324868e-05,
+      "loss": 0.959,
+      "step": 8409
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.700619848194189e-05,
+      "loss": 0.9658,
+      "step": 8410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7003280241539073e-05,
+      "loss": 0.8838,
+      "step": 8411
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7000361831187248e-05,
+      "loss": 0.877,
+      "step": 8412
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6997443250957222e-05,
+      "loss": 0.8164,
+      "step": 8413
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6994524500919835e-05,
+      "loss": 0.9072,
+      "step": 8414
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6991605581145898e-05,
+      "loss": 0.9512,
+      "step": 8415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6988686491706264e-05,
+      "loss": 1.0225,
+      "step": 8416
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.698576723267176e-05,
+      "loss": 1.0625,
+      "step": 8417
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6982847804113226e-05,
+      "loss": 0.9004,
+      "step": 8418
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6979928206101513e-05,
+      "loss": 1.04,
+      "step": 8419
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6977008438707472e-05,
+      "loss": 1.0234,
+      "step": 8420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.697408850200195e-05,
+      "loss": 1.0137,
+      "step": 8421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6971168396055817e-05,
+      "loss": 0.7979,
+      "step": 8422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.696824812093992e-05,
+      "loss": 0.8506,
+      "step": 8423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6965327676725147e-05,
+      "loss": 0.875,
+      "step": 8424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.696240706348234e-05,
+      "loss": 1.0635,
+      "step": 8425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.69594862812824e-05,
+      "loss": 0.8857,
+      "step": 8426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.695656533019619e-05,
+      "loss": 0.8018,
+      "step": 8427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.69536442102946e-05,
+      "loss": 0.7373,
+      "step": 8428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6950722921648513e-05,
+      "loss": 1.0156,
+      "step": 8429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6947801464328827e-05,
+      "loss": 0.8823,
+      "step": 8430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6944879838406424e-05,
+      "loss": 1.0664,
+      "step": 8431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6941958043952224e-05,
+      "loss": 1.0166,
+      "step": 8432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6939036081037105e-05,
+      "loss": 0.9434,
+      "step": 8433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6936113949731996e-05,
+      "loss": 0.9395,
+      "step": 8434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6933191650107792e-05,
+      "loss": 1.0088,
+      "step": 8435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6930269182235426e-05,
+      "loss": 0.8418,
+      "step": 8436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.69273465461858e-05,
+      "loss": 0.8081,
+      "step": 8437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6924423742029846e-05,
+      "loss": 0.9941,
+      "step": 8438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6921500769838494e-05,
+      "loss": 1.0039,
+      "step": 8439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6918577629682677e-05,
+      "loss": 1.0029,
+      "step": 8440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6915654321633323e-05,
+      "loss": 0.8438,
+      "step": 8441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.691273084576138e-05,
+      "loss": 0.8086,
+      "step": 8442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6909807202137783e-05,
+      "loss": 0.875,
+      "step": 8443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6906883390833492e-05,
+      "loss": 1.0723,
+      "step": 8444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6903959411919457e-05,
+      "loss": 1.125,
+      "step": 8445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6901035265466625e-05,
+      "loss": 0.9639,
+      "step": 8446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.689811095154596e-05,
+      "loss": 1.0508,
+      "step": 8447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6895186470228434e-05,
+      "loss": 0.8887,
+      "step": 8448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.689226182158501e-05,
+      "loss": 0.8379,
+      "step": 8449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.688933700568666e-05,
+      "loss": 1.0625,
+      "step": 8450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6886412022604362e-05,
+      "loss": 0.9336,
+      "step": 8451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6883486872409098e-05,
+      "loss": 0.9697,
+      "step": 8452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6880561555171846e-05,
+      "loss": 0.8311,
+      "step": 8453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6877636070963604e-05,
+      "loss": 1.0547,
+      "step": 8454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.687471041985536e-05,
+      "loss": 0.8027,
+      "step": 8455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6871784601918115e-05,
+      "loss": 0.9756,
+      "step": 8456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6868858617222864e-05,
+      "loss": 0.9844,
+      "step": 8457
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6865932465840617e-05,
+      "loss": 0.8955,
+      "step": 8458
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6863006147842377e-05,
+      "loss": 1.001,
+      "step": 8459
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6860079663299158e-05,
+      "loss": 0.9707,
+      "step": 8460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.685715301228199e-05,
+      "loss": 0.8037,
+      "step": 8461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.685422619486188e-05,
+      "loss": 0.8877,
+      "step": 8462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.685129921110986e-05,
+      "loss": 0.8848,
+      "step": 8463
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6848372061096953e-05,
+      "loss": 0.8672,
+      "step": 8464
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6845444744894197e-05,
+      "loss": 0.9746,
+      "step": 8465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.684251726257263e-05,
+      "loss": 0.8857,
+      "step": 8466
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6839589614203295e-05,
+      "loss": 0.9883,
+      "step": 8467
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.683666179985723e-05,
+      "loss": 0.9727,
+      "step": 8468
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6833733819605492e-05,
+      "loss": 1.0029,
+      "step": 8469
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6830805673519127e-05,
+      "loss": 0.8818,
+      "step": 8470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6827877361669202e-05,
+      "loss": 0.748,
+      "step": 8471
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6824948884126772e-05,
+      "loss": 0.9219,
+      "step": 8472
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6822020240962906e-05,
+      "loss": 1.041,
+      "step": 8473
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.681909143224867e-05,
+      "loss": 0.6304,
+      "step": 8474
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.681616245805514e-05,
+      "loss": 0.8838,
+      "step": 8475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.681323331845339e-05,
+      "loss": 0.8857,
+      "step": 8476
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6810304013514506e-05,
+      "loss": 0.9873,
+      "step": 8477
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6807374543309568e-05,
+      "loss": 0.9951,
+      "step": 8478
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6804444907909673e-05,
+      "loss": 0.8271,
+      "step": 8479
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.680151510738591e-05,
+      "loss": 0.8643,
+      "step": 8480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6798585141809382e-05,
+      "loss": 1.0127,
+      "step": 8481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.679565501125118e-05,
+      "loss": 0.8701,
+      "step": 8482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6792724715782425e-05,
+      "loss": 0.8984,
+      "step": 8483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.678979425547421e-05,
+      "loss": 1.0693,
+      "step": 8484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6786863630397662e-05,
+      "loss": 0.8252,
+      "step": 8485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6783932840623886e-05,
+      "loss": 0.9023,
+      "step": 8486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.678100188622402e-05,
+      "loss": 0.9492,
+      "step": 8487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6778070767269174e-05,
+      "loss": 1.0312,
+      "step": 8488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.677513948383049e-05,
+      "loss": 0.9707,
+      "step": 8489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6772208035979086e-05,
+      "loss": 0.8135,
+      "step": 8490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.676927642378612e-05,
+      "loss": 0.6729,
+      "step": 8491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6766344647322716e-05,
+      "loss": 1.0088,
+      "step": 8492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.676341270666003e-05,
+      "loss": 0.7393,
+      "step": 8493
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6760480601869202e-05,
+      "loss": 0.9561,
+      "step": 8494
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.67575483330214e-05,
+      "loss": 1.1309,
+      "step": 8495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6754615900187765e-05,
+      "loss": 1.0703,
+      "step": 8496
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6751683303439477e-05,
+      "loss": 0.959,
+      "step": 8497
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6748750542847682e-05,
+      "loss": 1.0137,
+      "step": 8498
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6745817618483563e-05,
+      "loss": 0.9893,
+      "step": 8499
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.674288453041829e-05,
+      "loss": 0.7725,
+      "step": 8500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6739951278723036e-05,
+      "loss": 1.041,
+      "step": 8501
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.673701786346899e-05,
+      "loss": 0.7334,
+      "step": 8502
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.673408428472733e-05,
+      "loss": 0.8594,
+      "step": 8503
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6731150542569247e-05,
+      "loss": 1.0186,
+      "step": 8504
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.672821663706594e-05,
+      "loss": 0.8096,
+      "step": 8505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.67252825682886e-05,
+      "loss": 0.8984,
+      "step": 8506
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6722348336308433e-05,
+      "loss": 0.8848,
+      "step": 8507
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6719413941196637e-05,
+      "loss": 0.9365,
+      "step": 8508
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6716479383024433e-05,
+      "loss": 0.8984,
+      "step": 8509
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.671354466186302e-05,
+      "loss": 1.0723,
+      "step": 8510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6710609777783626e-05,
+      "loss": 0.8506,
+      "step": 8511
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6707674730857463e-05,
+      "loss": 0.8926,
+      "step": 8512
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6704739521155766e-05,
+      "loss": 0.9795,
+      "step": 8513
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6701804148749756e-05,
+      "loss": 0.791,
+      "step": 8514
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.669886861371067e-05,
+      "loss": 0.7793,
+      "step": 8515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.669593291610974e-05,
+      "loss": 1.0322,
+      "step": 8516
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6692997056018218e-05,
+      "loss": 1.0352,
+      "step": 8517
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.669006103350733e-05,
+      "loss": 0.8398,
+      "step": 8518
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6687124848648344e-05,
+      "loss": 0.7852,
+      "step": 8519
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6684188501512496e-05,
+      "loss": 0.9238,
+      "step": 8520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.668125199217106e-05,
+      "loss": 1.0059,
+      "step": 8521
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.667831532069528e-05,
+      "loss": 1.0205,
+      "step": 8522
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6675378487156433e-05,
+      "loss": 0.8682,
+      "step": 8523
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6672441491625777e-05,
+      "loss": 0.9238,
+      "step": 8524
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.666950433417459e-05,
+      "loss": 1.0098,
+      "step": 8525
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.666656701487415e-05,
+      "loss": 1.0088,
+      "step": 8526
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.666362953379573e-05,
+      "loss": 0.957,
+      "step": 8527
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.666069189101062e-05,
+      "loss": 0.7998,
+      "step": 8528
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.665775408659011e-05,
+      "loss": 0.8887,
+      "step": 8529
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.665481612060548e-05,
+      "loss": 0.9678,
+      "step": 8530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.665187799312804e-05,
+      "loss": 0.9512,
+      "step": 8531
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6648939704229083e-05,
+      "loss": 0.835,
+      "step": 8532
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6646001253979913e-05,
+      "loss": 0.9277,
+      "step": 8533
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6643062642451836e-05,
+      "loss": 0.9033,
+      "step": 8534
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6640123869716167e-05,
+      "loss": 0.9199,
+      "step": 8535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.663718493584422e-05,
+      "loss": 0.8809,
+      "step": 8536
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.663424584090731e-05,
+      "loss": 0.9053,
+      "step": 8537
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.663130658497677e-05,
+      "loss": 0.875,
+      "step": 8538
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6628367168123918e-05,
+      "loss": 0.9561,
+      "step": 8539
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6625427590420082e-05,
+      "loss": 1.1328,
+      "step": 8540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6622487851936613e-05,
+      "loss": 0.957,
+      "step": 8541
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6619547952744834e-05,
+      "loss": 0.8428,
+      "step": 8542
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.66166078929161e-05,
+      "loss": 0.8682,
+      "step": 8543
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6613667672521743e-05,
+      "loss": 0.8057,
+      "step": 8544
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.661072729163313e-05,
+      "loss": 0.8188,
+      "step": 8545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6607786750321596e-05,
+      "loss": 0.915,
+      "step": 8546
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6604846048658518e-05,
+      "loss": 1.0957,
+      "step": 8547
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.660190518671525e-05,
+      "loss": 1.0215,
+      "step": 8548
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6598964164563156e-05,
+      "loss": 0.957,
+      "step": 8549
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6596022982273617e-05,
+      "loss": 0.8994,
+      "step": 8550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.659308163991799e-05,
+      "loss": 0.918,
+      "step": 8551
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6590140137567662e-05,
+      "loss": 1.0117,
+      "step": 8552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6587198475294012e-05,
+      "loss": 0.9541,
+      "step": 8553
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6584256653168436e-05,
+      "loss": 0.9971,
+      "step": 8554
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6581314671262305e-05,
+      "loss": 0.8242,
+      "step": 8555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6578372529647028e-05,
+      "loss": 0.9102,
+      "step": 8556
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6575430228393996e-05,
+      "loss": 0.9189,
+      "step": 8557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6572487767574604e-05,
+      "loss": 0.9541,
+      "step": 8558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6569545147260266e-05,
+      "loss": 0.8701,
+      "step": 8559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6566602367522388e-05,
+      "loss": 0.877,
+      "step": 8560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6563659428432384e-05,
+      "loss": 0.9893,
+      "step": 8561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6560716330061667e-05,
+      "loss": 0.9834,
+      "step": 8562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6557773072481657e-05,
+      "loss": 1.0156,
+      "step": 8563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.655482965576378e-05,
+      "loss": 0.7471,
+      "step": 8564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.655188607997947e-05,
+      "loss": 1.0547,
+      "step": 8565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6548942345200147e-05,
+      "loss": 1.0439,
+      "step": 8566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6545998451497256e-05,
+      "loss": 1.0117,
+      "step": 8567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6543054398942233e-05,
+      "loss": 0.8027,
+      "step": 8568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6540110187606526e-05,
+      "loss": 0.9297,
+      "step": 8569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6537165817561576e-05,
+      "loss": 1.0752,
+      "step": 8570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6534221288878837e-05,
+      "loss": 0.7559,
+      "step": 8571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.653127660162976e-05,
+      "loss": 0.9639,
+      "step": 8572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.652833175588581e-05,
+      "loss": 0.9209,
+      "step": 8573
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.652538675171845e-05,
+      "loss": 0.8711,
+      "step": 8574
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6522441589199138e-05,
+      "loss": 0.8398,
+      "step": 8575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.651949626839936e-05,
+      "loss": 0.9395,
+      "step": 8576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6516550789390572e-05,
+      "loss": 0.8057,
+      "step": 8577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.651360515224427e-05,
+      "loss": 0.7588,
+      "step": 8578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6510659357031917e-05,
+      "loss": 0.9814,
+      "step": 8579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6507713403825013e-05,
+      "loss": 1.0898,
+      "step": 8580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.650476729269504e-05,
+      "loss": 1.0293,
+      "step": 8581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.65018210237135e-05,
+      "loss": 0.8242,
+      "step": 8582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.649887459695188e-05,
+      "loss": 1.0986,
+      "step": 8583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.649592801248169e-05,
+      "loss": 1.0107,
+      "step": 8584
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6492981270374428e-05,
+      "loss": 1.0303,
+      "step": 8585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6490034370701606e-05,
+      "loss": 0.8267,
+      "step": 8586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6487087313534735e-05,
+      "loss": 0.9697,
+      "step": 8587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6484140098945343e-05,
+      "loss": 0.9316,
+      "step": 8588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6481192727004924e-05,
+      "loss": 0.9453,
+      "step": 8589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6478245197785032e-05,
+      "loss": 0.7637,
+      "step": 8590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.647529751135717e-05,
+      "loss": 0.9707,
+      "step": 8591
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.647234966779289e-05,
+      "loss": 0.7676,
+      "step": 8592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.646940166716371e-05,
+      "loss": 0.9531,
+      "step": 8593
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6466453509541183e-05,
+      "loss": 0.791,
+      "step": 8594
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6463505194996843e-05,
+      "loss": 0.9375,
+      "step": 8595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.646055672360224e-05,
+      "loss": 0.9219,
+      "step": 8596
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6457608095428925e-05,
+      "loss": 1.0156,
+      "step": 8597
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.645465931054846e-05,
+      "loss": 1.0449,
+      "step": 8598
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6451710369032387e-05,
+      "loss": 1.0869,
+      "step": 8599
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6448761270952284e-05,
+      "loss": 0.918,
+      "step": 8600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6445812016379703e-05,
+      "loss": 0.9248,
+      "step": 8601
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.644286260538623e-05,
+      "loss": 0.8818,
+      "step": 8602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.643991303804342e-05,
+      "loss": 0.8242,
+      "step": 8603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6436963314422862e-05,
+      "loss": 0.9287,
+      "step": 8604
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6434013434596132e-05,
+      "loss": 0.9082,
+      "step": 8605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6431063398634824e-05,
+      "loss": 0.9658,
+      "step": 8606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6428113206610517e-05,
+      "loss": 0.8379,
+      "step": 8607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6425162858594813e-05,
+      "loss": 0.9053,
+      "step": 8608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6422212354659294e-05,
+      "loss": 0.9043,
+      "step": 8609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6419261694875574e-05,
+      "loss": 0.8115,
+      "step": 8610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6416310879315247e-05,
+      "loss": 0.9326,
+      "step": 8611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6413359908049926e-05,
+      "loss": 0.9297,
+      "step": 8612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.641040878115122e-05,
+      "loss": 0.8867,
+      "step": 8613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.640745749869075e-05,
+      "loss": 0.6909,
+      "step": 8614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6404506060740124e-05,
+      "loss": 0.7285,
+      "step": 8615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6401554467370977e-05,
+      "loss": 0.7783,
+      "step": 8616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6398602718654923e-05,
+      "loss": 0.9199,
+      "step": 8617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6395650814663602e-05,
+      "loss": 0.998,
+      "step": 8618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6392698755468642e-05,
+      "loss": 0.9297,
+      "step": 8619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.638974654114169e-05,
+      "loss": 0.8252,
+      "step": 8620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.638679417175437e-05,
+      "loss": 0.9424,
+      "step": 8621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6383841647378356e-05,
+      "loss": 0.9697,
+      "step": 8622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6380888968085262e-05,
+      "loss": 0.9346,
+      "step": 8623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.637793613394677e-05,
+      "loss": 0.957,
+      "step": 8624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6374983145034515e-05,
+      "loss": 0.9131,
+      "step": 8625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6372030001420172e-05,
+      "loss": 0.96,
+      "step": 8626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.63690767031754e-05,
+      "loss": 1.0391,
+      "step": 8627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.636612325037187e-05,
+      "loss": 0.9082,
+      "step": 8628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6363169643081244e-05,
+      "loss": 1.0488,
+      "step": 8629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6360215881375214e-05,
+      "loss": 1.0186,
+      "step": 8630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.635726196532544e-05,
+      "loss": 0.8691,
+      "step": 8631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6354307895003617e-05,
+      "loss": 1.0332,
+      "step": 8632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6351353670481433e-05,
+      "loss": 1.0635,
+      "step": 8633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6348399291830566e-05,
+      "loss": 1.0469,
+      "step": 8634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6345444759122725e-05,
+      "loss": 1.0439,
+      "step": 8635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6342490072429598e-05,
+      "loss": 0.9326,
+      "step": 8636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6339535231822892e-05,
+      "loss": 0.8535,
+      "step": 8637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6336580237374305e-05,
+      "loss": 0.9854,
+      "step": 8638
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6333625089155557e-05,
+      "loss": 0.9814,
+      "step": 8639
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.633066978723835e-05,
+      "loss": 1.1641,
+      "step": 8640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.632771433169441e-05,
+      "loss": 0.8125,
+      "step": 8641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6324758722595443e-05,
+      "loss": 0.9385,
+      "step": 8642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.632180296001319e-05,
+      "loss": 1.0898,
+      "step": 8643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.631884704401937e-05,
+      "loss": 0.8398,
+      "step": 8644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.631589097468571e-05,
+      "loss": 0.8125,
+      "step": 8645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.631293475208396e-05,
+      "loss": 0.7705,
+      "step": 8646
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6309978376285842e-05,
+      "loss": 0.8809,
+      "step": 8647
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.630702184736311e-05,
+      "loss": 0.8789,
+      "step": 8648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.63040651653875e-05,
+      "loss": 1.0684,
+      "step": 8649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6301108330430775e-05,
+      "loss": 0.7861,
+      "step": 8650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.629815134256468e-05,
+      "loss": 0.8691,
+      "step": 8651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6295194201860964e-05,
+      "loss": 1.0176,
+      "step": 8652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.629223690839141e-05,
+      "loss": 0.9121,
+      "step": 8653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6289279462227768e-05,
+      "loss": 0.8389,
+      "step": 8654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6286321863441812e-05,
+      "loss": 1.0205,
+      "step": 8655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6283364112105307e-05,
+      "loss": 0.9336,
+      "step": 8656
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.628040620829004e-05,
+      "loss": 0.7852,
+      "step": 8657
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6277448152067776e-05,
+      "loss": 1.0625,
+      "step": 8658
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6274489943510312e-05,
+      "loss": 0.9277,
+      "step": 8659
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.627153158268943e-05,
+      "loss": 0.7139,
+      "step": 8660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6268573069676917e-05,
+      "loss": 0.8672,
+      "step": 8661
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6265614404544572e-05,
+      "loss": 0.9463,
+      "step": 8662
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6262655587364198e-05,
+      "loss": 0.9805,
+      "step": 8663
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.625969661820758e-05,
+      "loss": 0.8193,
+      "step": 8664
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6256737497146544e-05,
+      "loss": 1.0352,
+      "step": 8665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6253778224252883e-05,
+      "loss": 0.958,
+      "step": 8666
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.625081879959842e-05,
+      "loss": 0.873,
+      "step": 8667
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6247859223254962e-05,
+      "loss": 1.0215,
+      "step": 8668
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6244899495294344e-05,
+      "loss": 0.8916,
+      "step": 8669
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6241939615788372e-05,
+      "loss": 0.8848,
+      "step": 8670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.623897958480889e-05,
+      "loss": 0.9521,
+      "step": 8671
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6236019402427718e-05,
+      "loss": 0.7861,
+      "step": 8672
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6233059068716692e-05,
+      "loss": 0.8486,
+      "step": 8673
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6230098583747653e-05,
+      "loss": 0.9004,
+      "step": 8674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6227137947592448e-05,
+      "loss": 0.7827,
+      "step": 8675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.622417716032292e-05,
+      "loss": 0.874,
+      "step": 8676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6221216222010913e-05,
+      "loss": 0.7275,
+      "step": 8677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6218255132728284e-05,
+      "loss": 1.0117,
+      "step": 8678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6215293892546887e-05,
+      "loss": 0.7744,
+      "step": 8679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6212332501538593e-05,
+      "loss": 0.8838,
+      "step": 8680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6209370959775257e-05,
+      "loss": 0.8965,
+      "step": 8681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6206409267328746e-05,
+      "loss": 0.7686,
+      "step": 8682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6203447424270935e-05,
+      "loss": 0.96,
+      "step": 8683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6200485430673702e-05,
+      "loss": 0.9131,
+      "step": 8684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6197523286608923e-05,
+      "loss": 0.9102,
+      "step": 8685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6194560992148474e-05,
+      "loss": 0.9658,
+      "step": 8686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6191598547364247e-05,
+      "loss": 0.8242,
+      "step": 8687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6188635952328132e-05,
+      "loss": 0.8408,
+      "step": 8688
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6185673207112033e-05,
+      "loss": 0.8945,
+      "step": 8689
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6182710311787824e-05,
+      "loss": 0.9658,
+      "step": 8690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.617974726642743e-05,
+      "loss": 1.0127,
+      "step": 8691
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6176784071102733e-05,
+      "loss": 0.9658,
+      "step": 8692
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6173820725885658e-05,
+      "loss": 0.8625,
+      "step": 8693
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.617085723084811e-05,
+      "loss": 0.9336,
+      "step": 8694
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6167893586062006e-05,
+      "loss": 0.9541,
+      "step": 8695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.616492979159926e-05,
+      "loss": 1.0693,
+      "step": 8696
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6161965847531804e-05,
+      "loss": 0.9185,
+      "step": 8697
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6159001753931552e-05,
+      "loss": 0.9141,
+      "step": 8698
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6156037510870446e-05,
+      "loss": 0.9697,
+      "step": 8699
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.615307311842041e-05,
+      "loss": 0.918,
+      "step": 8700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6150108576653387e-05,
+      "loss": 0.9629,
+      "step": 8701
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6147143885641316e-05,
+      "loss": 1.1582,
+      "step": 8702
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6144179045456137e-05,
+      "loss": 1.0176,
+      "step": 8703
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6141214056169805e-05,
+      "loss": 1.1025,
+      "step": 8704
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.613824891785427e-05,
+      "loss": 0.8906,
+      "step": 8705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6135283630581484e-05,
+      "loss": 0.8945,
+      "step": 8706
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6132318194423406e-05,
+      "loss": 0.8164,
+      "step": 8707
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6129352609452003e-05,
+      "loss": 1.0283,
+      "step": 8708
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.612638687573924e-05,
+      "loss": 0.8096,
+      "step": 8709
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.612342099335708e-05,
+      "loss": 0.9365,
+      "step": 8710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.61204549623775e-05,
+      "loss": 1.0039,
+      "step": 8711
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.611748878287248e-05,
+      "loss": 0.7021,
+      "step": 8712
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6114522454914e-05,
+      "loss": 0.9189,
+      "step": 8713
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6111555978574036e-05,
+      "loss": 0.9863,
+      "step": 8714
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6108589353924588e-05,
+      "loss": 0.9941,
+      "step": 8715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.610562258103764e-05,
+      "loss": 0.835,
+      "step": 8716
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6102655659985186e-05,
+      "loss": 0.8477,
+      "step": 8717
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6099688590839227e-05,
+      "loss": 0.7817,
+      "step": 8718
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6096721373671764e-05,
+      "loss": 0.9043,
+      "step": 8719
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.60937540085548e-05,
+      "loss": 0.9248,
+      "step": 8720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6090786495560352e-05,
+      "loss": 1.0107,
+      "step": 8721
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6087818834760426e-05,
+      "loss": 0.7822,
+      "step": 8722
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.608485102622704e-05,
+      "loss": 0.9219,
+      "step": 8723
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.608188307003222e-05,
+      "loss": 0.8809,
+      "step": 8724
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.607891496624798e-05,
+      "loss": 1.0312,
+      "step": 8725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.607594671494635e-05,
+      "loss": 0.9092,
+      "step": 8726
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6072978316199363e-05,
+      "loss": 0.998,
+      "step": 8727
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6070009770079053e-05,
+      "loss": 0.9814,
+      "step": 8728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6067041076657456e-05,
+      "loss": 0.9199,
+      "step": 8729
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.606407223600662e-05,
+      "loss": 0.9619,
+      "step": 8730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6061103248198575e-05,
+      "loss": 0.8418,
+      "step": 8731
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.605813411330538e-05,
+      "loss": 0.8262,
+      "step": 8732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6055164831399093e-05,
+      "loss": 0.8604,
+      "step": 8733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6052195402551762e-05,
+      "loss": 0.9404,
+      "step": 8734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.604922582683545e-05,
+      "loss": 1.1221,
+      "step": 8735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.604625610432221e-05,
+      "loss": 0.9873,
+      "step": 8736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.604328623508412e-05,
+      "loss": 0.9707,
+      "step": 8737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6040316219193244e-05,
+      "loss": 1.0039,
+      "step": 8738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6037346056721658e-05,
+      "loss": 1.043,
+      "step": 8739
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6034375747741442e-05,
+      "loss": 0.8975,
+      "step": 8740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6031405292324667e-05,
+      "loss": 0.8232,
+      "step": 8741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6028434690543427e-05,
+      "loss": 1.0098,
+      "step": 8742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6025463942469805e-05,
+      "loss": 0.9941,
+      "step": 8743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.602249304817589e-05,
+      "loss": 0.959,
+      "step": 8744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6019522007733784e-05,
+      "loss": 0.9443,
+      "step": 8745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6016550821215584e-05,
+      "loss": 0.958,
+      "step": 8746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6013579488693386e-05,
+      "loss": 0.8057,
+      "step": 8747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.60106080102393e-05,
+      "loss": 1.0244,
+      "step": 8748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6007636385925433e-05,
+      "loss": 0.917,
+      "step": 8749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6004664615823903e-05,
+      "loss": 1.0596,
+      "step": 8750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.600169270000682e-05,
+      "loss": 0.957,
+      "step": 8751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.59987206385463e-05,
+      "loss": 1.0557,
+      "step": 8752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5995748431514474e-05,
+      "loss": 0.9736,
+      "step": 8753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.599277607898347e-05,
+      "loss": 1.042,
+      "step": 8754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5989803581025416e-05,
+      "loss": 1.0225,
+      "step": 8755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.598683093771244e-05,
+      "loss": 1.041,
+      "step": 8756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5983858149116686e-05,
+      "loss": 0.8271,
+      "step": 8757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5980885215310293e-05,
+      "loss": 0.8369,
+      "step": 8758
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5977912136365403e-05,
+      "loss": 1.1172,
+      "step": 8759
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5974938912354168e-05,
+      "loss": 0.9326,
+      "step": 8760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5971965543348738e-05,
+      "loss": 0.8838,
+      "step": 8761
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5968992029421266e-05,
+      "loss": 0.9014,
+      "step": 8762
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5966018370643906e-05,
+      "loss": 1.207,
+      "step": 8763
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.596304456708883e-05,
+      "loss": 0.7837,
+      "step": 8764
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.59600706188282e-05,
+      "loss": 1.0459,
+      "step": 8765
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.595709652593419e-05,
+      "loss": 0.8525,
+      "step": 8766
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5954122288478957e-05,
+      "loss": 0.8838,
+      "step": 8767
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.595114790653469e-05,
+      "loss": 0.9932,
+      "step": 8768
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.594817338017356e-05,
+      "loss": 0.8574,
+      "step": 8769
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5945198709467758e-05,
+      "loss": 1.0801,
+      "step": 8770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.594222389448947e-05,
+      "loss": 0.9033,
+      "step": 8771
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5939248935310882e-05,
+      "loss": 0.958,
+      "step": 8772
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5936273832004188e-05,
+      "loss": 1.0596,
+      "step": 8773
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5933298584641594e-05,
+      "loss": 1.0371,
+      "step": 8774
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5930323193295278e-05,
+      "loss": 0.9365,
+      "step": 8775
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.592734765803747e-05,
+      "loss": 0.7695,
+      "step": 8776
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5924371978940364e-05,
+      "loss": 0.916,
+      "step": 8777
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.592139615607617e-05,
+      "loss": 0.877,
+      "step": 8778
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5918420189517107e-05,
+      "loss": 1.0791,
+      "step": 8779
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5915444079335396e-05,
+      "loss": 0.9004,
+      "step": 8780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5912467825603247e-05,
+      "loss": 1.0195,
+      "step": 8781
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5909491428392895e-05,
+      "loss": 0.9629,
+      "step": 8782
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5906514887776562e-05,
+      "loss": 0.9678,
+      "step": 8783
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.590353820382649e-05,
+      "loss": 0.8291,
+      "step": 8784
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.59005613766149e-05,
+      "loss": 0.8691,
+      "step": 8785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5897584406214044e-05,
+      "loss": 1.0015,
+      "step": 8786
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5894607292696155e-05,
+      "loss": 1.0312,
+      "step": 8787
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5891630036133487e-05,
+      "loss": 0.999,
+      "step": 8788
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5888652636598278e-05,
+      "loss": 0.7959,
+      "step": 8789
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5885675094162793e-05,
+      "loss": 0.834,
+      "step": 8790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5882697408899277e-05,
+      "loss": 1.0312,
+      "step": 8791
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5879719580880004e-05,
+      "loss": 0.9756,
+      "step": 8792
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5876741610177217e-05,
+      "loss": 1.0312,
+      "step": 8793
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5873763496863203e-05,
+      "loss": 0.9541,
+      "step": 8794
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.587078524101022e-05,
+      "loss": 0.9014,
+      "step": 8795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5867806842690544e-05,
+      "loss": 0.9746,
+      "step": 8796
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.586482830197645e-05,
+      "loss": 1.0146,
+      "step": 8797
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.586184961894023e-05,
+      "loss": 0.8252,
+      "step": 8798
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5858870793654153e-05,
+      "loss": 0.876,
+      "step": 8799
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5855891826190512e-05,
+      "loss": 0.918,
+      "step": 8800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5852912716621594e-05,
+      "loss": 1.0352,
+      "step": 8801
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5849933465019707e-05,
+      "loss": 1.0225,
+      "step": 8802
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.584695407145713e-05,
+      "loss": 0.9854,
+      "step": 8803
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5843974536006173e-05,
+      "loss": 0.998,
+      "step": 8804
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5840994858739148e-05,
+      "loss": 0.8926,
+      "step": 8805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5838015039728346e-05,
+      "loss": 1.0127,
+      "step": 8806
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5835035079046095e-05,
+      "loss": 0.9736,
+      "step": 8807
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.58320549767647e-05,
+      "loss": 0.9102,
+      "step": 8808
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5829074732956487e-05,
+      "loss": 0.9668,
+      "step": 8809
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5826094347693767e-05,
+      "loss": 0.9346,
+      "step": 8810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5823113821048872e-05,
+      "loss": 0.9082,
+      "step": 8811
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5820133153094135e-05,
+      "loss": 1.0674,
+      "step": 8812
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5817152343901874e-05,
+      "loss": 0.9258,
+      "step": 8813
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5814171393544438e-05,
+      "loss": 0.9766,
+      "step": 8814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.581119030209416e-05,
+      "loss": 1.0107,
+      "step": 8815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5808209069623383e-05,
+      "loss": 0.9873,
+      "step": 8816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5805227696204455e-05,
+      "loss": 1.0137,
+      "step": 8817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.580224618190973e-05,
+      "loss": 0.834,
+      "step": 8818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.579926452681154e-05,
+      "loss": 0.9717,
+      "step": 8819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5796282730982264e-05,
+      "loss": 1.0762,
+      "step": 8820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.579330079449425e-05,
+      "loss": 0.8789,
+      "step": 8821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5790318717419866e-05,
+      "loss": 0.8301,
+      "step": 8822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5787336499831474e-05,
+      "loss": 0.9365,
+      "step": 8823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.578435414180144e-05,
+      "loss": 1.0361,
+      "step": 8824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5781371643402148e-05,
+      "loss": 1.1152,
+      "step": 8825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5778389004705963e-05,
+      "loss": 0.8516,
+      "step": 8826
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5775406225785276e-05,
+      "loss": 0.8838,
+      "step": 8827
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.577242330671246e-05,
+      "loss": 0.9668,
+      "step": 8828
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5769440247559914e-05,
+      "loss": 0.876,
+      "step": 8829
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5766457048400012e-05,
+      "loss": 0.9756,
+      "step": 8830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5763473709305164e-05,
+      "loss": 0.8184,
+      "step": 8831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5760490230347755e-05,
+      "loss": 0.8887,
+      "step": 8832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.575750661160019e-05,
+      "loss": 0.9922,
+      "step": 8833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5754522853134867e-05,
+      "loss": 0.7949,
+      "step": 8834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5751538955024203e-05,
+      "loss": 1.2109,
+      "step": 8835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5748554917340602e-05,
+      "loss": 0.8511,
+      "step": 8836
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.574557074015648e-05,
+      "loss": 0.8086,
+      "step": 8837
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5742586423544248e-05,
+      "loss": 0.7812,
+      "step": 8838
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.573960196757634e-05,
+      "loss": 0.7734,
+      "step": 8839
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5736617372325167e-05,
+      "loss": 0.9619,
+      "step": 8840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5733632637863163e-05,
+      "loss": 0.9453,
+      "step": 8841
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5730647764262754e-05,
+      "loss": 0.8555,
+      "step": 8842
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.572766275159638e-05,
+      "loss": 0.8496,
+      "step": 8843
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5724677599936474e-05,
+      "loss": 0.9873,
+      "step": 8844
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.572169230935548e-05,
+      "loss": 0.8047,
+      "step": 8845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5718706879925837e-05,
+      "loss": 0.8267,
+      "step": 8846
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.571572131172e-05,
+      "loss": 0.751,
+      "step": 8847
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.571273560481041e-05,
+      "loss": 0.9404,
+      "step": 8848
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.570974975926954e-05,
+      "loss": 0.9355,
+      "step": 8849
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.570676377516982e-05,
+      "loss": 0.8662,
+      "step": 8850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.570377765258374e-05,
+      "loss": 0.8047,
+      "step": 8851
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5700791391583736e-05,
+      "loss": 0.9321,
+      "step": 8852
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5697804992242303e-05,
+      "loss": 0.9189,
+      "step": 8853
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5694818454631886e-05,
+      "loss": 1.0381,
+      "step": 8854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5691831778824984e-05,
+      "loss": 0.9883,
+      "step": 8855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.568884496489406e-05,
+      "loss": 0.8608,
+      "step": 8856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5685858012911604e-05,
+      "loss": 0.7402,
+      "step": 8857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.568287092295009e-05,
+      "loss": 0.8125,
+      "step": 8858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5679883695082014e-05,
+      "loss": 0.7744,
+      "step": 8859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5676896329379865e-05,
+      "loss": 0.8672,
+      "step": 8860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5673908825916135e-05,
+      "loss": 0.8203,
+      "step": 8861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5670921184763323e-05,
+      "loss": 1.0967,
+      "step": 8862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5667933405993936e-05,
+      "loss": 0.8955,
+      "step": 8863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.566494548968047e-05,
+      "loss": 0.9443,
+      "step": 8864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.566195743589544e-05,
+      "loss": 0.9688,
+      "step": 8865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.565896924471135e-05,
+      "loss": 0.9648,
+      "step": 8866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.565598091620072e-05,
+      "loss": 0.8398,
+      "step": 8867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.565299245043607e-05,
+      "loss": 1.0098,
+      "step": 8868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.565000384748992e-05,
+      "loss": 0.915,
+      "step": 8869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5647015107434787e-05,
+      "loss": 0.9385,
+      "step": 8870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5644026230343205e-05,
+      "loss": 0.7354,
+      "step": 8871
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.564103721628771e-05,
+      "loss": 1.0078,
+      "step": 8872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5638048065340832e-05,
+      "loss": 0.7324,
+      "step": 8873
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5635058777575097e-05,
+      "loss": 0.8945,
+      "step": 8874
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5632069353063068e-05,
+      "loss": 0.915,
+      "step": 8875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5629079791877276e-05,
+      "loss": 1.0459,
+      "step": 8876
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5626090094090278e-05,
+      "loss": 0.9111,
+      "step": 8877
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.562310025977461e-05,
+      "loss": 0.9404,
+      "step": 8878
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5620110289002844e-05,
+      "loss": 1.0234,
+      "step": 8879
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.561712018184752e-05,
+      "loss": 0.8018,
+      "step": 8880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.561412993838121e-05,
+      "loss": 0.999,
+      "step": 8881
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.561113955867648e-05,
+      "loss": 0.8311,
+      "step": 8882
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5608149042805898e-05,
+      "loss": 1.1328,
+      "step": 8883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5605158390842027e-05,
+      "loss": 1.0371,
+      "step": 8884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.560216760285745e-05,
+      "loss": 0.8335,
+      "step": 8885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5599176678924735e-05,
+      "loss": 0.8413,
+      "step": 8886
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5596185619116476e-05,
+      "loss": 0.9766,
+      "step": 8887
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.559319442350524e-05,
+      "loss": 0.9072,
+      "step": 8888
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5590203092163628e-05,
+      "loss": 0.9229,
+      "step": 8889
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.558721162516423e-05,
+      "loss": 0.9014,
+      "step": 8890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5584220022579635e-05,
+      "loss": 0.7925,
+      "step": 8891
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5581228284482448e-05,
+      "loss": 0.9883,
+      "step": 8892
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5578236410945255e-05,
+      "loss": 0.8047,
+      "step": 8893
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.557524440204068e-05,
+      "loss": 0.813,
+      "step": 8894
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.557225225784131e-05,
+      "loss": 0.8438,
+      "step": 8895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.556925997841977e-05,
+      "loss": 0.7461,
+      "step": 8896
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.556626756384867e-05,
+      "loss": 0.9707,
+      "step": 8897
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.556327501420062e-05,
+      "loss": 0.8975,
+      "step": 8898
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5560282329548255e-05,
+      "loss": 0.9268,
+      "step": 8899
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5557289509964182e-05,
+      "loss": 0.8193,
+      "step": 8900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5554296555521042e-05,
+      "loss": 0.9893,
+      "step": 8901
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5551303466291453e-05,
+      "loss": 0.9092,
+      "step": 8902
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5548310242348064e-05,
+      "loss": 0.8555,
+      "step": 8903
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5545316883763492e-05,
+      "loss": 0.8838,
+      "step": 8904
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5542323390610396e-05,
+      "loss": 1.0078,
+      "step": 8905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5539329762961408e-05,
+      "loss": 1.0186,
+      "step": 8906
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5536336000889174e-05,
+      "loss": 0.8496,
+      "step": 8907
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.553334210446635e-05,
+      "loss": 1.0625,
+      "step": 8908
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5530348073765582e-05,
+      "loss": 0.9443,
+      "step": 8909
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5527353908859532e-05,
+      "loss": 0.9814,
+      "step": 8910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5524359609820862e-05,
+      "loss": 1.0195,
+      "step": 8911
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.552136517672223e-05,
+      "loss": 0.7434,
+      "step": 8912
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5518370609636298e-05,
+      "loss": 0.9658,
+      "step": 8913
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5515375908635747e-05,
+      "loss": 0.8145,
+      "step": 8914
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.551238107379324e-05,
+      "loss": 0.916,
+      "step": 8915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5509386105181456e-05,
+      "loss": 0.7627,
+      "step": 8916
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.550639100287307e-05,
+      "loss": 0.8057,
+      "step": 8917
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.550339576694077e-05,
+      "loss": 0.9717,
+      "step": 8918
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5500400397457245e-05,
+      "loss": 0.9619,
+      "step": 8919
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5497404894495173e-05,
+      "loss": 0.8555,
+      "step": 8920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5494409258127252e-05,
+      "loss": 0.874,
+      "step": 8921
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5491413488426177e-05,
+      "loss": 0.9531,
+      "step": 8922
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5488417585464648e-05,
+      "loss": 0.6748,
+      "step": 8923
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5485421549315367e-05,
+      "loss": 0.7803,
+      "step": 8924
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.548242538005103e-05,
+      "loss": 0.9609,
+      "step": 8925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5479429077744358e-05,
+      "loss": 0.957,
+      "step": 8926
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5476432642468053e-05,
+      "loss": 0.8496,
+      "step": 8927
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.547343607429484e-05,
+      "loss": 1.1387,
+      "step": 8928
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.547043937329742e-05,
+      "loss": 0.9365,
+      "step": 8929
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5467442539548533e-05,
+      "loss": 0.8467,
+      "step": 8930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5464445573120892e-05,
+      "loss": 0.9941,
+      "step": 8931
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.546144847408723e-05,
+      "loss": 1.042,
+      "step": 8932
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5458451242520267e-05,
+      "loss": 0.8481,
+      "step": 8933
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5455453878492752e-05,
+      "loss": 0.9941,
+      "step": 8934
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.545245638207741e-05,
+      "loss": 1.041,
+      "step": 8935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5449458753346993e-05,
+      "loss": 1.0518,
+      "step": 8936
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.544646099237423e-05,
+      "loss": 1.0957,
+      "step": 8937
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5443463099231885e-05,
+      "loss": 0.8262,
+      "step": 8938
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5440465073992694e-05,
+      "loss": 0.8184,
+      "step": 8939
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.543746691672942e-05,
+      "loss": 1.0137,
+      "step": 8940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.543446862751481e-05,
+      "loss": 0.9668,
+      "step": 8941
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.543147020642163e-05,
+      "loss": 0.9189,
+      "step": 8942
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5428471653522642e-05,
+      "loss": 0.9639,
+      "step": 8943
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5425472968890613e-05,
+      "loss": 0.8223,
+      "step": 8944
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.542247415259831e-05,
+      "loss": 0.9229,
+      "step": 8945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5419475204718506e-05,
+      "loss": 0.9521,
+      "step": 8946
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5416476125323977e-05,
+      "loss": 0.8506,
+      "step": 8947
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.54134769144875e-05,
+      "loss": 0.9922,
+      "step": 8948
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.541047757228186e-05,
+      "loss": 1.0146,
+      "step": 8949
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.540747809877984e-05,
+      "loss": 1.0938,
+      "step": 8950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.540447849405423e-05,
+      "loss": 1.041,
+      "step": 8951
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5401478758177826e-05,
+      "loss": 0.9414,
+      "step": 8952
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5398478891223412e-05,
+      "loss": 0.8242,
+      "step": 8953
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5395478893263796e-05,
+      "loss": 0.8281,
+      "step": 8954
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.539247876437177e-05,
+      "loss": 1.0439,
+      "step": 8955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.538947850462014e-05,
+      "loss": 0.7607,
+      "step": 8956
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5386478114081726e-05,
+      "loss": 1.0273,
+      "step": 8957
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5383477592829326e-05,
+      "loss": 1.0674,
+      "step": 8958
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5380476940935755e-05,
+      "loss": 0.9561,
+      "step": 8959
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5377476158473833e-05,
+      "loss": 0.9609,
+      "step": 8960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5374475245516376e-05,
+      "loss": 0.9414,
+      "step": 8961
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5371474202136215e-05,
+      "loss": 0.8184,
+      "step": 8962
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.536847302840617e-05,
+      "loss": 0.7661,
+      "step": 8963
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5365471724399077e-05,
+      "loss": 1.0059,
+      "step": 8964
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5362470290187756e-05,
+      "loss": 1.0684,
+      "step": 8965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.535946872584506e-05,
+      "loss": 0.9287,
+      "step": 8966
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5356467031443815e-05,
+      "loss": 0.8711,
+      "step": 8967
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.535346520705687e-05,
+      "loss": 0.9854,
+      "step": 8968
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5350463252757064e-05,
+      "loss": 1.002,
+      "step": 8969
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5347461168617257e-05,
+      "loss": 1.0508,
+      "step": 8970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.534445895471029e-05,
+      "loss": 0.8115,
+      "step": 8971
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5341456611109017e-05,
+      "loss": 0.9746,
+      "step": 8972
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5338454137886307e-05,
+      "loss": 0.8535,
+      "step": 8973
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5335451535115017e-05,
+      "loss": 1.0469,
+      "step": 8974
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5332448802868005e-05,
+      "loss": 0.9307,
+      "step": 8975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5329445941218137e-05,
+      "loss": 0.9385,
+      "step": 8976
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5326442950238298e-05,
+      "loss": 0.8828,
+      "step": 8977
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.532343983000135e-05,
+      "loss": 0.7568,
+      "step": 8978
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.532043658058017e-05,
+      "loss": 1.0742,
+      "step": 8979
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5317433202047647e-05,
+      "loss": 0.895,
+      "step": 8980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5314429694476657e-05,
+      "loss": 1.0518,
+      "step": 8981
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.531142605794009e-05,
+      "loss": 1.0039,
+      "step": 8982
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5308422292510824e-05,
+      "loss": 0.8086,
+      "step": 8983
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.530541839826177e-05,
+      "loss": 0.8408,
+      "step": 8984
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5302414375265806e-05,
+      "loss": 0.8721,
+      "step": 8985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5299410223595845e-05,
+      "loss": 1.042,
+      "step": 8986
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5296405943324782e-05,
+      "loss": 0.8003,
+      "step": 8987
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5293401534525522e-05,
+      "loss": 0.7803,
+      "step": 8988
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.529039699727097e-05,
+      "loss": 0.752,
+      "step": 8989
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5287392331634052e-05,
+      "loss": 1.0859,
+      "step": 8990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.528438753768766e-05,
+      "loss": 0.9648,
+      "step": 8991
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5281382615504733e-05,
+      "loss": 0.9404,
+      "step": 8992
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5278377565158172e-05,
+      "loss": 0.915,
+      "step": 8993
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5275372386720917e-05,
+      "loss": 0.9111,
+      "step": 8994
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.527236708026589e-05,
+      "loss": 0.8809,
+      "step": 8995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.526936164586601e-05,
+      "loss": 0.8848,
+      "step": 8996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5266356083594233e-05,
+      "loss": 0.771,
+      "step": 8997
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.526335039352347e-05,
+      "loss": 0.8955,
+      "step": 8998
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.526034457572668e-05,
+      "loss": 1.0635,
+      "step": 8999
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.525733863027679e-05,
+      "loss": 1.04,
+      "step": 9000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5254332557246758e-05,
+      "loss": 0.7969,
+      "step": 9001
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5251326356709526e-05,
+      "loss": 0.9443,
+      "step": 9002
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.524832002873805e-05,
+      "loss": 0.9043,
+      "step": 9003
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5245313573405275e-05,
+      "loss": 1.0342,
+      "step": 9004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5242306990784173e-05,
+      "loss": 0.8857,
+      "step": 9005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.523930028094769e-05,
+      "loss": 0.7168,
+      "step": 9006
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.52362934439688e-05,
+      "loss": 0.7578,
+      "step": 9007
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.523328647992047e-05,
+      "loss": 0.9805,
+      "step": 9008
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.523027938887567e-05,
+      "loss": 0.7568,
+      "step": 9009
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5227272170907364e-05,
+      "loss": 0.9482,
+      "step": 9010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.522426482608854e-05,
+      "loss": 0.8916,
+      "step": 9011
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5221257354492166e-05,
+      "loss": 1.0332,
+      "step": 9012
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.521824975619124e-05,
+      "loss": 0.9912,
+      "step": 9013
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5215242031258735e-05,
+      "loss": 0.9248,
+      "step": 9014
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5212234179767644e-05,
+      "loss": 0.833,
+      "step": 9015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5209226201790956e-05,
+      "loss": 1.0703,
+      "step": 9016
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.520621809740167e-05,
+      "loss": 0.9395,
+      "step": 9017
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.520320986667278e-05,
+      "loss": 0.9355,
+      "step": 9018
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5200201509677294e-05,
+      "loss": 0.8135,
+      "step": 9019
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5197193026488198e-05,
+      "loss": 0.8052,
+      "step": 9020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5194184417178522e-05,
+      "loss": 0.8633,
+      "step": 9021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.519117568182126e-05,
+      "loss": 0.8984,
+      "step": 9022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.518816682048944e-05,
+      "loss": 0.7539,
+      "step": 9023
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.518515783325606e-05,
+      "loss": 1.1406,
+      "step": 9024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5182148720194156e-05,
+      "loss": 1.0586,
+      "step": 9025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5179139481376736e-05,
+      "loss": 0.8955,
+      "step": 9026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5176130116876838e-05,
+      "loss": 1.0205,
+      "step": 9027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5173120626767477e-05,
+      "loss": 0.9102,
+      "step": 9028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5170111011121697e-05,
+      "loss": 1.1426,
+      "step": 9029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5167101270012526e-05,
+      "loss": 0.834,
+      "step": 9030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5164091403513005e-05,
+      "loss": 1.0244,
+      "step": 9031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.516108141169617e-05,
+      "loss": 0.8789,
+      "step": 9032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5158071294635074e-05,
+      "loss": 0.8936,
+      "step": 9033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5155061052402746e-05,
+      "loss": 1.043,
+      "step": 9034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5152050685072258e-05,
+      "loss": 0.875,
+      "step": 9035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.514904019271664e-05,
+      "loss": 1.0527,
+      "step": 9036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.514602957540897e-05,
+      "loss": 0.8398,
+      "step": 9037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.514301883322229e-05,
+      "loss": 0.9414,
+      "step": 9038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5140007966229672e-05,
+      "loss": 0.748,
+      "step": 9039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.513699697450417e-05,
+      "loss": 0.9854,
+      "step": 9040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5133985858118866e-05,
+      "loss": 0.8701,
+      "step": 9041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.513097461714682e-05,
+      "loss": 0.792,
+      "step": 9042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5127963251661115e-05,
+      "loss": 0.875,
+      "step": 9043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5124951761734812e-05,
+      "loss": 1.084,
+      "step": 9044
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5121940147441013e-05,
+      "loss": 0.9521,
+      "step": 9045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.511892840885278e-05,
+      "loss": 0.7891,
+      "step": 9046
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5115916546043217e-05,
+      "loss": 0.6846,
+      "step": 9047
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5112904559085404e-05,
+      "loss": 0.8594,
+      "step": 9048
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.510989244805243e-05,
+      "loss": 0.8604,
+      "step": 9049
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5106880213017392e-05,
+      "loss": 0.6582,
+      "step": 9050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5103867854053395e-05,
+      "loss": 1.0312,
+      "step": 9051
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5100855371233533e-05,
+      "loss": 0.9355,
+      "step": 9052
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5097842764630916e-05,
+      "loss": 0.8066,
+      "step": 9053
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.509483003431864e-05,
+      "loss": 1.0781,
+      "step": 9054
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5091817180369827e-05,
+      "loss": 0.7197,
+      "step": 9055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5088804202857584e-05,
+      "loss": 1.0342,
+      "step": 9056
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5085791101855033e-05,
+      "loss": 0.8574,
+      "step": 9057
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.508277787743528e-05,
+      "loss": 0.8574,
+      "step": 9058
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5079764529671465e-05,
+      "loss": 0.9111,
+      "step": 9059
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.50767510586367e-05,
+      "loss": 0.8594,
+      "step": 9060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5073737464404114e-05,
+      "loss": 0.7803,
+      "step": 9061
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.507072374704684e-05,
+      "loss": 0.9072,
+      "step": 9062
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5067709906638017e-05,
+      "loss": 0.9121,
+      "step": 9063
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5064695943250778e-05,
+      "loss": 0.7151,
+      "step": 9064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5061681856958265e-05,
+      "loss": 0.8945,
+      "step": 9065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5058667647833615e-05,
+      "loss": 0.8232,
+      "step": 9066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5055653315949987e-05,
+      "loss": 0.958,
+      "step": 9067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5052638861380508e-05,
+      "loss": 0.9502,
+      "step": 9068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.504962428419835e-05,
+      "loss": 0.9209,
+      "step": 9069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5046609584476664e-05,
+      "loss": 0.9873,
+      "step": 9070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.50435947622886e-05,
+      "loss": 0.9873,
+      "step": 9071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.504057981770733e-05,
+      "loss": 0.6899,
+      "step": 9072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.503756475080601e-05,
+      "loss": 0.9297,
+      "step": 9073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5034549561657803e-05,
+      "loss": 0.8604,
+      "step": 9074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5031534250335887e-05,
+      "loss": 0.9521,
+      "step": 9075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.502851881691343e-05,
+      "loss": 0.8857,
+      "step": 9076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5025503261463615e-05,
+      "loss": 1.0137,
+      "step": 9077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5022487584059606e-05,
+      "loss": 0.9287,
+      "step": 9078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5019471784774607e-05,
+      "loss": 0.917,
+      "step": 9079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5016455863681777e-05,
+      "loss": 0.7988,
+      "step": 9080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.501343982085432e-05,
+      "loss": 0.9473,
+      "step": 9081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5010423656365424e-05,
+      "loss": 0.9365,
+      "step": 9082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5007407370288282e-05,
+      "loss": 0.9727,
+      "step": 9083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.500439096269609e-05,
+      "loss": 0.8008,
+      "step": 9084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5001374433662044e-05,
+      "loss": 0.8447,
+      "step": 9085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.499835778325935e-05,
+      "loss": 1.0391,
+      "step": 9086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.499534101156121e-05,
+      "loss": 0.792,
+      "step": 9087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.499232411864084e-05,
+      "loss": 1.1035,
+      "step": 9088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4989307104571442e-05,
+      "loss": 0.8018,
+      "step": 9089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4986289969426238e-05,
+      "loss": 0.8926,
+      "step": 9090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4983272713278434e-05,
+      "loss": 0.9053,
+      "step": 9091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4980255336201264e-05,
+      "loss": 0.918,
+      "step": 9092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.497723783826793e-05,
+      "loss": 0.9502,
+      "step": 9093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4974220219551682e-05,
+      "loss": 0.9385,
+      "step": 9094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4971202480125737e-05,
+      "loss": 0.8643,
+      "step": 9095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4968184620063326e-05,
+      "loss": 0.9277,
+      "step": 9096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4965166639437684e-05,
+      "loss": 0.917,
+      "step": 9097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4962148538322053e-05,
+      "loss": 0.8057,
+      "step": 9098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4959130316789667e-05,
+      "loss": 0.7339,
+      "step": 9099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4956111974913773e-05,
+      "loss": 0.7939,
+      "step": 9100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4953093512767618e-05,
+      "loss": 1.1328,
+      "step": 9101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.495007493042445e-05,
+      "loss": 0.8569,
+      "step": 9102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4947056227957522e-05,
+      "loss": 0.8345,
+      "step": 9103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4944037405440084e-05,
+      "loss": 0.9336,
+      "step": 9104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4941018462945396e-05,
+      "loss": 1.001,
+      "step": 9105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.493799940054673e-05,
+      "loss": 0.9316,
+      "step": 9106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4934980218317334e-05,
+      "loss": 0.9121,
+      "step": 9107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4931960916330486e-05,
+      "loss": 0.998,
+      "step": 9108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4928941494659445e-05,
+      "loss": 0.8516,
+      "step": 9109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.49259219533775e-05,
+      "loss": 0.7812,
+      "step": 9110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4922902292557903e-05,
+      "loss": 0.8994,
+      "step": 9111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4919882512273952e-05,
+      "loss": 0.876,
+      "step": 9112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4916862612598918e-05,
+      "loss": 1.0752,
+      "step": 9113
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4913842593606094e-05,
+      "loss": 0.9189,
+      "step": 9114
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4910822455368754e-05,
+      "loss": 0.9424,
+      "step": 9115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4907802197960203e-05,
+      "loss": 0.834,
+      "step": 9116
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4904781821453718e-05,
+      "loss": 0.9795,
+      "step": 9117
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4901761325922614e-05,
+      "loss": 0.8955,
+      "step": 9118
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4898740711440166e-05,
+      "loss": 0.9609,
+      "step": 9119
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4895719978079696e-05,
+      "loss": 0.7852,
+      "step": 9120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4892699125914494e-05,
+      "loss": 0.9443,
+      "step": 9121
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4889678155017878e-05,
+      "loss": 0.9453,
+      "step": 9122
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4886657065463147e-05,
+      "loss": 0.9072,
+      "step": 9123
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4883635857323628e-05,
+      "loss": 0.9688,
+      "step": 9124
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.488061453067262e-05,
+      "loss": 0.8594,
+      "step": 9125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4877593085583458e-05,
+      "loss": 1.123,
+      "step": 9126
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4874571522129457e-05,
+      "loss": 0.9531,
+      "step": 9127
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4871549840383936e-05,
+      "loss": 0.9951,
+      "step": 9128
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4868528040420227e-05,
+      "loss": 1.0107,
+      "step": 9129
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4865506122311667e-05,
+      "loss": 0.8657,
+      "step": 9130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4862484086131576e-05,
+      "loss": 1.0059,
+      "step": 9131
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4859461931953294e-05,
+      "loss": 1.0352,
+      "step": 9132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4856439659850167e-05,
+      "loss": 1.0762,
+      "step": 9133
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4853417269895534e-05,
+      "loss": 0.8115,
+      "step": 9134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4850394762162728e-05,
+      "loss": 1.0342,
+      "step": 9135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4847372136725113e-05,
+      "loss": 0.9521,
+      "step": 9136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.484434939365603e-05,
+      "loss": 0.8838,
+      "step": 9137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4841326533028838e-05,
+      "loss": 0.917,
+      "step": 9138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.483830355491688e-05,
+      "loss": 0.8066,
+      "step": 9139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.483528045939353e-05,
+      "loss": 0.873,
+      "step": 9140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.483225724653214e-05,
+      "loss": 0.9414,
+      "step": 9141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4829233916406087e-05,
+      "loss": 0.9688,
+      "step": 9142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4826210469088717e-05,
+      "loss": 0.8213,
+      "step": 9143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4823186904653422e-05,
+      "loss": 0.9619,
+      "step": 9144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.482016322317355e-05,
+      "loss": 0.9414,
+      "step": 9145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4817139424722506e-05,
+      "loss": 0.8086,
+      "step": 9146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.481411550937365e-05,
+      "loss": 0.9658,
+      "step": 9147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4811091477200374e-05,
+      "loss": 0.8159,
+      "step": 9148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.480806732827605e-05,
+      "loss": 0.9336,
+      "step": 9149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.480504306267408e-05,
+      "loss": 0.8857,
+      "step": 9150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4802018680467835e-05,
+      "loss": 0.8848,
+      "step": 9151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.479899418173073e-05,
+      "loss": 1.1348,
+      "step": 9152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4795969566536136e-05,
+      "loss": 0.9424,
+      "step": 9153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.479294483495748e-05,
+      "loss": 0.8154,
+      "step": 9154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4789919987068137e-05,
+      "loss": 0.9551,
+      "step": 9155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.478689502294153e-05,
+      "loss": 0.9668,
+      "step": 9156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4783869942651055e-05,
+      "loss": 1.0557,
+      "step": 9157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4780844746270127e-05,
+      "loss": 0.8066,
+      "step": 9158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4777819433872153e-05,
+      "loss": 0.813,
+      "step": 9159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4774794005530555e-05,
+      "loss": 1.04,
+      "step": 9160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.477176846131875e-05,
+      "loss": 0.875,
+      "step": 9161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.476874280131016e-05,
+      "loss": 0.9326,
+      "step": 9162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.47657170255782e-05,
+      "loss": 1.0342,
+      "step": 9163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4762691134196312e-05,
+      "loss": 0.9111,
+      "step": 9164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4759665127237905e-05,
+      "loss": 0.9482,
+      "step": 9165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4756639004776432e-05,
+      "loss": 0.7637,
+      "step": 9166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4753612766885313e-05,
+      "loss": 0.8145,
+      "step": 9167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4750586413637994e-05,
+      "loss": 1.1777,
+      "step": 9168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.474755994510792e-05,
+      "loss": 0.7754,
+      "step": 9169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4744533361368518e-05,
+      "loss": 0.9736,
+      "step": 9170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4741506662493257e-05,
+      "loss": 0.8125,
+      "step": 9171
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.473847984855556e-05,
+      "loss": 0.9277,
+      "step": 9172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.47354529196289e-05,
+      "loss": 0.7559,
+      "step": 9173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.473242587578672e-05,
+      "loss": 0.8604,
+      "step": 9174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.472939871710249e-05,
+      "loss": 0.9844,
+      "step": 9175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4726371443649652e-05,
+      "loss": 0.9053,
+      "step": 9176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.472334405550169e-05,
+      "loss": 1.0957,
+      "step": 9177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4720316552732046e-05,
+      "loss": 1.0527,
+      "step": 9178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4717288935414213e-05,
+      "loss": 0.8945,
+      "step": 9179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.471426120362164e-05,
+      "loss": 1.0215,
+      "step": 9180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4711233357427817e-05,
+      "loss": 1.0195,
+      "step": 9181
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4708205396906214e-05,
+      "loss": 0.9492,
+      "step": 9182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.470517732213032e-05,
+      "loss": 1.0225,
+      "step": 9183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4702149133173596e-05,
+      "loss": 1.1562,
+      "step": 9184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.469912083010955e-05,
+      "loss": 1.0381,
+      "step": 9185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.469609241301166e-05,
+      "loss": 0.9219,
+      "step": 9186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4693063881953413e-05,
+      "loss": 1.0498,
+      "step": 9187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.469003523700831e-05,
+      "loss": 0.9834,
+      "step": 9188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.468700647824985e-05,
+      "loss": 0.96,
+      "step": 9189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.468397760575152e-05,
+      "loss": 0.8496,
+      "step": 9190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.468094861958683e-05,
+      "loss": 1.0127,
+      "step": 9191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4677919519829288e-05,
+      "loss": 0.9326,
+      "step": 9192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4674890306552393e-05,
+      "loss": 0.9414,
+      "step": 9193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.467186097982966e-05,
+      "loss": 1.1738,
+      "step": 9194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.46688315397346e-05,
+      "loss": 0.9111,
+      "step": 9195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4665801986340725e-05,
+      "loss": 1.0156,
+      "step": 9196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.466277231972156e-05,
+      "loss": 1.0117,
+      "step": 9197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4659742539950627e-05,
+      "loss": 0.8057,
+      "step": 9198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4656712647101445e-05,
+      "loss": 1.0078,
+      "step": 9199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4653682641247538e-05,
+      "loss": 0.8232,
+      "step": 9200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4650652522462447e-05,
+      "loss": 0.8311,
+      "step": 9201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4647622290819687e-05,
+      "loss": 0.958,
+      "step": 9202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4644591946392805e-05,
+      "loss": 0.8779,
+      "step": 9203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.464156148925534e-05,
+      "loss": 1.0625,
+      "step": 9204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4638530919480824e-05,
+      "loss": 0.9062,
+      "step": 9205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4635500237142806e-05,
+      "loss": 0.8936,
+      "step": 9206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4632469442314832e-05,
+      "loss": 0.8633,
+      "step": 9207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.462943853507044e-05,
+      "loss": 1.0459,
+      "step": 9208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4626407515483196e-05,
+      "loss": 0.8613,
+      "step": 9209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4623376383626646e-05,
+      "loss": 0.8867,
+      "step": 9210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4620345139574352e-05,
+      "loss": 0.8398,
+      "step": 9211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4617313783399863e-05,
+      "loss": 0.9141,
+      "step": 9212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.461428231517675e-05,
+      "loss": 0.8848,
+      "step": 9213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4611250734978575e-05,
+      "loss": 0.9395,
+      "step": 9214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4608219042878906e-05,
+      "loss": 0.8501,
+      "step": 9215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4605187238951312e-05,
+      "loss": 0.8711,
+      "step": 9216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4602155323269372e-05,
+      "loss": 0.5742,
+      "step": 9217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4599123295906653e-05,
+      "loss": 1.0059,
+      "step": 9218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.459609115693674e-05,
+      "loss": 0.9414,
+      "step": 9219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.45930589064332e-05,
+      "loss": 0.9434,
+      "step": 9220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4590026544469642e-05,
+      "loss": 0.7236,
+      "step": 9221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.458699407111963e-05,
+      "loss": 0.9854,
+      "step": 9222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4583961486456768e-05,
+      "loss": 0.9199,
+      "step": 9223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.458092879055464e-05,
+      "loss": 0.8838,
+      "step": 9224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.457789598348684e-05,
+      "loss": 0.7666,
+      "step": 9225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4574863065326965e-05,
+      "loss": 0.9307,
+      "step": 9226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4571830036148625e-05,
+      "loss": 0.8389,
+      "step": 9227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4568796896025407e-05,
+      "loss": 0.8799,
+      "step": 9228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4565763645030927e-05,
+      "loss": 1.1816,
+      "step": 9229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4562730283238795e-05,
+      "loss": 0.9961,
+      "step": 9230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4559696810722614e-05,
+      "loss": 0.8584,
+      "step": 9231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4556663227556e-05,
+      "loss": 0.8701,
+      "step": 9232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.455362953381258e-05,
+      "loss": 0.9531,
+      "step": 9233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.455059572956595e-05,
+      "loss": 0.9121,
+      "step": 9234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.454756181488975e-05,
+      "loss": 1.0586,
+      "step": 9235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4544527789857596e-05,
+      "loss": 0.9492,
+      "step": 9236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4541493654543124e-05,
+      "loss": 1.0127,
+      "step": 9237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.453845940901995e-05,
+      "loss": 0.9629,
+      "step": 9238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4535425053361718e-05,
+      "loss": 0.7979,
+      "step": 9239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4532390587642055e-05,
+      "loss": 0.8457,
+      "step": 9240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4529356011934605e-05,
+      "loss": 0.772,
+      "step": 9241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4526321326313002e-05,
+      "loss": 1.0088,
+      "step": 9242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4523286530850894e-05,
+      "loss": 0.9805,
+      "step": 9243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4520251625621922e-05,
+      "loss": 0.7754,
+      "step": 9244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.451721661069974e-05,
+      "loss": 1.0029,
+      "step": 9245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4514181486157993e-05,
+      "loss": 0.9727,
+      "step": 9246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.451114625207034e-05,
+      "loss": 0.9062,
+      "step": 9247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4508110908510427e-05,
+      "loss": 1.0244,
+      "step": 9248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.450507545555193e-05,
+      "loss": 0.8828,
+      "step": 9249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.450203989326849e-05,
+      "loss": 0.6748,
+      "step": 9250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.449900422173379e-05,
+      "loss": 0.7832,
+      "step": 9251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4495968441021483e-05,
+      "loss": 0.9883,
+      "step": 9252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.449293255120525e-05,
+      "loss": 0.7158,
+      "step": 9253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4489896552358757e-05,
+      "loss": 0.9131,
+      "step": 9254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.448686044455567e-05,
+      "loss": 0.9814,
+      "step": 9255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4483824227869684e-05,
+      "loss": 0.9121,
+      "step": 9256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4480787902374464e-05,
+      "loss": 0.8936,
+      "step": 9257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4477751468143706e-05,
+      "loss": 0.8535,
+      "step": 9258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4474714925251086e-05,
+      "loss": 1.0596,
+      "step": 9259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4471678273770296e-05,
+      "loss": 0.8467,
+      "step": 9260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4468641513775024e-05,
+      "loss": 0.8828,
+      "step": 9261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4465604645338962e-05,
+      "loss": 0.8701,
+      "step": 9262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.446256766853581e-05,
+      "loss": 0.96,
+      "step": 9263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4459530583439266e-05,
+      "loss": 0.8447,
+      "step": 9264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4456493390123023e-05,
+      "loss": 0.8203,
+      "step": 9265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.44534560886608e-05,
+      "loss": 0.9424,
+      "step": 9266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.445041867912629e-05,
+      "loss": 0.9609,
+      "step": 9267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4447381161593212e-05,
+      "loss": 0.8936,
+      "step": 9268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4444343536135264e-05,
+      "loss": 0.8242,
+      "step": 9269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4441305802826172e-05,
+      "loss": 0.9912,
+      "step": 9270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4438267961739656e-05,
+      "loss": 1.0322,
+      "step": 9271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.443523001294942e-05,
+      "loss": 0.8789,
+      "step": 9272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.44321919565292e-05,
+      "loss": 0.9229,
+      "step": 9273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4429153792552714e-05,
+      "loss": 0.8843,
+      "step": 9274
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.442611552109369e-05,
+      "loss": 0.8564,
+      "step": 9275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4423077142225865e-05,
+      "loss": 1.1055,
+      "step": 9276
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.442003865602296e-05,
+      "loss": 1.0488,
+      "step": 9277
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.441700006255872e-05,
+      "loss": 1.0771,
+      "step": 9278
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4413961361906875e-05,
+      "loss": 0.917,
+      "step": 9279
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.441092255414117e-05,
+      "loss": 1.0752,
+      "step": 9280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4407883639335346e-05,
+      "loss": 0.9854,
+      "step": 9281
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4404844617563147e-05,
+      "loss": 0.9648,
+      "step": 9282
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4401805488898328e-05,
+      "loss": 0.9131,
+      "step": 9283
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.439876625341463e-05,
+      "loss": 0.8525,
+      "step": 9284
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4395726911185812e-05,
+      "loss": 1.0127,
+      "step": 9285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4392687462285638e-05,
+      "loss": 0.9893,
+      "step": 9286
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.438964790678785e-05,
+      "loss": 0.9229,
+      "step": 9287
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.438660824476622e-05,
+      "loss": 0.9131,
+      "step": 9288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.43835684762945e-05,
+      "loss": 0.9072,
+      "step": 9289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4380528601446476e-05,
+      "loss": 0.8564,
+      "step": 9290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4377488620295897e-05,
+      "loss": 0.9551,
+      "step": 9291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4374448532916552e-05,
+      "loss": 0.9365,
+      "step": 9292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4371408339382196e-05,
+      "loss": 0.9678,
+      "step": 9293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4368368039766623e-05,
+      "loss": 0.8125,
+      "step": 9294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4365327634143602e-05,
+      "loss": 1.0371,
+      "step": 9295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4362287122586918e-05,
+      "loss": 0.8433,
+      "step": 9296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.435924650517035e-05,
+      "loss": 0.9307,
+      "step": 9297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4356205781967694e-05,
+      "loss": 0.8818,
+      "step": 9298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.435316495305273e-05,
+      "loss": 0.8877,
+      "step": 9299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4350124018499262e-05,
+      "loss": 0.7007,
+      "step": 9300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.434708297838107e-05,
+      "loss": 0.9648,
+      "step": 9301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4344041832771967e-05,
+      "loss": 0.999,
+      "step": 9302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4341000581745738e-05,
+      "loss": 0.9863,
+      "step": 9303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4337959225376193e-05,
+      "loss": 0.9395,
+      "step": 9304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4334917763737133e-05,
+      "loss": 0.8154,
+      "step": 9305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4331876196902367e-05,
+      "loss": 0.9863,
+      "step": 9306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4328834524945707e-05,
+      "loss": 0.9268,
+      "step": 9307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4325792747940967e-05,
+      "loss": 1.0947,
+      "step": 9308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4322750865961952e-05,
+      "loss": 0.9814,
+      "step": 9309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.431970887908249e-05,
+      "loss": 0.9707,
+      "step": 9310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.43166667873764e-05,
+      "loss": 0.9658,
+      "step": 9311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.43136245909175e-05,
+      "loss": 0.8223,
+      "step": 9312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.431058228977961e-05,
+      "loss": 1.0898,
+      "step": 9313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.430753988403657e-05,
+      "loss": 0.8857,
+      "step": 9314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4304497373762205e-05,
+      "loss": 0.874,
+      "step": 9315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4301454759030353e-05,
+      "loss": 0.8267,
+      "step": 9316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4298412039914836e-05,
+      "loss": 1.0342,
+      "step": 9317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.429536921648951e-05,
+      "loss": 0.8711,
+      "step": 9318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4292326288828195e-05,
+      "loss": 0.9199,
+      "step": 9319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4289283257004746e-05,
+      "loss": 0.8911,
+      "step": 9320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.428624012109301e-05,
+      "loss": 0.9111,
+      "step": 9321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4283196881166834e-05,
+      "loss": 0.9229,
+      "step": 9322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4280153537300063e-05,
+      "loss": 0.957,
+      "step": 9323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4277110089566557e-05,
+      "loss": 0.7842,
+      "step": 9324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.427406653804016e-05,
+      "loss": 1.0186,
+      "step": 9325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4271022882794738e-05,
+      "loss": 0.8311,
+      "step": 9326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4267979123904154e-05,
+      "loss": 1.0312,
+      "step": 9327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4264935261442273e-05,
+      "loss": 0.9971,
+      "step": 9328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.426189129548295e-05,
+      "loss": 0.9697,
+      "step": 9329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.425884722610006e-05,
+      "loss": 1.0225,
+      "step": 9330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4255803053367474e-05,
+      "loss": 1.0439,
+      "step": 9331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4252758777359064e-05,
+      "loss": 0.9385,
+      "step": 9332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.424971439814871e-05,
+      "loss": 0.832,
+      "step": 9333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.424666991581028e-05,
+      "loss": 0.8066,
+      "step": 9334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4243625330417654e-05,
+      "loss": 1.0342,
+      "step": 9335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.424058064204473e-05,
+      "loss": 0.9619,
+      "step": 9336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.423753585076538e-05,
+      "loss": 0.7588,
+      "step": 9337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4234490956653504e-05,
+      "loss": 0.7969,
+      "step": 9338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4231445959782976e-05,
+      "loss": 0.9619,
+      "step": 9339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4228400860227708e-05,
+      "loss": 0.915,
+      "step": 9340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4225355658061584e-05,
+      "loss": 0.9424,
+      "step": 9341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.42223103533585e-05,
+      "loss": 0.9668,
+      "step": 9342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4219264946192365e-05,
+      "loss": 0.8457,
+      "step": 9343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4216219436637074e-05,
+      "loss": 0.9385,
+      "step": 9344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.421317382476654e-05,
+      "loss": 1.0371,
+      "step": 9345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.421012811065467e-05,
+      "loss": 0.8818,
+      "step": 9346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.420708229437537e-05,
+      "loss": 0.9043,
+      "step": 9347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4204036376002556e-05,
+      "loss": 0.9414,
+      "step": 9348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4200990355610144e-05,
+      "loss": 1.0996,
+      "step": 9349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4197944233272045e-05,
+      "loss": 0.6553,
+      "step": 9350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4194898009062194e-05,
+      "loss": 0.915,
+      "step": 9351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4191851683054498e-05,
+      "loss": 0.9316,
+      "step": 9352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4188805255322895e-05,
+      "loss": 1.0557,
+      "step": 9353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4185758725941303e-05,
+      "loss": 0.8545,
+      "step": 9354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4182712094983653e-05,
+      "loss": 0.8457,
+      "step": 9355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4179665362523887e-05,
+      "loss": 0.8633,
+      "step": 9356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.417661852863593e-05,
+      "loss": 0.8447,
+      "step": 9357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.417357159339373e-05,
+      "loss": 0.8096,
+      "step": 9358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.417052455687122e-05,
+      "loss": 0.8037,
+      "step": 9359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4167477419142344e-05,
+      "loss": 0.9814,
+      "step": 9360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4164430180281046e-05,
+      "loss": 0.8223,
+      "step": 9361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4161382840361275e-05,
+      "loss": 1.1025,
+      "step": 9362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.415833539945698e-05,
+      "loss": 0.9229,
+      "step": 9363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4155287857642113e-05,
+      "loss": 0.8125,
+      "step": 9364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4152240214990635e-05,
+      "loss": 0.8398,
+      "step": 9365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.414919247157649e-05,
+      "loss": 1.0781,
+      "step": 9366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.414614462747365e-05,
+      "loss": 0.8799,
+      "step": 9367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.414309668275608e-05,
+      "loss": 0.8389,
+      "step": 9368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4140048637497732e-05,
+      "loss": 0.9238,
+      "step": 9369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4137000491772577e-05,
+      "loss": 0.8701,
+      "step": 9370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4133952245654593e-05,
+      "loss": 0.918,
+      "step": 9371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4130903899217737e-05,
+      "loss": 0.8418,
+      "step": 9372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4127855452536002e-05,
+      "loss": 1.0547,
+      "step": 9373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4124806905683348e-05,
+      "loss": 0.9102,
+      "step": 9374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.412175825873376e-05,
+      "loss": 0.8594,
+      "step": 9375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.411870951176122e-05,
+      "loss": 0.8691,
+      "step": 9376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.411566066483972e-05,
+      "loss": 1.0068,
+      "step": 9377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.411261171804323e-05,
+      "loss": 1.0518,
+      "step": 9378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4109562671445758e-05,
+      "loss": 0.8516,
+      "step": 9379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4106513525121277e-05,
+      "loss": 0.8652,
+      "step": 9380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.41034642791438e-05,
+      "loss": 1.0498,
+      "step": 9381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4100414933587303e-05,
+      "loss": 0.9014,
+      "step": 9382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.40973654885258e-05,
+      "loss": 1.0156,
+      "step": 9383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4094315944033282e-05,
+      "loss": 0.9707,
+      "step": 9384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.409126630018376e-05,
+      "loss": 0.874,
+      "step": 9385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.408821655705124e-05,
+      "loss": 1.0625,
+      "step": 9386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.408516671470972e-05,
+      "loss": 0.8936,
+      "step": 9387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.408211677323322e-05,
+      "loss": 0.9805,
+      "step": 9388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4079066732695756e-05,
+      "loss": 0.8672,
+      "step": 9389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4076016593171334e-05,
+      "loss": 0.8789,
+      "step": 9390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4072966354733978e-05,
+      "loss": 0.791,
+      "step": 9391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4069916017457703e-05,
+      "loss": 0.9502,
+      "step": 9392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.406686558141654e-05,
+      "loss": 0.77,
+      "step": 9393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4063815046684508e-05,
+      "loss": 0.8682,
+      "step": 9394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4060764413335632e-05,
+      "loss": 1.0996,
+      "step": 9395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.405771368144395e-05,
+      "loss": 1.0908,
+      "step": 9396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.405466285108349e-05,
+      "loss": 0.9482,
+      "step": 9397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4051611922328286e-05,
+      "loss": 1.0176,
+      "step": 9398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.404856089525238e-05,
+      "loss": 1.1748,
+      "step": 9399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4045509769929805e-05,
+      "loss": 0.8604,
+      "step": 9400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4042458546434606e-05,
+      "loss": 0.8828,
+      "step": 9401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4039407224840825e-05,
+      "loss": 0.9893,
+      "step": 9402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4036355805222513e-05,
+      "loss": 0.9307,
+      "step": 9403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.403330428765371e-05,
+      "loss": 1.082,
+      "step": 9404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.403025267220848e-05,
+      "loss": 0.9746,
+      "step": 9405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4027200958960863e-05,
+      "loss": 0.9297,
+      "step": 9406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.402414914798493e-05,
+      "loss": 0.9912,
+      "step": 9407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.402109723935473e-05,
+      "loss": 0.8975,
+      "step": 9408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4018045233144324e-05,
+      "loss": 1.0752,
+      "step": 9409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4014993129427777e-05,
+      "loss": 1.0254,
+      "step": 9410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4011940928279157e-05,
+      "loss": 0.9365,
+      "step": 9411
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4008888629772524e-05,
+      "loss": 0.9971,
+      "step": 9412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4005836233981957e-05,
+      "loss": 0.917,
+      "step": 9413
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4002783740981522e-05,
+      "loss": 0.9238,
+      "step": 9414
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3999731150845304e-05,
+      "loss": 0.9365,
+      "step": 9415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3996678463647367e-05,
+      "loss": 0.8535,
+      "step": 9416
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3993625679461802e-05,
+      "loss": 0.9414,
+      "step": 9417
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3990572798362682e-05,
+      "loss": 1.0059,
+      "step": 9418
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.39875198204241e-05,
+      "loss": 0.8135,
+      "step": 9419
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3984466745720137e-05,
+      "loss": 0.9922,
+      "step": 9420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.398141357432489e-05,
+      "loss": 0.8535,
+      "step": 9421
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3978360306312435e-05,
+      "loss": 0.8486,
+      "step": 9422
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3975306941756883e-05,
+      "loss": 0.8721,
+      "step": 9423
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3972253480732315e-05,
+      "loss": 0.9541,
+      "step": 9424
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.396919992331284e-05,
+      "loss": 0.9424,
+      "step": 9425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3966146269572562e-05,
+      "loss": 0.8564,
+      "step": 9426
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.396309251958557e-05,
+      "loss": 0.9258,
+      "step": 9427
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3960038673425984e-05,
+      "loss": 0.9893,
+      "step": 9428
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3956984731167897e-05,
+      "loss": 0.9375,
+      "step": 9429
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3953930692885433e-05,
+      "loss": 0.7852,
+      "step": 9430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3950876558652702e-05,
+      "loss": 0.9375,
+      "step": 9431
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.394782232854381e-05,
+      "loss": 0.9062,
+      "step": 9432
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3944768002632884e-05,
+      "loss": 1.082,
+      "step": 9433
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.394171358099404e-05,
+      "loss": 0.9697,
+      "step": 9434
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3938659063701394e-05,
+      "loss": 0.9863,
+      "step": 9435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3935604450829084e-05,
+      "loss": 1.0059,
+      "step": 9436
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3932549742451225e-05,
+      "loss": 0.918,
+      "step": 9437
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3929494938641944e-05,
+      "loss": 0.9229,
+      "step": 9438
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3926440039475382e-05,
+      "loss": 0.8037,
+      "step": 9439
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3923385045025667e-05,
+      "loss": 0.9326,
+      "step": 9440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3920329955366938e-05,
+      "loss": 0.9365,
+      "step": 9441
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3917274770573324e-05,
+      "loss": 0.9268,
+      "step": 9442
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3914219490718974e-05,
+      "loss": 0.873,
+      "step": 9443
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3911164115878028e-05,
+      "loss": 1.0752,
+      "step": 9444
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3908108646124634e-05,
+      "loss": 1.0732,
+      "step": 9445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3905053081532932e-05,
+      "loss": 0.8447,
+      "step": 9446
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.390199742217708e-05,
+      "loss": 0.8574,
+      "step": 9447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3898941668131225e-05,
+      "loss": 0.8438,
+      "step": 9448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.389588581946952e-05,
+      "loss": 0.7065,
+      "step": 9449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.389282987626613e-05,
+      "loss": 0.7578,
+      "step": 9450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.38897738385952e-05,
+      "loss": 0.9463,
+      "step": 9451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3886717706530906e-05,
+      "loss": 0.959,
+      "step": 9452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3883661480147392e-05,
+      "loss": 1.0156,
+      "step": 9453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.388060515951885e-05,
+      "loss": 0.7783,
+      "step": 9454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.387754874471942e-05,
+      "loss": 0.9814,
+      "step": 9455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3874492235823295e-05,
+      "loss": 0.8047,
+      "step": 9456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.387143563290463e-05,
+      "loss": 0.9053,
+      "step": 9457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3868378936037614e-05,
+      "loss": 0.8164,
+      "step": 9458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3865322145296413e-05,
+      "loss": 1.0127,
+      "step": 9459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3862265260755215e-05,
+      "loss": 0.7988,
+      "step": 9460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3859208282488194e-05,
+      "loss": 0.8877,
+      "step": 9461
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.385615121056954e-05,
+      "loss": 0.8379,
+      "step": 9462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3853094045073433e-05,
+      "loss": 0.9883,
+      "step": 9463
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3850036786074072e-05,
+      "loss": 0.9326,
+      "step": 9464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3846979433645636e-05,
+      "loss": 0.8809,
+      "step": 9465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3843921987862324e-05,
+      "loss": 0.9902,
+      "step": 9466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3840864448798326e-05,
+      "loss": 0.7539,
+      "step": 9467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.383780681652785e-05,
+      "loss": 0.9834,
+      "step": 9468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3834749091125083e-05,
+      "loss": 0.8467,
+      "step": 9469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.383169127266424e-05,
+      "loss": 0.7441,
+      "step": 9470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3828633361219512e-05,
+      "loss": 0.8525,
+      "step": 9471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3825575356865117e-05,
+      "loss": 0.9434,
+      "step": 9472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3822517259675254e-05,
+      "loss": 0.9082,
+      "step": 9473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3819459069724143e-05,
+      "loss": 1.0186,
+      "step": 9474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.381640078708599e-05,
+      "loss": 1.0537,
+      "step": 9475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3813342411835018e-05,
+      "loss": 0.874,
+      "step": 9476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.381028394404543e-05,
+      "loss": 0.7109,
+      "step": 9477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.380722538379147e-05,
+      "loss": 1.0654,
+      "step": 9478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.380416673114733e-05,
+      "loss": 0.8311,
+      "step": 9479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.380110798618726e-05,
+      "loss": 1.0732,
+      "step": 9480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3798049148985478e-05,
+      "loss": 1.0391,
+      "step": 9481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3794990219616217e-05,
+      "loss": 0.9258,
+      "step": 9482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3791931198153698e-05,
+      "loss": 0.8389,
+      "step": 9483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3788872084672163e-05,
+      "loss": 0.8398,
+      "step": 9484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3785812879245845e-05,
+      "loss": 0.8838,
+      "step": 9485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3782753581948982e-05,
+      "loss": 0.9219,
+      "step": 9486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3779694192855815e-05,
+      "loss": 0.9932,
+      "step": 9487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3776634712040586e-05,
+      "loss": 0.9463,
+      "step": 9488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.377357513957754e-05,
+      "loss": 1.0039,
+      "step": 9489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.377051547554092e-05,
+      "loss": 1.0137,
+      "step": 9490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3767455720004974e-05,
+      "loss": 0.9092,
+      "step": 9491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.376439587304397e-05,
+      "loss": 0.9238,
+      "step": 9492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3761335934732138e-05,
+      "loss": 1.0664,
+      "step": 9493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3758275905143752e-05,
+      "loss": 0.873,
+      "step": 9494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3755215784353055e-05,
+      "loss": 0.8936,
+      "step": 9495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3752155572434318e-05,
+      "loss": 0.9365,
+      "step": 9496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.37490952694618e-05,
+      "loss": 0.8848,
+      "step": 9497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3746034875509767e-05,
+      "loss": 0.8584,
+      "step": 9498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.374297439065248e-05,
+      "loss": 0.7695,
+      "step": 9499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.373991381496422e-05,
+      "loss": 0.79,
+      "step": 9500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3736853148519243e-05,
+      "loss": 0.8359,
+      "step": 9501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3733792391391832e-05,
+      "loss": 0.998,
+      "step": 9502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.373073154365626e-05,
+      "loss": 0.917,
+      "step": 9503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.372767060538681e-05,
+      "loss": 0.7881,
+      "step": 9504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.372460957665775e-05,
+      "loss": 1.0205,
+      "step": 9505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3721548457543374e-05,
+      "loss": 0.9102,
+      "step": 9506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3718487248117957e-05,
+      "loss": 0.9775,
+      "step": 9507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3715425948455793e-05,
+      "loss": 0.7559,
+      "step": 9508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.371236455863117e-05,
+      "loss": 0.8838,
+      "step": 9509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3709303078718372e-05,
+      "loss": 0.9316,
+      "step": 9510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.37062415087917e-05,
+      "loss": 0.8545,
+      "step": 9511
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3703179848925445e-05,
+      "loss": 0.7612,
+      "step": 9512
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3700118099193907e-05,
+      "loss": 0.9014,
+      "step": 9513
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3697056259671384e-05,
+      "loss": 0.9502,
+      "step": 9514
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3693994330432178e-05,
+      "loss": 0.9668,
+      "step": 9515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.369093231155059e-05,
+      "loss": 1.0654,
+      "step": 9516
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.368787020310094e-05,
+      "loss": 0.9102,
+      "step": 9517
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3684808005157515e-05,
+      "loss": 0.9102,
+      "step": 9518
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3681745717794643e-05,
+      "loss": 0.9951,
+      "step": 9519
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3678683341086623e-05,
+      "loss": 1.002,
+      "step": 9520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.367562087510778e-05,
+      "loss": 0.8203,
+      "step": 9521
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.367255831993243e-05,
+      "loss": 1.0107,
+      "step": 9522
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.366949567563489e-05,
+      "loss": 1.0449,
+      "step": 9523
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3666432942289478e-05,
+      "loss": 0.8262,
+      "step": 9524
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3663370119970527e-05,
+      "loss": 0.9912,
+      "step": 9525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3660307208752354e-05,
+      "loss": 0.812,
+      "step": 9526
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3657244208709286e-05,
+      "loss": 0.8604,
+      "step": 9527
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3654181119915663e-05,
+      "loss": 1.0713,
+      "step": 9528
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3651117942445806e-05,
+      "loss": 0.9473,
+      "step": 9529
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3648054676374056e-05,
+      "loss": 0.8389,
+      "step": 9530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.364499132177475e-05,
+      "loss": 0.7915,
+      "step": 9531
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3641927878722214e-05,
+      "loss": 0.6543,
+      "step": 9532
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.363886434729081e-05,
+      "loss": 0.895,
+      "step": 9533
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3635800727554866e-05,
+      "loss": 0.6694,
+      "step": 9534
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.363273701958873e-05,
+      "loss": 0.9004,
+      "step": 9535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.362967322346675e-05,
+      "loss": 0.9551,
+      "step": 9536
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.362660933926328e-05,
+      "loss": 1.0859,
+      "step": 9537
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.362354536705266e-05,
+      "loss": 0.9619,
+      "step": 9538
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3620481306909256e-05,
+      "loss": 0.9121,
+      "step": 9539
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3617417158907415e-05,
+      "loss": 0.8369,
+      "step": 9540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.36143529231215e-05,
+      "loss": 0.9551,
+      "step": 9541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3611288599625864e-05,
+      "loss": 0.9287,
+      "step": 9542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3608224188494878e-05,
+      "loss": 1.0176,
+      "step": 9543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3605159689802904e-05,
+      "loss": 0.9355,
+      "step": 9544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3602095103624305e-05,
+      "loss": 0.8691,
+      "step": 9545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.359903043003345e-05,
+      "loss": 0.9629,
+      "step": 9546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3595965669104715e-05,
+      "loss": 0.9561,
+      "step": 9547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.359290082091246e-05,
+      "loss": 1.0361,
+      "step": 9548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.358983588553108e-05,
+      "loss": 0.6787,
+      "step": 9549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.358677086303493e-05,
+      "loss": 0.8857,
+      "step": 9550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.358370575349841e-05,
+      "loss": 0.9131,
+      "step": 9551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3580640556995878e-05,
+      "loss": 0.8975,
+      "step": 9552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.357757527360174e-05,
+      "loss": 1.0684,
+      "step": 9553
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3574509903390363e-05,
+      "loss": 0.96,
+      "step": 9554
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3571444446436155e-05,
+      "loss": 1.0645,
+      "step": 9555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.356837890281348e-05,
+      "loss": 1.1602,
+      "step": 9556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3565313272596756e-05,
+      "loss": 0.9043,
+      "step": 9557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3562247555860355e-05,
+      "loss": 0.9717,
+      "step": 9558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3559181752678682e-05,
+      "loss": 0.8076,
+      "step": 9559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.355611586312614e-05,
+      "loss": 0.9707,
+      "step": 9560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3553049887277123e-05,
+      "loss": 0.7783,
+      "step": 9561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3549983825206026e-05,
+      "loss": 0.8115,
+      "step": 9562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3546917676987268e-05,
+      "loss": 0.8877,
+      "step": 9563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.354385144269525e-05,
+      "loss": 0.9111,
+      "step": 9564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.354078512240438e-05,
+      "loss": 0.9053,
+      "step": 9565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3537718716189067e-05,
+      "loss": 1.0078,
+      "step": 9566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3534652224123727e-05,
+      "loss": 0.9043,
+      "step": 9567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.353158564628277e-05,
+      "loss": 1.0664,
+      "step": 9568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3528518982740618e-05,
+      "loss": 0.7256,
+      "step": 9569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3525452233571686e-05,
+      "loss": 0.8408,
+      "step": 9570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.35223853988504e-05,
+      "loss": 0.8643,
+      "step": 9571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.351931847865117e-05,
+      "loss": 0.9873,
+      "step": 9572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3516251473048436e-05,
+      "loss": 0.7861,
+      "step": 9573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3513184382116617e-05,
+      "loss": 0.9434,
+      "step": 9574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.351011720593015e-05,
+      "loss": 0.9717,
+      "step": 9575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.350704994456346e-05,
+      "loss": 0.96,
+      "step": 9576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3503982598090982e-05,
+      "loss": 0.8945,
+      "step": 9577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3500915166587147e-05,
+      "loss": 0.8818,
+      "step": 9578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3497847650126402e-05,
+      "loss": 1.083,
+      "step": 9579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3494780048783175e-05,
+      "loss": 0.9531,
+      "step": 9580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.349171236263192e-05,
+      "loss": 0.9961,
+      "step": 9581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.348864459174707e-05,
+      "loss": 0.8291,
+      "step": 9582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3485576736203082e-05,
+      "loss": 1.0684,
+      "step": 9583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3482508796074387e-05,
+      "loss": 1.0107,
+      "step": 9584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3479440771435457e-05,
+      "loss": 0.6812,
+      "step": 9585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.347637266236072e-05,
+      "loss": 0.9326,
+      "step": 9586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3473304468924655e-05,
+      "loss": 1.0869,
+      "step": 9587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3470236191201695e-05,
+      "loss": 0.8584,
+      "step": 9588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3467167829266314e-05,
+      "loss": 0.8926,
+      "step": 9589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.346409938319296e-05,
+      "loss": 0.9678,
+      "step": 9590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.346103085305611e-05,
+      "loss": 0.9014,
+      "step": 9591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3457962238930218e-05,
+      "loss": 0.8809,
+      "step": 9592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.345489354088975e-05,
+      "loss": 0.8115,
+      "step": 9593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3451824759009175e-05,
+      "loss": 0.8174,
+      "step": 9594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3448755893362964e-05,
+      "loss": 1.0391,
+      "step": 9595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.344568694402559e-05,
+      "loss": 0.9736,
+      "step": 9596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.344261791107153e-05,
+      "loss": 1.0225,
+      "step": 9597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3439548794575254e-05,
+      "loss": 0.8809,
+      "step": 9598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3436479594611245e-05,
+      "loss": 0.8945,
+      "step": 9599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3433410311253985e-05,
+      "loss": 1.0186,
+      "step": 9600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.343034094457795e-05,
+      "loss": 1.001,
+      "step": 9601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3427271494657635e-05,
+      "loss": 0.9893,
+      "step": 9602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3424201961567517e-05,
+      "loss": 1.0693,
+      "step": 9603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3421132345382085e-05,
+      "loss": 0.9385,
+      "step": 9604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3418062646175836e-05,
+      "loss": 0.916,
+      "step": 9605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3414992864023253e-05,
+      "loss": 0.9336,
+      "step": 9606
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3411922998998842e-05,
+      "loss": 0.8926,
+      "step": 9607
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.340885305117709e-05,
+      "loss": 0.9893,
+      "step": 9608
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3405783020632505e-05,
+      "loss": 0.7861,
+      "step": 9609
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3402712907439574e-05,
+      "loss": 0.8906,
+      "step": 9610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.339964271167282e-05,
+      "loss": 0.8438,
+      "step": 9611
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3396572433406724e-05,
+      "loss": 1.0371,
+      "step": 9612
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.339350207271581e-05,
+      "loss": 0.8906,
+      "step": 9613
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3390431629674576e-05,
+      "loss": 0.918,
+      "step": 9614
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.338736110435754e-05,
+      "loss": 0.8682,
+      "step": 9615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.338429049683921e-05,
+      "loss": 0.9404,
+      "step": 9616
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3381219807194105e-05,
+      "loss": 0.9385,
+      "step": 9617
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.337814903549674e-05,
+      "loss": 1.0234,
+      "step": 9618
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.337507818182163e-05,
+      "loss": 0.8018,
+      "step": 9619
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.33720072462433e-05,
+      "loss": 0.8574,
+      "step": 9620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.336893622883627e-05,
+      "loss": 0.8828,
+      "step": 9621
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3365865129675066e-05,
+      "loss": 0.8779,
+      "step": 9622
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3362793948834215e-05,
+      "loss": 0.7305,
+      "step": 9623
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3359722686388242e-05,
+      "loss": 0.9922,
+      "step": 9624
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3356651342411682e-05,
+      "loss": 0.9629,
+      "step": 9625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3353579916979063e-05,
+      "loss": 0.96,
+      "step": 9626
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3350508410164925e-05,
+      "loss": 0.8857,
+      "step": 9627
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.33474368220438e-05,
+      "loss": 0.7549,
+      "step": 9628
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3344365152690225e-05,
+      "loss": 0.7803,
+      "step": 9629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.334129340217875e-05,
+      "loss": 1.0029,
+      "step": 9630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.33382215705839e-05,
+      "loss": 0.9365,
+      "step": 9631
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3335149657980242e-05,
+      "loss": 0.9229,
+      "step": 9632
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3332077664442304e-05,
+      "loss": 0.7832,
+      "step": 9633
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3329005590044642e-05,
+      "loss": 0.9238,
+      "step": 9634
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.33259334348618e-05,
+      "loss": 1.1025,
+      "step": 9635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3322861198968337e-05,
+      "loss": 0.9902,
+      "step": 9636
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3319788882438806e-05,
+      "loss": 1.0049,
+      "step": 9637
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3316716485347765e-05,
+      "loss": 1.0283,
+      "step": 9638
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3313644007769766e-05,
+      "loss": 0.9482,
+      "step": 9639
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3310571449779375e-05,
+      "loss": 0.9014,
+      "step": 9640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3307498811451144e-05,
+      "loss": 0.8379,
+      "step": 9641
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3304426092859647e-05,
+      "loss": 0.9932,
+      "step": 9642
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3301353294079446e-05,
+      "loss": 1.0996,
+      "step": 9643
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3298280415185114e-05,
+      "loss": 0.8228,
+      "step": 9644
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3295207456251206e-05,
+      "loss": 0.9395,
+      "step": 9645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3292134417352313e-05,
+      "loss": 0.7471,
+      "step": 9646
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3289061298562997e-05,
+      "loss": 0.7988,
+      "step": 9647
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.328598809995784e-05,
+      "loss": 1.0264,
+      "step": 9648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3282914821611407e-05,
+      "loss": 0.9795,
+      "step": 9649
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3279841463598292e-05,
+      "loss": 1.0264,
+      "step": 9650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.327676802599307e-05,
+      "loss": 0.8789,
+      "step": 9651
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3273694508870324e-05,
+      "loss": 0.8574,
+      "step": 9652
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.327062091230464e-05,
+      "loss": 0.8398,
+      "step": 9653
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3267547236370604e-05,
+      "loss": 0.9434,
+      "step": 9654
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3264473481142805e-05,
+      "loss": 0.8398,
+      "step": 9655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3261399646695843e-05,
+      "loss": 0.9717,
+      "step": 9656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.325832573310429e-05,
+      "loss": 0.8525,
+      "step": 9657
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.325525174044277e-05,
+      "loss": 1.0684,
+      "step": 9658
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3252177668785852e-05,
+      "loss": 0.958,
+      "step": 9659
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.324910351820815e-05,
+      "loss": 0.959,
+      "step": 9660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3246029288784255e-05,
+      "loss": 0.8994,
+      "step": 9661
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.324295498058878e-05,
+      "loss": 0.7109,
+      "step": 9662
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3239880593696326e-05,
+      "loss": 0.7622,
+      "step": 9663
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3236806128181497e-05,
+      "loss": 1.0566,
+      "step": 9664
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.32337315841189e-05,
+      "loss": 0.9346,
+      "step": 9665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3230656961583153e-05,
+      "loss": 0.9297,
+      "step": 9666
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.322758226064886e-05,
+      "loss": 0.8096,
+      "step": 9667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3224507481390637e-05,
+      "loss": 0.8291,
+      "step": 9668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3221432623883102e-05,
+      "loss": 0.9619,
+      "step": 9669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.321835768820087e-05,
+      "loss": 1.0508,
+      "step": 9670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3215282674418558e-05,
+      "loss": 1.0371,
+      "step": 9671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3212207582610796e-05,
+      "loss": 0.9121,
+      "step": 9672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.32091324128522e-05,
+      "loss": 0.8535,
+      "step": 9673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3206057165217405e-05,
+      "loss": 0.9102,
+      "step": 9674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.320298183978102e-05,
+      "loss": 0.8955,
+      "step": 9675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3199906436617696e-05,
+      "loss": 0.9033,
+      "step": 9676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3196830955802044e-05,
+      "loss": 1.0322,
+      "step": 9677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3193755397408714e-05,
+      "loss": 1.0508,
+      "step": 9678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3190679761512324e-05,
+      "loss": 1.0078,
+      "step": 9679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.318760404818753e-05,
+      "loss": 0.999,
+      "step": 9680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.318452825750895e-05,
+      "loss": 0.874,
+      "step": 9681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3181452389551242e-05,
+      "loss": 0.9189,
+      "step": 9682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3178376444389042e-05,
+      "loss": 0.9648,
+      "step": 9683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3175300422096987e-05,
+      "loss": 0.9541,
+      "step": 9684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3172224322749735e-05,
+      "loss": 1.0107,
+      "step": 9685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3169148146421923e-05,
+      "loss": 0.9023,
+      "step": 9686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3166071893188202e-05,
+      "loss": 1.0078,
+      "step": 9687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3162995563123232e-05,
+      "loss": 0.9658,
+      "step": 9688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3159919156301665e-05,
+      "loss": 0.8838,
+      "step": 9689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.315684267279815e-05,
+      "loss": 0.8203,
+      "step": 9690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3153766112687344e-05,
+      "loss": 0.915,
+      "step": 9691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3150689476043916e-05,
+      "loss": 0.915,
+      "step": 9692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3147612762942516e-05,
+      "loss": 0.8672,
+      "step": 9693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3144535973457813e-05,
+      "loss": 0.8486,
+      "step": 9694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3141459107664467e-05,
+      "loss": 0.8457,
+      "step": 9695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3138382165637153e-05,
+      "loss": 0.9287,
+      "step": 9696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3135305147450525e-05,
+      "loss": 1.0137,
+      "step": 9697
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3132228053179268e-05,
+      "loss": 0.7559,
+      "step": 9698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.312915088289805e-05,
+      "loss": 0.8379,
+      "step": 9699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.312607363668154e-05,
+      "loss": 0.8945,
+      "step": 9700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.312299631460441e-05,
+      "loss": 1.0039,
+      "step": 9701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3119918916741348e-05,
+      "loss": 0.8086,
+      "step": 9702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.311684144316703e-05,
+      "loss": 0.9092,
+      "step": 9703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3113763893956134e-05,
+      "loss": 0.8574,
+      "step": 9704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3110686269183348e-05,
+      "loss": 0.8809,
+      "step": 9705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3107608568923358e-05,
+      "loss": 0.8906,
+      "step": 9706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3104530793250842e-05,
+      "loss": 0.8984,
+      "step": 9707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3101452942240492e-05,
+      "loss": 1.0322,
+      "step": 9708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.309837501596701e-05,
+      "loss": 0.9219,
+      "step": 9709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3095297014505067e-05,
+      "loss": 0.9814,
+      "step": 9710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3092218937929376e-05,
+      "loss": 0.8418,
+      "step": 9711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3089140786314622e-05,
+      "loss": 0.75,
+      "step": 9712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3086062559735508e-05,
+      "loss": 0.8613,
+      "step": 9713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3082984258266724e-05,
+      "loss": 0.9287,
+      "step": 9714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.307990588198299e-05,
+      "loss": 1.0039,
+      "step": 9715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.307682743095899e-05,
+      "loss": 0.7178,
+      "step": 9716
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.307374890526944e-05,
+      "loss": 0.9609,
+      "step": 9717
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.307067030498904e-05,
+      "loss": 1.0977,
+      "step": 9718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3067591630192507e-05,
+      "loss": 1.0322,
+      "step": 9719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3064512880954545e-05,
+      "loss": 0.916,
+      "step": 9720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.306143405734987e-05,
+      "loss": 0.8301,
+      "step": 9721
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3058355159453186e-05,
+      "loss": 1.0156,
+      "step": 9722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3055276187339226e-05,
+      "loss": 0.7754,
+      "step": 9723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3052197141082693e-05,
+      "loss": 0.8633,
+      "step": 9724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3049118020758313e-05,
+      "loss": 0.9746,
+      "step": 9725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3046038826440804e-05,
+      "loss": 0.8125,
+      "step": 9726
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3042959558204894e-05,
+      "loss": 1.085,
+      "step": 9727
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.30398802161253e-05,
+      "loss": 0.7764,
+      "step": 9728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.303680080027675e-05,
+      "loss": 0.8828,
+      "step": 9729
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3033721310733986e-05,
+      "loss": 0.9561,
+      "step": 9730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3030641747571724e-05,
+      "loss": 0.8838,
+      "step": 9731
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.30275621108647e-05,
+      "loss": 0.7529,
+      "step": 9732
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3024482400687642e-05,
+      "loss": 1.0479,
+      "step": 9733
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3021402617115295e-05,
+      "loss": 0.958,
+      "step": 9734
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3018322760222397e-05,
+      "loss": 0.916,
+      "step": 9735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3015242830083673e-05,
+      "loss": 0.9014,
+      "step": 9736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3012162826773877e-05,
+      "loss": 0.8906,
+      "step": 9737
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3009082750367745e-05,
+      "loss": 0.8037,
+      "step": 9738
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3006002600940032e-05,
+      "loss": 0.9346,
+      "step": 9739
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3002922378565468e-05,
+      "loss": 0.8535,
+      "step": 9740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2999842083318814e-05,
+      "loss": 0.959,
+      "step": 9741
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.299676171527481e-05,
+      "loss": 0.9297,
+      "step": 9742
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.299368127450822e-05,
+      "loss": 0.9512,
+      "step": 9743
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.299060076109378e-05,
+      "loss": 0.917,
+      "step": 9744
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2987520175106265e-05,
+      "loss": 0.9932,
+      "step": 9745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.298443951662041e-05,
+      "loss": 0.9912,
+      "step": 9746
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2981358785710996e-05,
+      "loss": 0.8535,
+      "step": 9747
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.297827798245276e-05,
+      "loss": 1.0684,
+      "step": 9748
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.297519710692049e-05,
+      "loss": 0.9795,
+      "step": 9749
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2972116159188923e-05,
+      "loss": 0.9932,
+      "step": 9750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.296903513933285e-05,
+      "loss": 1.0283,
+      "step": 9751
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2965954047427012e-05,
+      "loss": 0.9297,
+      "step": 9752
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2962872883546203e-05,
+      "loss": 0.7891,
+      "step": 9753
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2959791647765176e-05,
+      "loss": 0.9072,
+      "step": 9754
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2956710340158715e-05,
+      "loss": 1.0273,
+      "step": 9755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2953628960801584e-05,
+      "loss": 0.9297,
+      "step": 9756
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2950547509768573e-05,
+      "loss": 0.874,
+      "step": 9757
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2947465987134444e-05,
+      "loss": 0.9902,
+      "step": 9758
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2944384392973986e-05,
+      "loss": 0.8779,
+      "step": 9759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2941302727361973e-05,
+      "loss": 0.8721,
+      "step": 9760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2938220990373204e-05,
+      "loss": 1.0098,
+      "step": 9761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.293513918208244e-05,
+      "loss": 0.7676,
+      "step": 9762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2932057302564488e-05,
+      "loss": 0.8516,
+      "step": 9763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2928975351894124e-05,
+      "loss": 0.6719,
+      "step": 9764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2925893330146146e-05,
+      "loss": 0.8135,
+      "step": 9765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.292281123739534e-05,
+      "loss": 0.9658,
+      "step": 9766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2919729073716494e-05,
+      "loss": 0.9438,
+      "step": 9767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.291664683918442e-05,
+      "loss": 0.9043,
+      "step": 9768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.29135645338739e-05,
+      "loss": 1.0,
+      "step": 9769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2910482157859736e-05,
+      "loss": 0.9775,
+      "step": 9770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2907399711216734e-05,
+      "loss": 0.7646,
+      "step": 9771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2904317194019686e-05,
+      "loss": 0.8223,
+      "step": 9772
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.290123460634341e-05,
+      "loss": 0.9346,
+      "step": 9773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2898151948262695e-05,
+      "loss": 0.959,
+      "step": 9774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2895069219852357e-05,
+      "loss": 0.9043,
+      "step": 9775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.28919864211872e-05,
+      "loss": 0.8262,
+      "step": 9776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2888903552342042e-05,
+      "loss": 0.8008,
+      "step": 9777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.288582061339169e-05,
+      "loss": 0.8896,
+      "step": 9778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.288273760441096e-05,
+      "loss": 0.916,
+      "step": 9779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.287965452547466e-05,
+      "loss": 0.8379,
+      "step": 9780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2876571376657623e-05,
+      "loss": 1.0557,
+      "step": 9781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.287348815803465e-05,
+      "loss": 0.8896,
+      "step": 9782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2870404869680573e-05,
+      "loss": 0.9492,
+      "step": 9783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.286732151167021e-05,
+      "loss": 0.8994,
+      "step": 9784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2864238084078392e-05,
+      "loss": 0.8906,
+      "step": 9785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2861154586979932e-05,
+      "loss": 1.084,
+      "step": 9786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2858071020449667e-05,
+      "loss": 0.8652,
+      "step": 9787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2854987384562426e-05,
+      "loss": 1.0098,
+      "step": 9788
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2851903679393033e-05,
+      "loss": 0.811,
+      "step": 9789
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.284881990501633e-05,
+      "loss": 0.8584,
+      "step": 9790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2845736061507137e-05,
+      "loss": 0.9717,
+      "step": 9791
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.284265214894031e-05,
+      "loss": 1.0361,
+      "step": 9792
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2839568167390668e-05,
+      "loss": 0.8584,
+      "step": 9793
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.283648411693306e-05,
+      "loss": 1.0312,
+      "step": 9794
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.283339999764232e-05,
+      "loss": 0.9395,
+      "step": 9795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2830315809593303e-05,
+      "loss": 0.8701,
+      "step": 9796
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.282723155286084e-05,
+      "loss": 1.1523,
+      "step": 9797
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2824147227519788e-05,
+      "loss": 0.7979,
+      "step": 9798
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2821062833644982e-05,
+      "loss": 0.8457,
+      "step": 9799
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.281797837131128e-05,
+      "loss": 0.9209,
+      "step": 9800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2814893840593527e-05,
+      "loss": 0.6738,
+      "step": 9801
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2811809241566582e-05,
+      "loss": 1.0293,
+      "step": 9802
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2808724574305292e-05,
+      "loss": 0.9004,
+      "step": 9803
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.280563983888452e-05,
+      "loss": 0.873,
+      "step": 9804
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.280255503537912e-05,
+      "loss": 0.8857,
+      "step": 9805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.279947016386395e-05,
+      "loss": 1.0088,
+      "step": 9806
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.279638522441387e-05,
+      "loss": 0.9668,
+      "step": 9807
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2793300217103747e-05,
+      "loss": 0.9404,
+      "step": 9808
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2790215142008447e-05,
+      "loss": 0.9648,
+      "step": 9809
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.278712999920283e-05,
+      "loss": 0.7212,
+      "step": 9810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2784044788761754e-05,
+      "loss": 0.9297,
+      "step": 9811
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2780959510760107e-05,
+      "loss": 1.0479,
+      "step": 9812
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2777874165272747e-05,
+      "loss": 0.8877,
+      "step": 9813
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2774788752374554e-05,
+      "loss": 0.8994,
+      "step": 9814
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.27717032721404e-05,
+      "loss": 1.0742,
+      "step": 9815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2768617724645157e-05,
+      "loss": 0.9229,
+      "step": 9816
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2765532109963703e-05,
+      "loss": 0.8359,
+      "step": 9817
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.276244642817092e-05,
+      "loss": 0.7549,
+      "step": 9818
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2759360679341686e-05,
+      "loss": 0.9023,
+      "step": 9819
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.275627486355088e-05,
+      "loss": 0.8906,
+      "step": 9820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2753188980873394e-05,
+      "loss": 0.8838,
+      "step": 9821
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2750103031384105e-05,
+      "loss": 1.0586,
+      "step": 9822
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2747017015157903e-05,
+      "loss": 1.0195,
+      "step": 9823
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.274393093226968e-05,
+      "loss": 0.959,
+      "step": 9824
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2740844782794315e-05,
+      "loss": 0.96,
+      "step": 9825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.273775856680672e-05,
+      "loss": 0.9863,
+      "step": 9826
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2734672284381762e-05,
+      "loss": 0.9404,
+      "step": 9827
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2731585935594362e-05,
+      "loss": 1.0,
+      "step": 9828
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2728499520519396e-05,
+      "loss": 0.8594,
+      "step": 9829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2725413039231778e-05,
+      "loss": 0.8057,
+      "step": 9830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2722326491806393e-05,
+      "loss": 1.0977,
+      "step": 9831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2719239878318158e-05,
+      "loss": 1.0762,
+      "step": 9832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2716153198841963e-05,
+      "loss": 0.9062,
+      "step": 9833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2713066453452727e-05,
+      "loss": 0.8691,
+      "step": 9834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2709979642225334e-05,
+      "loss": 0.9854,
+      "step": 9835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2706892765234714e-05,
+      "loss": 1.002,
+      "step": 9836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2703805822555756e-05,
+      "loss": 0.6904,
+      "step": 9837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2700718814263398e-05,
+      "loss": 0.8115,
+      "step": 9838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2697631740432525e-05,
+      "loss": 0.9512,
+      "step": 9839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2694544601138068e-05,
+      "loss": 0.6904,
+      "step": 9840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2691457396454933e-05,
+      "loss": 0.957,
+      "step": 9841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.268837012645805e-05,
+      "loss": 0.8633,
+      "step": 9842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2685282791222324e-05,
+      "loss": 0.9502,
+      "step": 9843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.268219539082269e-05,
+      "loss": 0.9258,
+      "step": 9844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.267910792533405e-05,
+      "loss": 0.9854,
+      "step": 9845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.267602039483135e-05,
+      "loss": 0.833,
+      "step": 9846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.26729327993895e-05,
+      "loss": 0.6924,
+      "step": 9847
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.266984513908343e-05,
+      "loss": 1.0859,
+      "step": 9848
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.266675741398807e-05,
+      "loss": 1.0098,
+      "step": 9849
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.266366962417836e-05,
+      "loss": 0.9092,
+      "step": 9850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.266058176972921e-05,
+      "loss": 0.9922,
+      "step": 9851
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.265749385071557e-05,
+      "loss": 0.9248,
+      "step": 9852
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.265440586721237e-05,
+      "loss": 0.8486,
+      "step": 9853
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.265131781929454e-05,
+      "loss": 1.0215,
+      "step": 9854
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.264822970703703e-05,
+      "loss": 0.8271,
+      "step": 9855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2645141530514773e-05,
+      "loss": 0.874,
+      "step": 9856
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2642053289802707e-05,
+      "loss": 0.998,
+      "step": 9857
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2638964984975786e-05,
+      "loss": 0.9277,
+      "step": 9858
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2635876616108935e-05,
+      "loss": 0.9023,
+      "step": 9859
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2632788183277118e-05,
+      "loss": 0.9688,
+      "step": 9860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2629699686555266e-05,
+      "loss": 0.8945,
+      "step": 9861
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2626611126018344e-05,
+      "loss": 0.9863,
+      "step": 9862
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.262352250174129e-05,
+      "loss": 1.0234,
+      "step": 9863
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.262043381379907e-05,
+      "loss": 1.0391,
+      "step": 9864
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2617345062266613e-05,
+      "loss": 0.8496,
+      "step": 9865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.26142562472189e-05,
+      "loss": 0.8545,
+      "step": 9866
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.261116736873087e-05,
+      "loss": 0.8682,
+      "step": 9867
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2608078426877495e-05,
+      "loss": 0.7783,
+      "step": 9868
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2604989421733718e-05,
+      "loss": 0.9199,
+      "step": 9869
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.260190035337451e-05,
+      "loss": 0.8335,
+      "step": 9870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2598811221874834e-05,
+      "loss": 0.7656,
+      "step": 9871
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2595722027309655e-05,
+      "loss": 0.9043,
+      "step": 9872
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2592632769753932e-05,
+      "loss": 0.9111,
+      "step": 9873
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2589543449282635e-05,
+      "loss": 0.8779,
+      "step": 9874
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.258645406597074e-05,
+      "loss": 0.6118,
+      "step": 9875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2583364619893208e-05,
+      "loss": 1.0283,
+      "step": 9876
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2580275111125013e-05,
+      "loss": 0.9531,
+      "step": 9877
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.257718553974113e-05,
+      "loss": 0.9082,
+      "step": 9878
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2574095905816537e-05,
+      "loss": 0.8091,
+      "step": 9879
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.25710062094262e-05,
+      "loss": 1.1465,
+      "step": 9880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.256791645064511e-05,
+      "loss": 0.9004,
+      "step": 9881
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2564826629548235e-05,
+      "loss": 1.0215,
+      "step": 9882
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2561736746210564e-05,
+      "loss": 0.7471,
+      "step": 9883
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2558646800707072e-05,
+      "loss": 0.8389,
+      "step": 9884
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2555556793112752e-05,
+      "loss": 0.8311,
+      "step": 9885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.255246672350258e-05,
+      "loss": 0.8887,
+      "step": 9886
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.254937659195155e-05,
+      "loss": 0.834,
+      "step": 9887
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2546286398534646e-05,
+      "loss": 0.959,
+      "step": 9888
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.254319614332686e-05,
+      "loss": 0.9023,
+      "step": 9889
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.254010582640318e-05,
+      "loss": 0.9277,
+      "step": 9890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2537015447838603e-05,
+      "loss": 1.0771,
+      "step": 9891
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2533925007708125e-05,
+      "loss": 0.9248,
+      "step": 9892
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2530834506086737e-05,
+      "loss": 0.8701,
+      "step": 9893
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2527743943049436e-05,
+      "loss": 0.9609,
+      "step": 9894
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2524653318671223e-05,
+      "loss": 0.959,
+      "step": 9895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2521562633027102e-05,
+      "loss": 0.8037,
+      "step": 9896
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2518471886192072e-05,
+      "loss": 0.877,
+      "step": 9897
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2515381078241134e-05,
+      "loss": 0.8916,
+      "step": 9898
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.25122902092493e-05,
+      "loss": 0.7969,
+      "step": 9899
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2509199279291566e-05,
+      "loss": 1.0732,
+      "step": 9900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2506108288442948e-05,
+      "loss": 1.2051,
+      "step": 9901
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2503017236778443e-05,
+      "loss": 0.9521,
+      "step": 9902
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2499926124373082e-05,
+      "loss": 0.9941,
+      "step": 9903
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.249683495130186e-05,
+      "loss": 1.0234,
+      "step": 9904
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2493743717639803e-05,
+      "loss": 0.8667,
+      "step": 9905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.249065242346191e-05,
+      "loss": 0.8623,
+      "step": 9906
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2487561068843214e-05,
+      "loss": 1.1035,
+      "step": 9907
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.248446965385872e-05,
+      "loss": 0.9355,
+      "step": 9908
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2481378178583464e-05,
+      "loss": 0.7578,
+      "step": 9909
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.247828664309245e-05,
+      "loss": 0.7451,
+      "step": 9910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2475195047460714e-05,
+      "loss": 0.9229,
+      "step": 9911
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2472103391763268e-05,
+      "loss": 0.8545,
+      "step": 9912
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2469011676075147e-05,
+      "loss": 0.7881,
+      "step": 9913
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2465919900471365e-05,
+      "loss": 0.873,
+      "step": 9914
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2462828065026968e-05,
+      "loss": 0.8984,
+      "step": 9915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2459736169816973e-05,
+      "loss": 1.041,
+      "step": 9916
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2456644214916414e-05,
+      "loss": 0.9365,
+      "step": 9917
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2453552200400325e-05,
+      "loss": 0.9062,
+      "step": 9918
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2450460126343745e-05,
+      "loss": 0.9795,
+      "step": 9919
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2447367992821694e-05,
+      "loss": 0.9619,
+      "step": 9920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.244427579990923e-05,
+      "loss": 0.8809,
+      "step": 9921
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2441183547681375e-05,
+      "loss": 0.9502,
+      "step": 9922
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.243809123621318e-05,
+      "loss": 1.0322,
+      "step": 9923
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.243499886557967e-05,
+      "loss": 0.8914,
+      "step": 9924
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.243190643585591e-05,
+      "loss": 0.7739,
+      "step": 9925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2428813947116924e-05,
+      "loss": 1.0146,
+      "step": 9926
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2425721399437775e-05,
+      "loss": 0.7979,
+      "step": 9927
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2422628792893492e-05,
+      "loss": 0.9336,
+      "step": 9928
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2419536127559143e-05,
+      "loss": 0.8311,
+      "step": 9929
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.241644340350976e-05,
+      "loss": 0.8164,
+      "step": 9930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2413350620820403e-05,
+      "loss": 0.9033,
+      "step": 9931
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2410257779566126e-05,
+      "loss": 0.8818,
+      "step": 9932
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.240716487982198e-05,
+      "loss": 1.0205,
+      "step": 9933
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2404071921663023e-05,
+      "loss": 0.8867,
+      "step": 9934
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.240097890516431e-05,
+      "loss": 0.7998,
+      "step": 9935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.23978858304009e-05,
+      "loss": 0.8701,
+      "step": 9936
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2394792697447846e-05,
+      "loss": 0.9121,
+      "step": 9937
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.239169950638022e-05,
+      "loss": 0.8574,
+      "step": 9938
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2388606257273086e-05,
+      "loss": 0.9512,
+      "step": 9939
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.23855129502015e-05,
+      "loss": 0.9814,
+      "step": 9940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.238241958524053e-05,
+      "loss": 0.9453,
+      "step": 9941
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2379326162465245e-05,
+      "loss": 1.0566,
+      "step": 9942
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.237623268195071e-05,
+      "loss": 0.9785,
+      "step": 9943
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2373139143771992e-05,
+      "loss": 0.8867,
+      "step": 9944
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2370045548004175e-05,
+      "loss": 0.8818,
+      "step": 9945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2366951894722315e-05,
+      "loss": 0.9072,
+      "step": 9946
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.23638581840015e-05,
+      "loss": 1.001,
+      "step": 9947
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.236076441591679e-05,
+      "loss": 1.0186,
+      "step": 9948
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.235767059054328e-05,
+      "loss": 0.9512,
+      "step": 9949
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.235457670795603e-05,
+      "loss": 0.9072,
+      "step": 9950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2351482768230133e-05,
+      "loss": 0.8926,
+      "step": 9951
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.234838877144066e-05,
+      "loss": 0.9854,
+      "step": 9952
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2345294717662703e-05,
+      "loss": 0.9033,
+      "step": 9953
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.234220060697134e-05,
+      "loss": 0.9443,
+      "step": 9954
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2339106439441656e-05,
+      "loss": 0.6924,
+      "step": 9955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.233601221514873e-05,
+      "loss": 0.8301,
+      "step": 9956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.233291793416767e-05,
+      "loss": 0.8711,
+      "step": 9957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2329823596573542e-05,
+      "loss": 0.8428,
+      "step": 9958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.232672920244145e-05,
+      "loss": 0.8887,
+      "step": 9959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2323634751846485e-05,
+      "loss": 0.8984,
+      "step": 9960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2320540244863737e-05,
+      "loss": 0.7866,
+      "step": 9961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2317445681568307e-05,
+      "loss": 0.9609,
+      "step": 9962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2314351062035277e-05,
+      "loss": 1.0947,
+      "step": 9963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.231125638633976e-05,
+      "loss": 0.9453,
+      "step": 9964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2308161654556844e-05,
+      "loss": 0.7002,
+      "step": 9965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2305066866761633e-05,
+      "loss": 0.8545,
+      "step": 9966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.230197202302923e-05,
+      "loss": 1.0684,
+      "step": 9967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2298877123434735e-05,
+      "loss": 0.7344,
+      "step": 9968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.229578216805325e-05,
+      "loss": 0.7229,
+      "step": 9969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2292687156959886e-05,
+      "loss": 0.9141,
+      "step": 9970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2289592090229747e-05,
+      "loss": 0.877,
+      "step": 9971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2286496967937942e-05,
+      "loss": 0.8438,
+      "step": 9972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.228340179015958e-05,
+      "loss": 0.8896,
+      "step": 9973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2280306556969774e-05,
+      "loss": 0.9854,
+      "step": 9974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.227721126844363e-05,
+      "loss": 0.8555,
+      "step": 9975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.227411592465627e-05,
+      "loss": 0.8975,
+      "step": 9976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2271020525682795e-05,
+      "loss": 0.999,
+      "step": 9977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.226792507159834e-05,
+      "loss": 0.9785,
+      "step": 9978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2264829562478005e-05,
+      "loss": 1.041,
+      "step": 9979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2261733998396923e-05,
+      "loss": 0.8057,
+      "step": 9980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.225863837943021e-05,
+      "loss": 0.9023,
+      "step": 9981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2255542705652983e-05,
+      "loss": 1.0098,
+      "step": 9982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2252446977140365e-05,
+      "loss": 0.9746,
+      "step": 9983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.224935119396749e-05,
+      "loss": 0.9717,
+      "step": 9984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2246255356209468e-05,
+      "loss": 0.9512,
+      "step": 9985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2243159463941445e-05,
+      "loss": 0.8555,
+      "step": 9986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.224006351723853e-05,
+      "loss": 0.8359,
+      "step": 9987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2236967516175868e-05,
+      "loss": 0.8916,
+      "step": 9988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2233871460828578e-05,
+      "loss": 0.917,
+      "step": 9989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.22307753512718e-05,
+      "loss": 0.8535,
+      "step": 9990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2227679187580664e-05,
+      "loss": 0.6631,
+      "step": 9991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2224582969830307e-05,
+      "loss": 0.8887,
+      "step": 9992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.222148669809586e-05,
+      "loss": 0.7568,
+      "step": 9993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.221839037245247e-05,
+      "loss": 0.9424,
+      "step": 9994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2215293992975262e-05,
+      "loss": 0.9385,
+      "step": 9995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2212197559739388e-05,
+      "loss": 0.9023,
+      "step": 9996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.220910107281998e-05,
+      "loss": 0.7803,
+      "step": 9997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2206004532292192e-05,
+      "loss": 0.833,
+      "step": 9998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2202907938231158e-05,
+      "loss": 0.9033,
+      "step": 9999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.219981129071203e-05,
+      "loss": 0.8613,
+      "step": 10000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.219671458980995e-05,
+      "loss": 1.0127,
+      "step": 10001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.219361783560007e-05,
+      "loss": 0.8438,
+      "step": 10002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.219052102815753e-05,
+      "loss": 1.0361,
+      "step": 10003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2187424167557496e-05,
+      "loss": 0.8506,
+      "step": 10004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2184327253875098e-05,
+      "loss": 0.9893,
+      "step": 10005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.218123028718551e-05,
+      "loss": 0.835,
+      "step": 10006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2178133267563875e-05,
+      "loss": 0.9893,
+      "step": 10007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2175036195085354e-05,
+      "loss": 0.7812,
+      "step": 10008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2171939069825097e-05,
+      "loss": 1.0273,
+      "step": 10009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2168841891858272e-05,
+      "loss": 0.9209,
+      "step": 10010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2165744661260025e-05,
+      "loss": 0.9746,
+      "step": 10011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2162647378105532e-05,
+      "loss": 1.0156,
+      "step": 10012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2159550042469938e-05,
+      "loss": 0.917,
+      "step": 10013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2156452654428425e-05,
+      "loss": 0.999,
+      "step": 10014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.215335521405614e-05,
+      "loss": 0.877,
+      "step": 10015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2150257721428264e-05,
+      "loss": 0.8984,
+      "step": 10016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2147160176619953e-05,
+      "loss": 0.9805,
+      "step": 10017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2144062579706377e-05,
+      "loss": 0.9932,
+      "step": 10018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.214096493076271e-05,
+      "loss": 0.9229,
+      "step": 10019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2137867229864122e-05,
+      "loss": 1.0078,
+      "step": 10020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2134769477085777e-05,
+      "loss": 0.9766,
+      "step": 10021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.213167167250286e-05,
+      "loss": 0.8936,
+      "step": 10022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.212857381619054e-05,
+      "loss": 0.8516,
+      "step": 10023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2125475908224e-05,
+      "loss": 1.0059,
+      "step": 10024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2122377948678402e-05,
+      "loss": 0.9141,
+      "step": 10025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2119279937628938e-05,
+      "loss": 0.7881,
+      "step": 10026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2116181875150777e-05,
+      "loss": 0.9268,
+      "step": 10027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.211308376131911e-05,
+      "loss": 0.7988,
+      "step": 10028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2109985596209106e-05,
+      "loss": 0.7617,
+      "step": 10029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.210688737989597e-05,
+      "loss": 0.8594,
+      "step": 10030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2103789112454862e-05,
+      "loss": 0.957,
+      "step": 10031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2100690793960986e-05,
+      "loss": 0.9072,
+      "step": 10032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2097592424489523e-05,
+      "loss": 0.8701,
+      "step": 10033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2094494004115655e-05,
+      "loss": 0.8066,
+      "step": 10034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.209139553291458e-05,
+      "loss": 0.9766,
+      "step": 10035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.208829701096149e-05,
+      "loss": 0.9121,
+      "step": 10036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.208519843833157e-05,
+      "loss": 1.0039,
+      "step": 10037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2082099815100016e-05,
+      "loss": 0.8145,
+      "step": 10038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.207900114134202e-05,
+      "loss": 0.8809,
+      "step": 10039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2075902417132785e-05,
+      "loss": 0.9238,
+      "step": 10040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.20728036425475e-05,
+      "loss": 0.8838,
+      "step": 10041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2069704817661366e-05,
+      "loss": 1.0273,
+      "step": 10042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2066605942549587e-05,
+      "loss": 0.7197,
+      "step": 10043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.206350701728736e-05,
+      "loss": 0.9248,
+      "step": 10044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2060408041949878e-05,
+      "loss": 1.0908,
+      "step": 10045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2057309016612357e-05,
+      "loss": 0.876,
+      "step": 10046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2054209941349996e-05,
+      "loss": 0.9883,
+      "step": 10047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2051110816238e-05,
+      "loss": 1.0576,
+      "step": 10048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2048011641351582e-05,
+      "loss": 0.9043,
+      "step": 10049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.204491241676594e-05,
+      "loss": 0.9033,
+      "step": 10050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.204181314255629e-05,
+      "loss": 0.814,
+      "step": 10051
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2038713818797834e-05,
+      "loss": 0.999,
+      "step": 10052
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.20356144455658e-05,
+      "loss": 0.8345,
+      "step": 10053
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.203251502293538e-05,
+      "loss": 0.8691,
+      "step": 10054
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.20294155509818e-05,
+      "loss": 0.7202,
+      "step": 10055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2026316029780276e-05,
+      "loss": 0.9668,
+      "step": 10056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2023216459406022e-05,
+      "loss": 0.9277,
+      "step": 10057
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2020116839934245e-05,
+      "loss": 0.8535,
+      "step": 10058
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2017017171440186e-05,
+      "loss": 0.9697,
+      "step": 10059
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2013917453999044e-05,
+      "loss": 1.0186,
+      "step": 10060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2010817687686052e-05,
+      "loss": 0.9619,
+      "step": 10061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2007717872576424e-05,
+      "loss": 0.8623,
+      "step": 10062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2004618008745387e-05,
+      "loss": 0.7822,
+      "step": 10063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.200151809626817e-05,
+      "loss": 0.9512,
+      "step": 10064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1998418135219996e-05,
+      "loss": 0.9189,
+      "step": 10065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1995318125676083e-05,
+      "loss": 0.9287,
+      "step": 10066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.199221806771168e-05,
+      "loss": 1.0508,
+      "step": 10067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1989117961401995e-05,
+      "loss": 0.7832,
+      "step": 10068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.198601780682227e-05,
+      "loss": 0.9072,
+      "step": 10069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.198291760404773e-05,
+      "loss": 1.0283,
+      "step": 10070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1979817353153615e-05,
+      "loss": 0.876,
+      "step": 10071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1976717054215153e-05,
+      "loss": 0.957,
+      "step": 10072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1973616707307584e-05,
+      "loss": 0.958,
+      "step": 10073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1970516312506138e-05,
+      "loss": 0.7637,
+      "step": 10074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1967415869886062e-05,
+      "loss": 0.9619,
+      "step": 10075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1964315379522585e-05,
+      "loss": 0.9287,
+      "step": 10076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.196121484149095e-05,
+      "loss": 1.0205,
+      "step": 10077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1958114255866397e-05,
+      "loss": 0.8584,
+      "step": 10078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1955013622724177e-05,
+      "loss": 0.9033,
+      "step": 10079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1951912942139516e-05,
+      "loss": 0.77,
+      "step": 10080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1948812214187675e-05,
+      "loss": 0.9287,
+      "step": 10081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1945711438943887e-05,
+      "loss": 0.8018,
+      "step": 10082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1942610616483413e-05,
+      "loss": 0.8701,
+      "step": 10083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1939509746881486e-05,
+      "loss": 0.9473,
+      "step": 10084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1936408830213364e-05,
+      "loss": 0.8232,
+      "step": 10085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1933307866554293e-05,
+      "loss": 0.7842,
+      "step": 10086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.193020685597953e-05,
+      "loss": 0.8057,
+      "step": 10087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1927105798564315e-05,
+      "loss": 1.0156,
+      "step": 10088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1924004694383915e-05,
+      "loss": 0.7803,
+      "step": 10089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1920903543513575e-05,
+      "loss": 0.8916,
+      "step": 10090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1917802346028566e-05,
+      "loss": 0.9316,
+      "step": 10091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1914701102004124e-05,
+      "loss": 0.8691,
+      "step": 10092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.191159981151552e-05,
+      "loss": 0.7539,
+      "step": 10093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1908498474638005e-05,
+      "loss": 0.9072,
+      "step": 10094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1905397091446857e-05,
+      "loss": 0.9902,
+      "step": 10095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1902295662017314e-05,
+      "loss": 0.8945,
+      "step": 10096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.189919418642466e-05,
+      "loss": 0.8389,
+      "step": 10097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.189609266474414e-05,
+      "loss": 1.04,
+      "step": 10098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.189299109705103e-05,
+      "loss": 0.9072,
+      "step": 10099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.188988948342059e-05,
+      "loss": 0.8262,
+      "step": 10100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1886787823928097e-05,
+      "loss": 0.79,
+      "step": 10101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1883686118648804e-05,
+      "loss": 0.918,
+      "step": 10102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1880584367657996e-05,
+      "loss": 0.9746,
+      "step": 10103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1877482571030927e-05,
+      "loss": 0.9248,
+      "step": 10104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1874380728842883e-05,
+      "loss": 0.96,
+      "step": 10105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.187127884116913e-05,
+      "loss": 0.9648,
+      "step": 10106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1868176908084947e-05,
+      "loss": 1.0537,
+      "step": 10107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.18650749296656e-05,
+      "loss": 0.8145,
+      "step": 10108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1861972905986374e-05,
+      "loss": 1.0264,
+      "step": 10109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1858870837122536e-05,
+      "loss": 0.7617,
+      "step": 10110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1855768723149375e-05,
+      "loss": 1.0176,
+      "step": 10111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.185266656414216e-05,
+      "loss": 0.8398,
+      "step": 10112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.184956436017618e-05,
+      "loss": 0.7549,
+      "step": 10113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1846462111326703e-05,
+      "loss": 0.8652,
+      "step": 10114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1843359817669035e-05,
+      "loss": 1.0156,
+      "step": 10115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1840257479278435e-05,
+      "loss": 1.04,
+      "step": 10116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1837155096230202e-05,
+      "loss": 0.8545,
+      "step": 10117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.183405266859962e-05,
+      "loss": 0.9053,
+      "step": 10118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1830950196461972e-05,
+      "loss": 0.834,
+      "step": 10119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.182784767989254e-05,
+      "loss": 1.0312,
+      "step": 10120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.182474511896663e-05,
+      "loss": 1.041,
+      "step": 10121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1821642513759516e-05,
+      "loss": 0.7969,
+      "step": 10122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.18185398643465e-05,
+      "loss": 0.96,
+      "step": 10123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1815437170802865e-05,
+      "loss": 0.9893,
+      "step": 10124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1812334433203916e-05,
+      "loss": 0.8789,
+      "step": 10125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1809231651624933e-05,
+      "loss": 1.043,
+      "step": 10126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1806128826141223e-05,
+      "loss": 0.9668,
+      "step": 10127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1803025956828075e-05,
+      "loss": 0.9336,
+      "step": 10128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1799923043760792e-05,
+      "loss": 0.7578,
+      "step": 10129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.179682008701467e-05,
+      "loss": 0.8486,
+      "step": 10130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1793717086665004e-05,
+      "loss": 0.7949,
+      "step": 10131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1790614042787106e-05,
+      "loss": 0.8574,
+      "step": 10132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1787510955456263e-05,
+      "loss": 1.043,
+      "step": 10133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1784407824747795e-05,
+      "loss": 0.8154,
+      "step": 10134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.178130465073699e-05,
+      "loss": 0.916,
+      "step": 10135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1778201433499167e-05,
+      "loss": 0.8213,
+      "step": 10136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.177509817310962e-05,
+      "loss": 0.9727,
+      "step": 10137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1771994869643663e-05,
+      "loss": 0.9375,
+      "step": 10138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1768891523176598e-05,
+      "loss": 0.7168,
+      "step": 10139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.176578813378374e-05,
+      "loss": 0.8975,
+      "step": 10140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1762684701540392e-05,
+      "loss": 0.9385,
+      "step": 10141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.175958122652188e-05,
+      "loss": 0.9043,
+      "step": 10142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.17564777088035e-05,
+      "loss": 0.9072,
+      "step": 10143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1753374148460576e-05,
+      "loss": 0.9297,
+      "step": 10144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.175027054556841e-05,
+      "loss": 0.8652,
+      "step": 10145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.174716690020233e-05,
+      "loss": 0.8906,
+      "step": 10146
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1744063212437648e-05,
+      "loss": 0.8115,
+      "step": 10147
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.174095948234968e-05,
+      "loss": 0.9736,
+      "step": 10148
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1737855710013748e-05,
+      "loss": 0.8457,
+      "step": 10149
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1734751895505167e-05,
+      "loss": 0.8311,
+      "step": 10150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1731648038899255e-05,
+      "loss": 0.8428,
+      "step": 10151
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1728544140271347e-05,
+      "loss": 0.877,
+      "step": 10152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.172544019969675e-05,
+      "loss": 0.7373,
+      "step": 10153
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1722336217250794e-05,
+      "loss": 1.0449,
+      "step": 10154
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1719232193008798e-05,
+      "loss": 1.0098,
+      "step": 10155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1716128127046106e-05,
+      "loss": 0.8936,
+      "step": 10156
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.171302401943802e-05,
+      "loss": 0.7412,
+      "step": 10157
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1709919870259887e-05,
+      "loss": 0.8574,
+      "step": 10158
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1706815679587017e-05,
+      "loss": 1.084,
+      "step": 10159
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.170371144749476e-05,
+      "loss": 0.875,
+      "step": 10160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1700607174058433e-05,
+      "loss": 0.9707,
+      "step": 10161
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.169750285935337e-05,
+      "loss": 0.9238,
+      "step": 10162
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1694398503454904e-05,
+      "loss": 0.9434,
+      "step": 10163
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.169129410643837e-05,
+      "loss": 0.9258,
+      "step": 10164
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1688189668379106e-05,
+      "loss": 0.8525,
+      "step": 10165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1685085189352442e-05,
+      "loss": 0.8447,
+      "step": 10166
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.168198066943371e-05,
+      "loss": 0.8838,
+      "step": 10167
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1678876108698262e-05,
+      "loss": 0.7568,
+      "step": 10168
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1675771507221426e-05,
+      "loss": 0.8486,
+      "step": 10169
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1672666865078544e-05,
+      "loss": 0.8262,
+      "step": 10170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.166956218234495e-05,
+      "loss": 0.8569,
+      "step": 10171
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1666457459096e-05,
+      "loss": 0.8887,
+      "step": 10172
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1663352695407025e-05,
+      "loss": 0.959,
+      "step": 10173
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1660247891353375e-05,
+      "loss": 0.9814,
+      "step": 10174
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1657143047010383e-05,
+      "loss": 1.0039,
+      "step": 10175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.165403816245341e-05,
+      "loss": 0.8691,
+      "step": 10176
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.165093323775779e-05,
+      "loss": 0.7891,
+      "step": 10177
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.164782827299888e-05,
+      "loss": 0.9902,
+      "step": 10178
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.164472326825202e-05,
+      "loss": 0.8164,
+      "step": 10179
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.164161822359257e-05,
+      "loss": 1.0811,
+      "step": 10180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1638513139095866e-05,
+      "loss": 0.9365,
+      "step": 10181
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1635408014837274e-05,
+      "loss": 0.9111,
+      "step": 10182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1632302850892133e-05,
+      "loss": 0.834,
+      "step": 10183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1629197647335807e-05,
+      "loss": 0.9551,
+      "step": 10184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.162609240424364e-05,
+      "loss": 1.0547,
+      "step": 10185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1622987121690998e-05,
+      "loss": 0.8984,
+      "step": 10186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.161988179975322e-05,
+      "loss": 0.9131,
+      "step": 10187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1616776438505688e-05,
+      "loss": 0.7344,
+      "step": 10188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.161367103802374e-05,
+      "loss": 0.8799,
+      "step": 10189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1610565598382743e-05,
+      "loss": 1.0146,
+      "step": 10190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1607460119658054e-05,
+      "loss": 0.8916,
+      "step": 10191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1604354601925038e-05,
+      "loss": 0.9551,
+      "step": 10192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1601249045259057e-05,
+      "loss": 0.9727,
+      "step": 10193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1598143449735467e-05,
+      "loss": 0.9297,
+      "step": 10194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1595037815429633e-05,
+      "loss": 0.9062,
+      "step": 10195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1591932142416928e-05,
+      "loss": 1.0166,
+      "step": 10196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1588826430772704e-05,
+      "loss": 0.8521,
+      "step": 10197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1585720680572342e-05,
+      "loss": 0.9336,
+      "step": 10198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1582614891891196e-05,
+      "loss": 0.9043,
+      "step": 10199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1579509064804648e-05,
+      "loss": 0.8643,
+      "step": 10200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1576403199388056e-05,
+      "loss": 1.041,
+      "step": 10201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1573297295716794e-05,
+      "loss": 1.042,
+      "step": 10202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1570191353866236e-05,
+      "loss": 0.8232,
+      "step": 10203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1567085373911752e-05,
+      "loss": 0.8672,
+      "step": 10204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.156397935592871e-05,
+      "loss": 0.625,
+      "step": 10205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1560873299992492e-05,
+      "loss": 0.9697,
+      "step": 10206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1557767206178467e-05,
+      "loss": 0.7373,
+      "step": 10207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.155466107456202e-05,
+      "loss": 0.8447,
+      "step": 10208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1551554905218513e-05,
+      "loss": 1.0205,
+      "step": 10209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1548448698223336e-05,
+      "loss": 0.7295,
+      "step": 10210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.154534245365186e-05,
+      "loss": 0.7842,
+      "step": 10211
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.154223617157947e-05,
+      "loss": 1.0293,
+      "step": 10212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.153912985208154e-05,
+      "loss": 1.0762,
+      "step": 10213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1536023495233464e-05,
+      "loss": 0.9238,
+      "step": 10214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1532917101110602e-05,
+      "loss": 0.8398,
+      "step": 10215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1529810669788354e-05,
+      "loss": 0.918,
+      "step": 10216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1526704201342103e-05,
+      "loss": 0.8408,
+      "step": 10217
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.152359769584723e-05,
+      "loss": 0.8652,
+      "step": 10218
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1520491153379127e-05,
+      "loss": 1.1074,
+      "step": 10219
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1517384574013166e-05,
+      "loss": 0.8369,
+      "step": 10220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1514277957824747e-05,
+      "loss": 0.8662,
+      "step": 10221
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1511171304889258e-05,
+      "loss": 0.9551,
+      "step": 10222
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1508064615282087e-05,
+      "loss": 0.9463,
+      "step": 10223
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1504957889078616e-05,
+      "loss": 0.8721,
+      "step": 10224
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1501851126354247e-05,
+      "loss": 0.876,
+      "step": 10225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.149874432718437e-05,
+      "loss": 1.0117,
+      "step": 10226
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1495637491644373e-05,
+      "loss": 0.6836,
+      "step": 10227
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1492530619809654e-05,
+      "loss": 0.8311,
+      "step": 10228
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1489423711755603e-05,
+      "loss": 0.7217,
+      "step": 10229
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1486316767557623e-05,
+      "loss": 0.9932,
+      "step": 10230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.14832097872911e-05,
+      "loss": 1.0469,
+      "step": 10231
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1480102771031442e-05,
+      "loss": 0.8809,
+      "step": 10232
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1476995718854044e-05,
+      "loss": 0.8477,
+      "step": 10233
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.14738886308343e-05,
+      "loss": 0.8682,
+      "step": 10234
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1470781507047613e-05,
+      "loss": 0.7686,
+      "step": 10235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.146767434756939e-05,
+      "loss": 0.7939,
+      "step": 10236
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.146456715247502e-05,
+      "loss": 0.6318,
+      "step": 10237
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1461459921839918e-05,
+      "loss": 0.8652,
+      "step": 10238
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1458352655739477e-05,
+      "loss": 0.7656,
+      "step": 10239
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1455245354249108e-05,
+      "loss": 0.8906,
+      "step": 10240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1452138017444218e-05,
+      "loss": 0.8867,
+      "step": 10241
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1449030645400203e-05,
+      "loss": 0.9863,
+      "step": 10242
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.144592323819248e-05,
+      "loss": 0.8613,
+      "step": 10243
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.144281579589645e-05,
+      "loss": 0.8105,
+      "step": 10244
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1439708318587522e-05,
+      "loss": 0.8564,
+      "step": 10245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.143660080634111e-05,
+      "loss": 0.9287,
+      "step": 10246
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.143349325923262e-05,
+      "loss": 0.7686,
+      "step": 10247
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1430385677337465e-05,
+      "loss": 0.8311,
+      "step": 10248
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1427278060731057e-05,
+      "loss": 0.9766,
+      "step": 10249
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1424170409488804e-05,
+      "loss": 0.8682,
+      "step": 10250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1421062723686132e-05,
+      "loss": 0.9932,
+      "step": 10251
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1417955003398436e-05,
+      "loss": 0.7793,
+      "step": 10252
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1414847248701154e-05,
+      "loss": 1.0195,
+      "step": 10253
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1411739459669677e-05,
+      "loss": 1.0293,
+      "step": 10254
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1408631636379446e-05,
+      "loss": 0.8105,
+      "step": 10255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.140552377890586e-05,
+      "loss": 0.9443,
+      "step": 10256
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1402415887324354e-05,
+      "loss": 0.8418,
+      "step": 10257
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.139930796171033e-05,
+      "loss": 0.9688,
+      "step": 10258
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1396200002139223e-05,
+      "loss": 0.7451,
+      "step": 10259
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1393092008686443e-05,
+      "loss": 0.8291,
+      "step": 10260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1389983981427423e-05,
+      "loss": 0.8047,
+      "step": 10261
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1386875920437573e-05,
+      "loss": 1.0693,
+      "step": 10262
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1383767825792332e-05,
+      "loss": 0.8936,
+      "step": 10263
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1380659697567107e-05,
+      "loss": 0.7998,
+      "step": 10264
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1377551535837335e-05,
+      "loss": 0.9531,
+      "step": 10265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1374443340678438e-05,
+      "loss": 0.7393,
+      "step": 10266
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1371335112165843e-05,
+      "loss": 0.917,
+      "step": 10267
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1368226850374974e-05,
+      "loss": 0.916,
+      "step": 10268
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.136511855538127e-05,
+      "loss": 0.9482,
+      "step": 10269
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1362010227260143e-05,
+      "loss": 0.9121,
+      "step": 10270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1358901866087044e-05,
+      "loss": 0.8242,
+      "step": 10271
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1355793471937387e-05,
+      "loss": 0.8193,
+      "step": 10272
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1352685044886616e-05,
+      "loss": 1.0146,
+      "step": 10273
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1349576585010146e-05,
+      "loss": 0.8936,
+      "step": 10274
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1346468092383433e-05,
+      "loss": 0.9668,
+      "step": 10275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1343359567081895e-05,
+      "loss": 0.8286,
+      "step": 10276
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.134025100918097e-05,
+      "loss": 1.1113,
+      "step": 10277
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1337142418756095e-05,
+      "loss": 0.7158,
+      "step": 10278
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1334033795882708e-05,
+      "loss": 1.0127,
+      "step": 10279
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1330925140636242e-05,
+      "loss": 0.9473,
+      "step": 10280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.132781645309214e-05,
+      "loss": 0.9727,
+      "step": 10281
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.132470773332583e-05,
+      "loss": 0.9316,
+      "step": 10282
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1321598981412766e-05,
+      "loss": 0.9521,
+      "step": 10283
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1318490197428376e-05,
+      "loss": 0.9424,
+      "step": 10284
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1315381381448112e-05,
+      "loss": 0.8955,
+      "step": 10285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1312272533547408e-05,
+      "loss": 0.8467,
+      "step": 10286
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.130916365380171e-05,
+      "loss": 0.8652,
+      "step": 10287
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.130605474228646e-05,
+      "loss": 0.875,
+      "step": 10288
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.13029457990771e-05,
+      "loss": 0.8447,
+      "step": 10289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.129983682424908e-05,
+      "loss": 0.9287,
+      "step": 10290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1296727817877842e-05,
+      "loss": 0.8955,
+      "step": 10291
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1293618780038833e-05,
+      "loss": 0.8379,
+      "step": 10292
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.12905097108075e-05,
+      "loss": 0.7246,
+      "step": 10293
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.128740061025929e-05,
+      "loss": 0.8818,
+      "step": 10294
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.128429147846966e-05,
+      "loss": 0.9189,
+      "step": 10295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1281182315514046e-05,
+      "loss": 1.0293,
+      "step": 10296
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.127807312146791e-05,
+      "loss": 0.8237,
+      "step": 10297
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1274963896406695e-05,
+      "loss": 0.6982,
+      "step": 10298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1271854640405856e-05,
+      "loss": 0.9258,
+      "step": 10299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1268745353540844e-05,
+      "loss": 0.7451,
+      "step": 10300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1265636035887116e-05,
+      "loss": 0.957,
+      "step": 10301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1262526687520122e-05,
+      "loss": 0.9219,
+      "step": 10302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1259417308515317e-05,
+      "loss": 0.8311,
+      "step": 10303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.125630789894816e-05,
+      "loss": 0.876,
+      "step": 10304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1253198458894105e-05,
+      "loss": 0.8232,
+      "step": 10305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.125008898842861e-05,
+      "loss": 1.0781,
+      "step": 10306
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1246979487627128e-05,
+      "loss": 1.0215,
+      "step": 10307
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1243869956565122e-05,
+      "loss": 0.9199,
+      "step": 10308
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.124076039531805e-05,
+      "loss": 0.9121,
+      "step": 10309
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1237650803961378e-05,
+      "loss": 0.9248,
+      "step": 10310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1234541182570556e-05,
+      "loss": 0.8906,
+      "step": 10311
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1231431531221046e-05,
+      "loss": 1.0391,
+      "step": 10312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1228321849988322e-05,
+      "loss": 0.9121,
+      "step": 10313
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1225212138947834e-05,
+      "loss": 0.8867,
+      "step": 10314
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1222102398175054e-05,
+      "loss": 1.0137,
+      "step": 10315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.121899262774544e-05,
+      "loss": 0.6367,
+      "step": 10316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1215882827734462e-05,
+      "loss": 0.9766,
+      "step": 10317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.121277299821758e-05,
+      "loss": 0.832,
+      "step": 10318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.120966313927027e-05,
+      "loss": 1.0449,
+      "step": 10319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1206553250967988e-05,
+      "loss": 0.9639,
+      "step": 10320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.120344333338621e-05,
+      "loss": 0.9287,
+      "step": 10321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.12003333866004e-05,
+      "loss": 0.9102,
+      "step": 10322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1197223410686022e-05,
+      "loss": 1.001,
+      "step": 10323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1194113405718566e-05,
+      "loss": 0.7632,
+      "step": 10324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.119100337177348e-05,
+      "loss": 0.8643,
+      "step": 10325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.118789330892625e-05,
+      "loss": 1.0527,
+      "step": 10326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1184783217252337e-05,
+      "loss": 0.8779,
+      "step": 10327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.118167309682723e-05,
+      "loss": 0.918,
+      "step": 10328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.117856294772638e-05,
+      "loss": 0.876,
+      "step": 10329
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.117545277002528e-05,
+      "loss": 0.8887,
+      "step": 10330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1172342563799396e-05,
+      "loss": 0.9365,
+      "step": 10331
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.116923232912421e-05,
+      "loss": 0.8965,
+      "step": 10332
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1166122066075188e-05,
+      "loss": 0.8906,
+      "step": 10333
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1163011774727817e-05,
+      "loss": 0.9482,
+      "step": 10334
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1159901455157565e-05,
+      "loss": 0.8711,
+      "step": 10335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1156791107439925e-05,
+      "loss": 0.9844,
+      "step": 10336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1153680731650355e-05,
+      "loss": 1.0166,
+      "step": 10337
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1150570327864358e-05,
+      "loss": 1.0098,
+      "step": 10338
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1147459896157397e-05,
+      "loss": 0.9092,
+      "step": 10339
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1144349436604963e-05,
+      "loss": 0.9307,
+      "step": 10340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1141238949282527e-05,
+      "loss": 1.0039,
+      "step": 10341
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1138128434265583e-05,
+      "loss": 0.6992,
+      "step": 10342
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1135017891629608e-05,
+      "loss": 0.8516,
+      "step": 10343
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1131907321450087e-05,
+      "loss": 1.042,
+      "step": 10344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1128796723802504e-05,
+      "loss": 1.001,
+      "step": 10345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1125686098762343e-05,
+      "loss": 0.9883,
+      "step": 10346
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1122575446405088e-05,
+      "loss": 0.9229,
+      "step": 10347
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1119464766806237e-05,
+      "loss": 0.9385,
+      "step": 10348
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1116354060041257e-05,
+      "loss": 0.7817,
+      "step": 10349
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1113243326185656e-05,
+      "loss": 0.9219,
+      "step": 10350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1110132565314912e-05,
+      "loss": 0.875,
+      "step": 10351
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1107021777504513e-05,
+      "loss": 0.8965,
+      "step": 10352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.110391096282995e-05,
+      "loss": 1.0186,
+      "step": 10353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1100800121366712e-05,
+      "loss": 0.8867,
+      "step": 10354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1097689253190297e-05,
+      "loss": 0.9932,
+      "step": 10355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1094578358376197e-05,
+      "loss": 0.9678,
+      "step": 10356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.109146743699989e-05,
+      "loss": 0.9639,
+      "step": 10357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1088356489136884e-05,
+      "loss": 0.667,
+      "step": 10358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.108524551486266e-05,
+      "loss": 0.9883,
+      "step": 10359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.108213451425273e-05,
+      "loss": 0.8809,
+      "step": 10360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1079023487382567e-05,
+      "loss": 1.0664,
+      "step": 10361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1075912434327686e-05,
+      "loss": 0.875,
+      "step": 10362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1072801355163566e-05,
+      "loss": 1.084,
+      "step": 10363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.106969024996572e-05,
+      "loss": 0.7622,
+      "step": 10364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.106657911880963e-05,
+      "loss": 0.8564,
+      "step": 10365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1063467961770813e-05,
+      "loss": 1.0127,
+      "step": 10366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.106035677892474e-05,
+      "loss": 1.0088,
+      "step": 10367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.105724557034694e-05,
+      "loss": 1.0127,
+      "step": 10368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1054134336112894e-05,
+      "loss": 0.8604,
+      "step": 10369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.105102307629811e-05,
+      "loss": 1.0732,
+      "step": 10370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1047911790978085e-05,
+      "loss": 0.9297,
+      "step": 10371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1044800480228323e-05,
+      "loss": 0.9043,
+      "step": 10372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.104168914412432e-05,
+      "loss": 0.8242,
+      "step": 10373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1038577782741597e-05,
+      "loss": 0.8867,
+      "step": 10374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1035466396155635e-05,
+      "loss": 0.7812,
+      "step": 10375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1032354984441955e-05,
+      "loss": 0.9961,
+      "step": 10376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.102924354767605e-05,
+      "loss": 1.0254,
+      "step": 10377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.102613208593344e-05,
+      "loss": 0.8481,
+      "step": 10378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.102302059928961e-05,
+      "loss": 0.8945,
+      "step": 10379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1019909087820088e-05,
+      "loss": 0.9717,
+      "step": 10380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1016797551600368e-05,
+      "loss": 0.9932,
+      "step": 10381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.101368599070596e-05,
+      "loss": 0.8535,
+      "step": 10382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1010574405212377e-05,
+      "loss": 0.6997,
+      "step": 10383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.100746279519513e-05,
+      "loss": 0.9521,
+      "step": 10384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1004351160729714e-05,
+      "loss": 0.958,
+      "step": 10385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1001239501891658e-05,
+      "loss": 0.8281,
+      "step": 10386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.099812781875646e-05,
+      "loss": 0.8291,
+      "step": 10387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0995016111399627e-05,
+      "loss": 0.833,
+      "step": 10388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.099190437989669e-05,
+      "loss": 0.8252,
+      "step": 10389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.098879262432315e-05,
+      "loss": 0.9443,
+      "step": 10390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.098568084475452e-05,
+      "loss": 0.9668,
+      "step": 10391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0982569041266313e-05,
+      "loss": 0.6987,
+      "step": 10392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0979457213934045e-05,
+      "loss": 1.0615,
+      "step": 10393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.097634536283323e-05,
+      "loss": 0.7812,
+      "step": 10394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0973233488039383e-05,
+      "loss": 0.8682,
+      "step": 10395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0970121589628025e-05,
+      "loss": 0.9756,
+      "step": 10396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0967009667674665e-05,
+      "loss": 0.9844,
+      "step": 10397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0963897722254833e-05,
+      "loss": 0.8945,
+      "step": 10398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.096078575344403e-05,
+      "loss": 0.8447,
+      "step": 10399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0957673761317787e-05,
+      "loss": 0.877,
+      "step": 10400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.095456174595161e-05,
+      "loss": 0.8398,
+      "step": 10401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0951449707421038e-05,
+      "loss": 0.8193,
+      "step": 10402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.094833764580157e-05,
+      "loss": 1.002,
+      "step": 10403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0945225561168746e-05,
+      "loss": 0.9863,
+      "step": 10404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.094211345359807e-05,
+      "loss": 0.749,
+      "step": 10405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0939001323165072e-05,
+      "loss": 0.7422,
+      "step": 10406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0935889169945276e-05,
+      "loss": 0.7803,
+      "step": 10407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0932776994014195e-05,
+      "loss": 0.8408,
+      "step": 10408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0929664795447372e-05,
+      "loss": 0.9678,
+      "step": 10409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0926552574320308e-05,
+      "loss": 0.8623,
+      "step": 10410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0923440330708544e-05,
+      "loss": 0.8193,
+      "step": 10411
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0920328064687595e-05,
+      "loss": 0.9844,
+      "step": 10412
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0917215776332992e-05,
+      "loss": 0.9893,
+      "step": 10413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.091410346572026e-05,
+      "loss": 0.8691,
+      "step": 10414
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0910991132924927e-05,
+      "loss": 1.0273,
+      "step": 10415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0907878778022513e-05,
+      "loss": 0.7764,
+      "step": 10416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0904766401088556e-05,
+      "loss": 0.7988,
+      "step": 10417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0901654002198577e-05,
+      "loss": 0.9082,
+      "step": 10418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.089854158142811e-05,
+      "loss": 0.8984,
+      "step": 10419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0895429138852674e-05,
+      "loss": 0.8213,
+      "step": 10420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0892316674547817e-05,
+      "loss": 0.8643,
+      "step": 10421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.088920418858905e-05,
+      "loss": 1.0283,
+      "step": 10422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0886091681051917e-05,
+      "loss": 0.9639,
+      "step": 10423
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.088297915201194e-05,
+      "loss": 0.9492,
+      "step": 10424
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.087986660154466e-05,
+      "loss": 0.8745,
+      "step": 10425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0876754029725602e-05,
+      "loss": 0.9443,
+      "step": 10426
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0873641436630304e-05,
+      "loss": 0.7427,
+      "step": 10427
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0870528822334296e-05,
+      "loss": 0.8877,
+      "step": 10428
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0867416186913117e-05,
+      "loss": 0.9697,
+      "step": 10429
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0864303530442295e-05,
+      "loss": 0.9092,
+      "step": 10430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0861190852997373e-05,
+      "loss": 1.0439,
+      "step": 10431
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0858078154653873e-05,
+      "loss": 0.8389,
+      "step": 10432
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0854965435487347e-05,
+      "loss": 0.9551,
+      "step": 10433
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0851852695573324e-05,
+      "loss": 0.9951,
+      "step": 10434
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0848739934987337e-05,
+      "loss": 0.7314,
+      "step": 10435
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0845627153804927e-05,
+      "loss": 0.9844,
+      "step": 10436
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0842514352101636e-05,
+      "loss": 0.7617,
+      "step": 10437
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0839401529953e-05,
+      "loss": 0.8945,
+      "step": 10438
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0836288687434555e-05,
+      "loss": 0.9346,
+      "step": 10439
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0833175824621845e-05,
+      "loss": 0.7817,
+      "step": 10440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0830062941590408e-05,
+      "loss": 0.7817,
+      "step": 10441
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0826950038415782e-05,
+      "loss": 0.8633,
+      "step": 10442
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0823837115173516e-05,
+      "loss": 1.0293,
+      "step": 10443
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.082072417193914e-05,
+      "loss": 1.0879,
+      "step": 10444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0817611208788206e-05,
+      "loss": 0.9229,
+      "step": 10445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0814498225796253e-05,
+      "loss": 0.9023,
+      "step": 10446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0811385223038823e-05,
+      "loss": 0.9229,
+      "step": 10447
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0808272200591455e-05,
+      "loss": 0.9727,
+      "step": 10448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0805159158529707e-05,
+      "loss": 0.8916,
+      "step": 10449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.08020460969291e-05,
+      "loss": 0.6931,
+      "step": 10450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0798933015865207e-05,
+      "loss": 0.8301,
+      "step": 10451
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.079581991541355e-05,
+      "loss": 0.9316,
+      "step": 10452
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.079270679564969e-05,
+      "loss": 0.8701,
+      "step": 10453
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0789593656649166e-05,
+      "loss": 0.8203,
+      "step": 10454
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0786480498487526e-05,
+      "loss": 0.9365,
+      "step": 10455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0783367321240316e-05,
+      "loss": 0.8027,
+      "step": 10456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.078025412498309e-05,
+      "loss": 0.8975,
+      "step": 10457
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.077714090979138e-05,
+      "loss": 0.8379,
+      "step": 10458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0774027675740753e-05,
+      "loss": 0.8643,
+      "step": 10459
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.077091442290675e-05,
+      "loss": 0.8926,
+      "step": 10460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0767801151364924e-05,
+      "loss": 1.0547,
+      "step": 10461
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0764687861190814e-05,
+      "loss": 0.9043,
+      "step": 10462
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0761574552459988e-05,
+      "loss": 1.0488,
+      "step": 10463
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.075846122524798e-05,
+      "loss": 0.9219,
+      "step": 10464
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0755347879630352e-05,
+      "loss": 1.0762,
+      "step": 10465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.075223451568265e-05,
+      "loss": 1.1846,
+      "step": 10466
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0749121133480434e-05,
+      "loss": 0.918,
+      "step": 10467
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0746007733099244e-05,
+      "loss": 1.0801,
+      "step": 10468
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0742894314614645e-05,
+      "loss": 0.8584,
+      "step": 10469
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0739780878102183e-05,
+      "loss": 0.9565,
+      "step": 10470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.073666742363742e-05,
+      "loss": 0.8438,
+      "step": 10471
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0733553951295895e-05,
+      "loss": 0.9023,
+      "step": 10472
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0730440461153183e-05,
+      "loss": 0.918,
+      "step": 10473
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0727326953284823e-05,
+      "loss": 0.9004,
+      "step": 10474
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0724213427766376e-05,
+      "loss": 0.7637,
+      "step": 10475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0721099884673402e-05,
+      "loss": 0.8682,
+      "step": 10476
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0717986324081455e-05,
+      "loss": 0.9756,
+      "step": 10477
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0714872746066095e-05,
+      "loss": 1.0869,
+      "step": 10478
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.071175915070287e-05,
+      "loss": 0.8262,
+      "step": 10479
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0708645538067343e-05,
+      "loss": 0.8027,
+      "step": 10480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0705531908235077e-05,
+      "loss": 0.8379,
+      "step": 10481
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0702418261281623e-05,
+      "loss": 0.7979,
+      "step": 10482
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0699304597282546e-05,
+      "loss": 0.8613,
+      "step": 10483
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0696190916313398e-05,
+      "loss": 1.0127,
+      "step": 10484
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0693077218449754e-05,
+      "loss": 0.8936,
+      "step": 10485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0689963503767156e-05,
+      "loss": 1.0762,
+      "step": 10486
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0686849772341176e-05,
+      "loss": 1.0088,
+      "step": 10487
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0683736024247367e-05,
+      "loss": 0.7988,
+      "step": 10488
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0680622259561304e-05,
+      "loss": 1.0547,
+      "step": 10489
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0677508478358535e-05,
+      "loss": 0.96,
+      "step": 10490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.067439468071463e-05,
+      "loss": 1.0098,
+      "step": 10491
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0671280866705144e-05,
+      "loss": 0.7319,
+      "step": 10492
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.066816703640565e-05,
+      "loss": 0.8311,
+      "step": 10493
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0665053189891706e-05,
+      "loss": 0.9365,
+      "step": 10494
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0661939327238876e-05,
+      "loss": 0.9658,
+      "step": 10495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.065882544852273e-05,
+      "loss": 0.8135,
+      "step": 10496
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0655711553818824e-05,
+      "loss": 0.8906,
+      "step": 10497
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0652597643202727e-05,
+      "loss": 0.8828,
+      "step": 10498
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0649483716750006e-05,
+      "loss": 0.7754,
+      "step": 10499
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0646369774536222e-05,
+      "loss": 0.9297,
+      "step": 10500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.064325581663695e-05,
+      "loss": 0.791,
+      "step": 10501
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0640141843127747e-05,
+      "loss": 0.9102,
+      "step": 10502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0637027854084188e-05,
+      "loss": 0.998,
+      "step": 10503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.063391384958183e-05,
+      "loss": 0.7051,
+      "step": 10504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.063079982969625e-05,
+      "loss": 0.9219,
+      "step": 10505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0627685794503018e-05,
+      "loss": 0.8203,
+      "step": 10506
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.062457174407769e-05,
+      "loss": 0.8916,
+      "step": 10507
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0621457678495845e-05,
+      "loss": 0.9355,
+      "step": 10508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0618343597833052e-05,
+      "loss": 0.8887,
+      "step": 10509
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.061522950216488e-05,
+      "loss": 0.8672,
+      "step": 10510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0612115391566892e-05,
+      "loss": 0.9688,
+      "step": 10511
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.060900126611467e-05,
+      "loss": 0.9473,
+      "step": 10512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.060588712588377e-05,
+      "loss": 0.998,
+      "step": 10513
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0602772970949777e-05,
+      "loss": 0.9404,
+      "step": 10514
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0599658801388256e-05,
+      "loss": 1.0283,
+      "step": 10515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0596544617274778e-05,
+      "loss": 0.8447,
+      "step": 10516
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0593430418684917e-05,
+      "loss": 0.8594,
+      "step": 10517
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.059031620569425e-05,
+      "loss": 0.9238,
+      "step": 10518
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0587201978378336e-05,
+      "loss": 0.9492,
+      "step": 10519
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0584087736812757e-05,
+      "loss": 0.8945,
+      "step": 10520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.058097348107309e-05,
+      "loss": 0.9912,
+      "step": 10521
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.057785921123491e-05,
+      "loss": 0.8237,
+      "step": 10522
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0574744927373777e-05,
+      "loss": 0.8438,
+      "step": 10523
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.057163062956528e-05,
+      "loss": 0.9268,
+      "step": 10524
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.056851631788499e-05,
+      "loss": 0.8721,
+      "step": 10525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.056540199240848e-05,
+      "loss": 0.7998,
+      "step": 10526
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0562287653211323e-05,
+      "loss": 0.9092,
+      "step": 10527
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0559173300369105e-05,
+      "loss": 0.9678,
+      "step": 10528
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.055605893395739e-05,
+      "loss": 0.9473,
+      "step": 10529
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0552944554051764e-05,
+      "loss": 0.9238,
+      "step": 10530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.054983016072779e-05,
+      "loss": 1.0322,
+      "step": 10531
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0546715754061068e-05,
+      "loss": 0.8916,
+      "step": 10532
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0543601334127156e-05,
+      "loss": 1.0176,
+      "step": 10533
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.054048690100164e-05,
+      "loss": 0.9189,
+      "step": 10534
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0537372454760095e-05,
+      "loss": 1.0156,
+      "step": 10535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.05342579954781e-05,
+      "loss": 0.834,
+      "step": 10536
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0531143523231237e-05,
+      "loss": 1.0166,
+      "step": 10537
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0528029038095087e-05,
+      "loss": 0.7588,
+      "step": 10538
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0524914540145215e-05,
+      "loss": 0.9688,
+      "step": 10539
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0521800029457218e-05,
+      "loss": 1.0195,
+      "step": 10540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0518685506106665e-05,
+      "loss": 0.8193,
+      "step": 10541
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0515570970169144e-05,
+      "loss": 0.8145,
+      "step": 10542
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0512456421720225e-05,
+      "loss": 0.8428,
+      "step": 10543
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0509341860835504e-05,
+      "loss": 0.7285,
+      "step": 10544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0506227287590546e-05,
+      "loss": 0.7539,
+      "step": 10545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.050311270206095e-05,
+      "loss": 0.9619,
+      "step": 10546
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0499998104322283e-05,
+      "loss": 0.8066,
+      "step": 10547
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0496883494450133e-05,
+      "loss": 0.96,
+      "step": 10548
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.049376887252008e-05,
+      "loss": 0.959,
+      "step": 10549
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0490654238607712e-05,
+      "loss": 1.0098,
+      "step": 10550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0487539592788605e-05,
+      "loss": 0.7715,
+      "step": 10551
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.048442493513835e-05,
+      "loss": 1.0137,
+      "step": 10552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.048131026573252e-05,
+      "loss": 1.0107,
+      "step": 10553
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.047819558464671e-05,
+      "loss": 0.8818,
+      "step": 10554
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0475080891956497e-05,
+      "loss": 0.7178,
+      "step": 10555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0471966187737477e-05,
+      "loss": 1.0039,
+      "step": 10556
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0468851472065213e-05,
+      "loss": 0.9316,
+      "step": 10557
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.046573674501531e-05,
+      "loss": 0.9043,
+      "step": 10558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.046262200666334e-05,
+      "loss": 0.9746,
+      "step": 10559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.04595072570849e-05,
+      "loss": 0.7803,
+      "step": 10560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.045639249635557e-05,
+      "loss": 0.9121,
+      "step": 10561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0453277724550933e-05,
+      "loss": 0.9062,
+      "step": 10562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0450162941746578e-05,
+      "loss": 0.9609,
+      "step": 10563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0447048148018097e-05,
+      "loss": 0.9688,
+      "step": 10564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0443933343441068e-05,
+      "loss": 0.5071,
+      "step": 10565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0440818528091085e-05,
+      "loss": 0.9414,
+      "step": 10566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.043770370204373e-05,
+      "loss": 1.0664,
+      "step": 10567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.04345888653746e-05,
+      "loss": 1.0723,
+      "step": 10568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0431474018159266e-05,
+      "loss": 0.79,
+      "step": 10569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0428359160473333e-05,
+      "loss": 0.9492,
+      "step": 10570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0425244292392384e-05,
+      "loss": 0.8838,
+      "step": 10571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0422129413992006e-05,
+      "loss": 1.0723,
+      "step": 10572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0419014525347785e-05,
+      "loss": 0.8877,
+      "step": 10573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.041589962653532e-05,
+      "loss": 0.8945,
+      "step": 10574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0412784717630187e-05,
+      "loss": 0.877,
+      "step": 10575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0409669798707984e-05,
+      "loss": 0.915,
+      "step": 10576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0406554869844303e-05,
+      "loss": 0.6851,
+      "step": 10577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0403439931114732e-05,
+      "loss": 0.8057,
+      "step": 10578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0400324982594856e-05,
+      "loss": 0.9561,
+      "step": 10579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0397210024360267e-05,
+      "loss": 1.0322,
+      "step": 10580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.039409505648657e-05,
+      "loss": 0.9873,
+      "step": 10581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0390980079049333e-05,
+      "loss": 0.8438,
+      "step": 10582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0387865092124167e-05,
+      "loss": 0.9473,
+      "step": 10583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0384750095786655e-05,
+      "loss": 0.8799,
+      "step": 10584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.038163509011239e-05,
+      "loss": 0.875,
+      "step": 10585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.037852007517696e-05,
+      "loss": 0.9463,
+      "step": 10586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0375405051055966e-05,
+      "loss": 0.874,
+      "step": 10587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0372290017824993e-05,
+      "loss": 0.8628,
+      "step": 10588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0369174975559636e-05,
+      "loss": 0.7617,
+      "step": 10589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0366059924335487e-05,
+      "loss": 1.0195,
+      "step": 10590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0362944864228145e-05,
+      "loss": 0.8701,
+      "step": 10591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.035982979531319e-05,
+      "loss": 0.9082,
+      "step": 10592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0356714717666235e-05,
+      "loss": 0.8184,
+      "step": 10593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0353599631362853e-05,
+      "loss": 1.0361,
+      "step": 10594
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0350484536478655e-05,
+      "loss": 0.8779,
+      "step": 10595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.034736943308922e-05,
+      "loss": 0.9551,
+      "step": 10596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.034425432127016e-05,
+      "loss": 0.9961,
+      "step": 10597
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.034113920109705e-05,
+      "loss": 0.835,
+      "step": 10598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0338024072645503e-05,
+      "loss": 0.9072,
+      "step": 10599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.03349089359911e-05,
+      "loss": 0.9053,
+      "step": 10600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0331793791209446e-05,
+      "loss": 0.9316,
+      "step": 10601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0328678638376132e-05,
+      "loss": 0.7705,
+      "step": 10602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0325563477566757e-05,
+      "loss": 0.8945,
+      "step": 10603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.03224483088569e-05,
+      "loss": 1.0117,
+      "step": 10604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0319333132322187e-05,
+      "loss": 0.8594,
+      "step": 10605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0316217948038187e-05,
+      "loss": 0.9707,
+      "step": 10606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0313102756080517e-05,
+      "loss": 0.833,
+      "step": 10607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0309987556524758e-05,
+      "loss": 1.0371,
+      "step": 10608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0306872349446512e-05,
+      "loss": 0.8594,
+      "step": 10609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0303757134921377e-05,
+      "loss": 1.0264,
+      "step": 10610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0300641913024958e-05,
+      "loss": 0.9824,
+      "step": 10611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0297526683832834e-05,
+      "loss": 0.8887,
+      "step": 10612
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0294411447420614e-05,
+      "loss": 0.8496,
+      "step": 10613
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0291296203863896e-05,
+      "loss": 1.1152,
+      "step": 10614
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.028818095323828e-05,
+      "loss": 0.9473,
+      "step": 10615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0285065695619348e-05,
+      "loss": 0.8418,
+      "step": 10616
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.028195043108272e-05,
+      "loss": 0.8613,
+      "step": 10617
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.027883515970398e-05,
+      "loss": 1.043,
+      "step": 10618
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0275719881558734e-05,
+      "loss": 1.002,
+      "step": 10619
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0272604596722572e-05,
+      "loss": 0.9072,
+      "step": 10620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0269489305271107e-05,
+      "loss": 0.9355,
+      "step": 10621
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0266374007279918e-05,
+      "loss": 0.8066,
+      "step": 10622
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0263258702824627e-05,
+      "loss": 0.7725,
+      "step": 10623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0260143391980807e-05,
+      "loss": 1.0264,
+      "step": 10624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0257028074824084e-05,
+      "loss": 0.9316,
+      "step": 10625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.025391275143004e-05,
+      "loss": 1.0488,
+      "step": 10626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0250797421874285e-05,
+      "loss": 0.8652,
+      "step": 10627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.024768208623241e-05,
+      "loss": 0.8281,
+      "step": 10628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0244566744580023e-05,
+      "loss": 0.8271,
+      "step": 10629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0241451396992713e-05,
+      "loss": 0.8545,
+      "step": 10630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0238336043546093e-05,
+      "loss": 1.0244,
+      "step": 10631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.023522068431576e-05,
+      "loss": 0.8311,
+      "step": 10632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.023210531937731e-05,
+      "loss": 0.8809,
+      "step": 10633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0228989948806347e-05,
+      "loss": 0.5754,
+      "step": 10634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.022587457267847e-05,
+      "loss": 0.9033,
+      "step": 10635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.022275919106928e-05,
+      "loss": 0.8662,
+      "step": 10636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0219643804054387e-05,
+      "loss": 0.8281,
+      "step": 10637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0216528411709375e-05,
+      "loss": 0.9219,
+      "step": 10638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0213413014109866e-05,
+      "loss": 1.0791,
+      "step": 10639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.021029761133144e-05,
+      "loss": 1.0127,
+      "step": 10640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0207182203449717e-05,
+      "loss": 0.9014,
+      "step": 10641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0204066790540287e-05,
+      "loss": 0.9121,
+      "step": 10642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.020095137267876e-05,
+      "loss": 0.9541,
+      "step": 10643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0197835949940726e-05,
+      "loss": 1.1035,
+      "step": 10644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0194720522401803e-05,
+      "loss": 0.7588,
+      "step": 10645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0191605090137582e-05,
+      "loss": 0.9453,
+      "step": 10646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0188489653223672e-05,
+      "loss": 0.9854,
+      "step": 10647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0185374211735666e-05,
+      "loss": 0.9336,
+      "step": 10648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0182258765749182e-05,
+      "loss": 1.043,
+      "step": 10649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0179143315339808e-05,
+      "loss": 0.8975,
+      "step": 10650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0176027860583155e-05,
+      "loss": 0.8789,
+      "step": 10651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0172912401554824e-05,
+      "loss": 0.959,
+      "step": 10652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.016979693833042e-05,
+      "loss": 0.9521,
+      "step": 10653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0166681470985538e-05,
+      "loss": 0.9424,
+      "step": 10654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.016356599959579e-05,
+      "loss": 0.8457,
+      "step": 10655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0160450524236773e-05,
+      "loss": 0.9561,
+      "step": 10656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.01573350449841e-05,
+      "loss": 0.8325,
+      "step": 10657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.015421956191336e-05,
+      "loss": 0.8926,
+      "step": 10658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.015110407510017e-05,
+      "loss": 1.04,
+      "step": 10659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0147988584620126e-05,
+      "loss": 0.875,
+      "step": 10660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0144873090548835e-05,
+      "loss": 0.9678,
+      "step": 10661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0141757592961898e-05,
+      "loss": 0.9795,
+      "step": 10662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.013864209193493e-05,
+      "loss": 0.8662,
+      "step": 10663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0135526587543515e-05,
+      "loss": 0.96,
+      "step": 10664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0132411079863275e-05,
+      "loss": 0.9287,
+      "step": 10665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0129295568969805e-05,
+      "loss": 0.9551,
+      "step": 10666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0126180054938712e-05,
+      "loss": 1.0254,
+      "step": 10667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.01230645378456e-05,
+      "loss": 0.918,
+      "step": 10668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0119949017766073e-05,
+      "loss": 0.8398,
+      "step": 10669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0116833494775743e-05,
+      "loss": 0.9453,
+      "step": 10670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.01137179689502e-05,
+      "loss": 1.0195,
+      "step": 10671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.011060244036506e-05,
+      "loss": 0.8975,
+      "step": 10672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.010748690909592e-05,
+      "loss": 0.8525,
+      "step": 10673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0104371375218395e-05,
+      "loss": 0.8682,
+      "step": 10674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0101255838808075e-05,
+      "loss": 0.8496,
+      "step": 10675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.009814029994058e-05,
+      "loss": 1.04,
+      "step": 10676
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.009502475869151e-05,
+      "loss": 0.8223,
+      "step": 10677
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0091909215136463e-05,
+      "loss": 0.9033,
+      "step": 10678
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.008879366935105e-05,
+      "loss": 0.8711,
+      "step": 10679
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.008567812141088e-05,
+      "loss": 0.9121,
+      "step": 10680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0082562571391552e-05,
+      "loss": 0.9355,
+      "step": 10681
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0079447019368677e-05,
+      "loss": 0.8516,
+      "step": 10682
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.007633146541785e-05,
+      "loss": 1.0215,
+      "step": 10683
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0073215909614686e-05,
+      "loss": 0.9316,
+      "step": 10684
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0070100352034783e-05,
+      "loss": 0.8809,
+      "step": 10685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0066984792753757e-05,
+      "loss": 0.9844,
+      "step": 10686
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.00638692318472e-05,
+      "loss": 0.9785,
+      "step": 10687
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0060753669390732e-05,
+      "loss": 0.667,
+      "step": 10688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0057638105459945e-05,
+      "loss": 1.0547,
+      "step": 10689
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0054522540130453e-05,
+      "loss": 0.9785,
+      "step": 10690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.005140697347786e-05,
+      "loss": 0.9355,
+      "step": 10691
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0048291405577774e-05,
+      "loss": 0.9609,
+      "step": 10692
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.004517583650579e-05,
+      "loss": 0.8115,
+      "step": 10693
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.004206026633753e-05,
+      "loss": 0.957,
+      "step": 10694
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0038944695148585e-05,
+      "loss": 0.9404,
+      "step": 10695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.003582912301457e-05,
+      "loss": 0.8086,
+      "step": 10696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0032713550011084e-05,
+      "loss": 0.8438,
+      "step": 10697
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0029597976213743e-05,
+      "loss": 0.8926,
+      "step": 10698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0026482401698138e-05,
+      "loss": 0.8174,
+      "step": 10699
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0023366826539892e-05,
+      "loss": 1.1025,
+      "step": 10700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0020251250814592e-05,
+      "loss": 0.9727,
+      "step": 10701
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.001713567459786e-05,
+      "loss": 0.8975,
+      "step": 10702
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0014020097965293e-05,
+      "loss": 0.9834,
+      "step": 10703
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0010904520992503e-05,
+      "loss": 1.1084,
+      "step": 10704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.000778894375509e-05,
+      "loss": 0.8311,
+      "step": 10705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0004673366328663e-05,
+      "loss": 0.8096,
+      "step": 10706
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0001557788788825e-05,
+      "loss": 0.9688,
+      "step": 10707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998442211211185e-05,
+      "loss": 0.9277,
+      "step": 10708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9995326633671347e-05,
+      "loss": 0.9424,
+      "step": 10709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999221105624492e-05,
+      "loss": 0.8223,
+      "step": 10710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9989095479007504e-05,
+      "loss": 0.834,
+      "step": 10711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9985979902034714e-05,
+      "loss": 0.7305,
+      "step": 10712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9982864325402146e-05,
+      "loss": 0.9824,
+      "step": 10713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9979748749185415e-05,
+      "loss": 0.8877,
+      "step": 10714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9976633173460114e-05,
+      "loss": 0.9443,
+      "step": 10715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.997351759830187e-05,
+      "loss": 1.0098,
+      "step": 10716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9970402023786264e-05,
+      "loss": 1.0957,
+      "step": 10717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.996728644998892e-05,
+      "loss": 0.9736,
+      "step": 10718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9964170876985437e-05,
+      "loss": 0.9453,
+      "step": 10719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.996105530485142e-05,
+      "loss": 0.9629,
+      "step": 10720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9957939733662474e-05,
+      "loss": 1.0195,
+      "step": 10721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9954824163494215e-05,
+      "loss": 0.8594,
+      "step": 10722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9951708594422233e-05,
+      "loss": 0.8916,
+      "step": 10723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9948593026522145e-05,
+      "loss": 0.8662,
+      "step": 10724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.994547745986955e-05,
+      "loss": 0.9453,
+      "step": 10725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.994236189454006e-05,
+      "loss": 0.8496,
+      "step": 10726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9939246330609278e-05,
+      "loss": 0.8496,
+      "step": 10727
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9936130768152807e-05,
+      "loss": 0.7451,
+      "step": 10728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.993301520724625e-05,
+      "loss": 1.0215,
+      "step": 10729
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9929899647965227e-05,
+      "loss": 0.8789,
+      "step": 10730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9926784090385324e-05,
+      "loss": 1.0215,
+      "step": 10731
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9923668534582157e-05,
+      "loss": 1.0088,
+      "step": 10732
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9920552980631336e-05,
+      "loss": 0.8896,
+      "step": 10733
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9917437428608454e-05,
+      "loss": 1.0127,
+      "step": 10734
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.991432187858913e-05,
+      "loss": 0.9004,
+      "step": 10735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9911206330648957e-05,
+      "loss": 0.8525,
+      "step": 10736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9908090784863544e-05,
+      "loss": 0.9697,
+      "step": 10737
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.99049752413085e-05,
+      "loss": 0.6777,
+      "step": 10738
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.990185970005943e-05,
+      "loss": 0.9473,
+      "step": 10739
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.989874416119193e-05,
+      "loss": 0.9463,
+      "step": 10740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.989562862478162e-05,
+      "loss": 0.8701,
+      "step": 10741
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9892513090904086e-05,
+      "loss": 0.9512,
+      "step": 10742
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.988939755963495e-05,
+      "loss": 1.0264,
+      "step": 10743
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.988628203104981e-05,
+      "loss": 0.9277,
+      "step": 10744
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.988316650522427e-05,
+      "loss": 0.6924,
+      "step": 10745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.988005098223393e-05,
+      "loss": 0.7676,
+      "step": 10746
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.987693546215441e-05,
+      "loss": 1.0127,
+      "step": 10747
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9873819945061295e-05,
+      "loss": 0.9004,
+      "step": 10748
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.98707044310302e-05,
+      "loss": 0.9834,
+      "step": 10749
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.986758892013673e-05,
+      "loss": 0.9814,
+      "step": 10750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9864473412456485e-05,
+      "loss": 0.9971,
+      "step": 10751
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9861357908065078e-05,
+      "loss": 0.8799,
+      "step": 10752
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.98582424070381e-05,
+      "loss": 0.8389,
+      "step": 10753
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9855126909451168e-05,
+      "loss": 0.6436,
+      "step": 10754
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9852011415379874e-05,
+      "loss": 0.8721,
+      "step": 10755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9848895924899834e-05,
+      "loss": 0.7988,
+      "step": 10756
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.984578043808664e-05,
+      "loss": 0.8213,
+      "step": 10757
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.984266495501591e-05,
+      "loss": 0.9434,
+      "step": 10758
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.983954947576323e-05,
+      "loss": 0.9541,
+      "step": 10759
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9836434000404216e-05,
+      "loss": 0.9268,
+      "step": 10760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9833318529014466e-05,
+      "loss": 0.96,
+      "step": 10761
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.983020306166959e-05,
+      "loss": 0.7588,
+      "step": 10762
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982708759844518e-05,
+      "loss": 0.9482,
+      "step": 10763
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9823972139416848e-05,
+      "loss": 1.0127,
+      "step": 10764
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9820856684660192e-05,
+      "loss": 1.0596,
+      "step": 10765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.981774123425082e-05,
+      "loss": 0.9424,
+      "step": 10766
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.981462578826433e-05,
+      "loss": 1.001,
+      "step": 10767
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9811510346776335e-05,
+      "loss": 0.8311,
+      "step": 10768
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.980839490986242e-05,
+      "loss": 0.8945,
+      "step": 10769
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9805279477598203e-05,
+      "loss": 0.8223,
+      "step": 10770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9802164050059278e-05,
+      "loss": 0.8818,
+      "step": 10771
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9799048627321247e-05,
+      "loss": 0.8906,
+      "step": 10772
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.979593320945972e-05,
+      "loss": 0.8525,
+      "step": 10773
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9792817796550286e-05,
+      "loss": 0.9268,
+      "step": 10774
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9789702388668566e-05,
+      "loss": 0.9531,
+      "step": 10775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.978658698589014e-05,
+      "loss": 0.8491,
+      "step": 10776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9783471588290628e-05,
+      "loss": 0.7725,
+      "step": 10777
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.978035619594562e-05,
+      "loss": 0.9141,
+      "step": 10778
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9777240808930726e-05,
+      "loss": 0.8936,
+      "step": 10779
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9774125427321535e-05,
+      "loss": 1.0713,
+      "step": 10780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.977101005119366e-05,
+      "loss": 0.9404,
+      "step": 10781
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9767894680622694e-05,
+      "loss": 0.957,
+      "step": 10782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.976477931568425e-05,
+      "loss": 0.9512,
+      "step": 10783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.976166395645391e-05,
+      "loss": 0.7051,
+      "step": 10784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.975854860300729e-05,
+      "loss": 1.0244,
+      "step": 10785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9755433255419984e-05,
+      "loss": 0.8867,
+      "step": 10786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9752317913767597e-05,
+      "loss": 0.9658,
+      "step": 10787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.974920257812572e-05,
+      "loss": 1.0811,
+      "step": 10788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9746087248569965e-05,
+      "loss": 0.8125,
+      "step": 10789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.974297192517592e-05,
+      "loss": 0.8877,
+      "step": 10790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9739856608019196e-05,
+      "loss": 0.7725,
+      "step": 10791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9736741297175383e-05,
+      "loss": 0.9043,
+      "step": 10792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9733625992720085e-05,
+      "loss": 0.9336,
+      "step": 10793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9730510694728903e-05,
+      "loss": 0.791,
+      "step": 10794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.972739540327743e-05,
+      "loss": 0.9023,
+      "step": 10795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9724280118441273e-05,
+      "loss": 0.6973,
+      "step": 10796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9721164840296025e-05,
+      "loss": 0.8945,
+      "step": 10797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9718049568917286e-05,
+      "loss": 1.0264,
+      "step": 10798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714934304380656e-05,
+      "loss": 0.8711,
+      "step": 10799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971181904676173e-05,
+      "loss": 0.9883,
+      "step": 10800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970870379613611e-05,
+      "loss": 0.7339,
+      "step": 10801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970558855257939e-05,
+      "loss": 0.9639,
+      "step": 10802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9702473316167173e-05,
+      "loss": 0.9795,
+      "step": 10803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9699358086975052e-05,
+      "loss": 0.8398,
+      "step": 10804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9696242865078626e-05,
+      "loss": 0.9463,
+      "step": 10805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.969312765055349e-05,
+      "loss": 0.9639,
+      "step": 10806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.969001244347525e-05,
+      "loss": 0.7471,
+      "step": 10807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.968689724391949e-05,
+      "loss": 0.834,
+      "step": 10808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9683782051961816e-05,
+      "loss": 1.0176,
+      "step": 10809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.968066686767782e-05,
+      "loss": 1.042,
+      "step": 10810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9677551691143105e-05,
+      "loss": 0.8623,
+      "step": 10811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9674436522433257e-05,
+      "loss": 0.9834,
+      "step": 10812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9671321361623878e-05,
+      "loss": 0.8662,
+      "step": 10813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.966820620879056e-05,
+      "loss": 1.0039,
+      "step": 10814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.966509106400891e-05,
+      "loss": 0.8975,
+      "step": 10815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9661975927354507e-05,
+      "loss": 1.1094,
+      "step": 10816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9658860798902956e-05,
+      "loss": 0.9023,
+      "step": 10817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.965574567872985e-05,
+      "loss": 0.8076,
+      "step": 10818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9652630566910784e-05,
+      "loss": 1.0078,
+      "step": 10819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9649515463521355e-05,
+      "loss": 0.6924,
+      "step": 10820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9646400368637154e-05,
+      "loss": 0.9072,
+      "step": 10821
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.964328528233378e-05,
+      "loss": 0.8662,
+      "step": 10822
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9640170204686816e-05,
+      "loss": 0.9482,
+      "step": 10823
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9637055135771866e-05,
+      "loss": 0.8477,
+      "step": 10824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.963394007566452e-05,
+      "loss": 0.8262,
+      "step": 10825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9630825024440374e-05,
+      "loss": 0.9697,
+      "step": 10826
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9627709982175014e-05,
+      "loss": 0.8662,
+      "step": 10827
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9624594948944044e-05,
+      "loss": 0.9473,
+      "step": 10828
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9621479924823045e-05,
+      "loss": 0.8682,
+      "step": 10829
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.961836490988762e-05,
+      "loss": 1.0781,
+      "step": 10830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9615249904213352e-05,
+      "loss": 0.7549,
+      "step": 10831
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.961213490787584e-05,
+      "loss": 1.0342,
+      "step": 10832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9609019920950663e-05,
+      "loss": 0.8535,
+      "step": 10833
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.960590494351343e-05,
+      "loss": 0.8271,
+      "step": 10834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.960278997563973e-05,
+      "loss": 0.8223,
+      "step": 10835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9599675017405148e-05,
+      "loss": 0.8262,
+      "step": 10836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9596560068885275e-05,
+      "loss": 0.6816,
+      "step": 10837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.95934451301557e-05,
+      "loss": 0.9268,
+      "step": 10838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.959033020129202e-05,
+      "loss": 0.9277,
+      "step": 10839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9587215282369816e-05,
+      "loss": 0.8125,
+      "step": 10840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9584100373464688e-05,
+      "loss": 0.7725,
+      "step": 10841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9580985474652215e-05,
+      "loss": 0.791,
+      "step": 10842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9577870586008e-05,
+      "loss": 0.9297,
+      "step": 10843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.957475570760762e-05,
+      "loss": 0.9951,
+      "step": 10844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.957164083952667e-05,
+      "loss": 0.9561,
+      "step": 10845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9568525981840734e-05,
+      "loss": 0.874,
+      "step": 10846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.956541113462541e-05,
+      "loss": 0.876,
+      "step": 10847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.956229629795627e-05,
+      "loss": 0.835,
+      "step": 10848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.955918147190892e-05,
+      "loss": 0.9473,
+      "step": 10849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9556066656558935e-05,
+      "loss": 0.8672,
+      "step": 10850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.955295185198191e-05,
+      "loss": 0.875,
+      "step": 10851
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9549837058253425e-05,
+      "loss": 0.9067,
+      "step": 10852
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9546722275449074e-05,
+      "loss": 0.9824,
+      "step": 10853
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9543607503644438e-05,
+      "loss": 0.9707,
+      "step": 10854
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9540492742915107e-05,
+      "loss": 0.8818,
+      "step": 10855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9537377993336663e-05,
+      "loss": 0.791,
+      "step": 10856
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9534263254984696e-05,
+      "loss": 1.002,
+      "step": 10857
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9531148527934794e-05,
+      "loss": 0.8584,
+      "step": 10858
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9528033812262533e-05,
+      "loss": 0.7617,
+      "step": 10859
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9524919108043506e-05,
+      "loss": 0.791,
+      "step": 10860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9521804415353293e-05,
+      "loss": 0.8633,
+      "step": 10861
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9518689734267485e-05,
+      "loss": 0.8174,
+      "step": 10862
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9515575064861655e-05,
+      "loss": 0.9854,
+      "step": 10863
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.95124604072114e-05,
+      "loss": 0.9414,
+      "step": 10864
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9509345761392295e-05,
+      "loss": 0.9648,
+      "step": 10865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9506231127479927e-05,
+      "loss": 0.812,
+      "step": 10866
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9503116505549873e-05,
+      "loss": 1.0117,
+      "step": 10867
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9500001895677723e-05,
+      "loss": 0.9902,
+      "step": 10868
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9496887297939056e-05,
+      "loss": 0.9531,
+      "step": 10869
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9493772712409457e-05,
+      "loss": 0.9707,
+      "step": 10870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.94906581391645e-05,
+      "loss": 0.9209,
+      "step": 10871
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9487543578279778e-05,
+      "loss": 1.0234,
+      "step": 10872
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9484429029830863e-05,
+      "loss": 0.8867,
+      "step": 10873
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.948131449389334e-05,
+      "loss": 0.9082,
+      "step": 10874
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.947819997054279e-05,
+      "loss": 0.9873,
+      "step": 10875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.947508545985479e-05,
+      "loss": 0.8662,
+      "step": 10876
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.947197096190492e-05,
+      "loss": 0.8223,
+      "step": 10877
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.946885647676877e-05,
+      "loss": 0.8496,
+      "step": 10878
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9465742004521903e-05,
+      "loss": 0.9873,
+      "step": 10879
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9462627545239912e-05,
+      "loss": 0.9277,
+      "step": 10880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9459513098998363e-05,
+      "loss": 0.9727,
+      "step": 10881
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.945639866587285e-05,
+      "loss": 0.8291,
+      "step": 10882
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.945328424593894e-05,
+      "loss": 0.877,
+      "step": 10883
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9450169839272212e-05,
+      "loss": 0.8652,
+      "step": 10884
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9447055445948243e-05,
+      "loss": 0.9199,
+      "step": 10885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9443941066042614e-05,
+      "loss": 0.8906,
+      "step": 10886
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.94408266996309e-05,
+      "loss": 0.9307,
+      "step": 10887
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9437712346788683e-05,
+      "loss": 0.915,
+      "step": 10888
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9434598007591526e-05,
+      "loss": 1.0127,
+      "step": 10889
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9431483682115018e-05,
+      "loss": 0.9746,
+      "step": 10890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9428369370434722e-05,
+      "loss": 0.8857,
+      "step": 10891
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9425255072626227e-05,
+      "loss": 0.9072,
+      "step": 10892
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9422140788765098e-05,
+      "loss": 0.8662,
+      "step": 10893
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9419026518926915e-05,
+      "loss": 1.0059,
+      "step": 10894
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9415912263187246e-05,
+      "loss": 0.959,
+      "step": 10895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9412798021621674e-05,
+      "loss": 0.8037,
+      "step": 10896
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.940968379430576e-05,
+      "loss": 0.8857,
+      "step": 10897
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9406569581315093e-05,
+      "loss": 0.7935,
+      "step": 10898
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.940345538272523e-05,
+      "loss": 0.9268,
+      "step": 10899
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9400341198611754e-05,
+      "loss": 0.8369,
+      "step": 10900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9397227029050233e-05,
+      "loss": 0.9023,
+      "step": 10901
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.939411287411624e-05,
+      "loss": 0.9707,
+      "step": 10902
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9390998733885343e-05,
+      "loss": 1.0479,
+      "step": 10903
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9387884608433115e-05,
+      "loss": 0.9414,
+      "step": 10904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.938477049783513e-05,
+      "loss": 0.873,
+      "step": 10905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.938165640216695e-05,
+      "loss": 0.9385,
+      "step": 10906
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.937854232150416e-05,
+      "loss": 0.8682,
+      "step": 10907
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9375428255922315e-05,
+      "loss": 0.9385,
+      "step": 10908
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9372314205496992e-05,
+      "loss": 0.96,
+      "step": 10909
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9369200170303754e-05,
+      "loss": 0.9824,
+      "step": 10910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9366086150418177e-05,
+      "loss": 0.8955,
+      "step": 10911
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9362972145915822e-05,
+      "loss": 0.7617,
+      "step": 10912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9359858156872263e-05,
+      "loss": 0.9102,
+      "step": 10913
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9356744183363057e-05,
+      "loss": 0.9502,
+      "step": 10914
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9353630225463785e-05,
+      "loss": 0.833,
+      "step": 10915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9350516283249997e-05,
+      "loss": 0.8896,
+      "step": 10916
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.934740235679727e-05,
+      "loss": 0.6919,
+      "step": 10917
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.934428844618118e-05,
+      "loss": 0.916,
+      "step": 10918
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.934117455147727e-05,
+      "loss": 0.9238,
+      "step": 10919
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9338060672761124e-05,
+      "loss": 0.9375,
+      "step": 10920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9334946810108294e-05,
+      "loss": 0.791,
+      "step": 10921
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.933183296359435e-05,
+      "loss": 0.9072,
+      "step": 10922
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9328719133294856e-05,
+      "loss": 0.8467,
+      "step": 10923
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9325605319285377e-05,
+      "loss": 0.8921,
+      "step": 10924
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.932249152164147e-05,
+      "loss": 1.0693,
+      "step": 10925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9319377740438703e-05,
+      "loss": 0.7861,
+      "step": 10926
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9316263975752633e-05,
+      "loss": 0.71,
+      "step": 10927
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.931315022765883e-05,
+      "loss": 0.8623,
+      "step": 10928
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9310036496232848e-05,
+      "loss": 0.6333,
+      "step": 10929
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9306922781550253e-05,
+      "loss": 0.8359,
+      "step": 10930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9303809083686602e-05,
+      "loss": 0.8398,
+      "step": 10931
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.930069540271746e-05,
+      "loss": 0.8154,
+      "step": 10932
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.929758173871838e-05,
+      "loss": 0.8428,
+      "step": 10933
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.929446809176493e-05,
+      "loss": 0.8975,
+      "step": 10934
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.929135446193266e-05,
+      "loss": 0.9561,
+      "step": 10935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9288240849297136e-05,
+      "loss": 1.0137,
+      "step": 10936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.928512725393391e-05,
+      "loss": 0.8379,
+      "step": 10937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9282013675918548e-05,
+      "loss": 0.7441,
+      "step": 10938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.92789001153266e-05,
+      "loss": 0.9482,
+      "step": 10939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9275786572233627e-05,
+      "loss": 0.873,
+      "step": 10940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9272673046715184e-05,
+      "loss": 0.7939,
+      "step": 10941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9269559538846823e-05,
+      "loss": 1.0234,
+      "step": 10942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9266446048704108e-05,
+      "loss": 1.1406,
+      "step": 10943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9263332576362586e-05,
+      "loss": 0.9941,
+      "step": 10944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9260219121897824e-05,
+      "loss": 0.7949,
+      "step": 10945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.925710568538536e-05,
+      "loss": 0.9775,
+      "step": 10946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9253992266900763e-05,
+      "loss": 0.9355,
+      "step": 10947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9250878866519573e-05,
+      "loss": 0.793,
+      "step": 10948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9247765484317356e-05,
+      "loss": 1.0713,
+      "step": 10949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9244652120369654e-05,
+      "loss": 0.8369,
+      "step": 10950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9241538774752027e-05,
+      "loss": 0.9609,
+      "step": 10951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.923842544754002e-05,
+      "loss": 0.8311,
+      "step": 10952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.923531213880919e-05,
+      "loss": 0.9805,
+      "step": 10953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9232198848635083e-05,
+      "loss": 0.8701,
+      "step": 10954
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9229085577093255e-05,
+      "loss": 0.8691,
+      "step": 10955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.922597232425925e-05,
+      "loss": 0.8916,
+      "step": 10956
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9222859090208626e-05,
+      "loss": 0.7588,
+      "step": 10957
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9219745875016918e-05,
+      "loss": 0.9072,
+      "step": 10958
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9216632678759694e-05,
+      "loss": 0.9922,
+      "step": 10959
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9213519501512477e-05,
+      "loss": 0.7275,
+      "step": 10960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.921040634335084e-05,
+      "loss": 0.8154,
+      "step": 10961
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9207293204350316e-05,
+      "loss": 0.8457,
+      "step": 10962
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9204180084586455e-05,
+      "loss": 0.9189,
+      "step": 10963
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9201066984134796e-05,
+      "loss": 0.7871,
+      "step": 10964
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9197953903070902e-05,
+      "loss": 0.9746,
+      "step": 10965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9194840841470303e-05,
+      "loss": 0.9121,
+      "step": 10966
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9191727799408552e-05,
+      "loss": 1.0508,
+      "step": 10967
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9188614776961184e-05,
+      "loss": 1.1699,
+      "step": 10968
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9185501774203754e-05,
+      "loss": 0.9717,
+      "step": 10969
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9182388791211797e-05,
+      "loss": 0.7705,
+      "step": 10970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9179275828060866e-05,
+      "loss": 0.8213,
+      "step": 10971
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.917616288482649e-05,
+      "loss": 1.1553,
+      "step": 10972
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.917304996158422e-05,
+      "loss": 1.0176,
+      "step": 10973
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.91699370584096e-05,
+      "loss": 0.7734,
+      "step": 10974
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.916682417537816e-05,
+      "loss": 0.8613,
+      "step": 10975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9163711312565448e-05,
+      "loss": 0.9902,
+      "step": 10976
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.916059847004701e-05,
+      "loss": 0.8613,
+      "step": 10977
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.915748564789837e-05,
+      "loss": 0.8994,
+      "step": 10978
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.915437284619508e-05,
+      "loss": 0.9355,
+      "step": 10979
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.915126006501267e-05,
+      "loss": 0.9131,
+      "step": 10980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.914814730442669e-05,
+      "loss": 1.0537,
+      "step": 10981
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.914503456451266e-05,
+      "loss": 0.918,
+      "step": 10982
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9141921845346137e-05,
+      "loss": 1.0967,
+      "step": 10983
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.913880914700264e-05,
+      "loss": 0.9355,
+      "step": 10984
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9135696469557715e-05,
+      "loss": 0.7817,
+      "step": 10985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9132583813086893e-05,
+      "loss": 0.8828,
+      "step": 10986
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9129471177665714e-05,
+      "loss": 0.9082,
+      "step": 10987
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9126358563369706e-05,
+      "loss": 0.9717,
+      "step": 10988
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9123245970274405e-05,
+      "loss": 0.8281,
+      "step": 10989
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.912013339845535e-05,
+      "loss": 1.0049,
+      "step": 10990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9117020847988066e-05,
+      "loss": 0.9805,
+      "step": 10991
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9113908318948097e-05,
+      "loss": 0.7681,
+      "step": 10992
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.911079581141096e-05,
+      "loss": 0.8496,
+      "step": 10993
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9107683325452197e-05,
+      "loss": 0.9502,
+      "step": 10994
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.910457086114733e-05,
+      "loss": 1.0332,
+      "step": 10995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.91014584185719e-05,
+      "loss": 0.8545,
+      "step": 10996
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.909834599780143e-05,
+      "loss": 0.915,
+      "step": 10997
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9095233598911454e-05,
+      "loss": 1.04,
+      "step": 10998
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9092121221977493e-05,
+      "loss": 1.0088,
+      "step": 10999
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9089008867075073e-05,
+      "loss": 0.8457,
+      "step": 11000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9085896534279742e-05,
+      "loss": 0.8604,
+      "step": 11001
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9082784223667008e-05,
+      "loss": 0.8613,
+      "step": 11002
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.907967193531241e-05,
+      "loss": 0.8467,
+      "step": 11003
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9076559669291456e-05,
+      "loss": 0.9482,
+      "step": 11004
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9073447425679695e-05,
+      "loss": 0.8984,
+      "step": 11005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.907033520455263e-05,
+      "loss": 0.8545,
+      "step": 11006
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9067223005985805e-05,
+      "loss": 0.998,
+      "step": 11007
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9064110830054727e-05,
+      "loss": 1.0352,
+      "step": 11008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.906099867683493e-05,
+      "loss": 0.5837,
+      "step": 11009
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.905788654640193e-05,
+      "loss": 1.1338,
+      "step": 11010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.905477443883126e-05,
+      "loss": 0.8867,
+      "step": 11011
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.905166235419843e-05,
+      "loss": 0.9912,
+      "step": 11012
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.904855029257897e-05,
+      "loss": 0.877,
+      "step": 11013
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.904543825404839e-05,
+      "loss": 0.9932,
+      "step": 11014
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.904232623868222e-05,
+      "loss": 0.8369,
+      "step": 11015
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9039214246555974e-05,
+      "loss": 1.0,
+      "step": 11016
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9036102277745174e-05,
+      "loss": 0.8369,
+      "step": 11017
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9032990332325335e-05,
+      "loss": 0.876,
+      "step": 11018
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.902987841037198e-05,
+      "loss": 1.002,
+      "step": 11019
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.902676651196062e-05,
+      "loss": 0.8867,
+      "step": 11020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.902365463716678e-05,
+      "loss": 0.834,
+      "step": 11021
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9020542786065962e-05,
+      "loss": 0.8491,
+      "step": 11022
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9017430958733697e-05,
+      "loss": 0.8506,
+      "step": 11023
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9014319155245487e-05,
+      "loss": 0.9746,
+      "step": 11024
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.901120737567686e-05,
+      "loss": 0.8352,
+      "step": 11025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9008095620103312e-05,
+      "loss": 1.0566,
+      "step": 11026
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9004983888600376e-05,
+      "loss": 0.958,
+      "step": 11027
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9001872181243548e-05,
+      "loss": 1.042,
+      "step": 11028
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.899876049810835e-05,
+      "loss": 0.9043,
+      "step": 11029
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.899564883927029e-05,
+      "loss": 1.0029,
+      "step": 11030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.899253720480488e-05,
+      "loss": 0.9541,
+      "step": 11031
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8989425594787626e-05,
+      "loss": 0.7773,
+      "step": 11032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.898631400929404e-05,
+      "loss": 0.9375,
+      "step": 11033
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.898320244839964e-05,
+      "loss": 0.8467,
+      "step": 11034
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8980090912179916e-05,
+      "loss": 0.8086,
+      "step": 11035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8976979400710393e-05,
+      "loss": 0.874,
+      "step": 11036
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8973867914066567e-05,
+      "loss": 0.9033,
+      "step": 11037
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8970756452323957e-05,
+      "loss": 0.9668,
+      "step": 11038
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.896764501555805e-05,
+      "loss": 1.0449,
+      "step": 11039
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8964533603844372e-05,
+      "loss": 0.8652,
+      "step": 11040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8961422217258413e-05,
+      "loss": 0.9902,
+      "step": 11041
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8958310855875685e-05,
+      "loss": 0.9072,
+      "step": 11042
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8955199519771683e-05,
+      "loss": 0.8599,
+      "step": 11043
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8952088209021922e-05,
+      "loss": 0.7881,
+      "step": 11044
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8948976923701896e-05,
+      "loss": 0.9199,
+      "step": 11045
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8945865663887113e-05,
+      "loss": 0.6924,
+      "step": 11046
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8942754429653067e-05,
+      "loss": 1.0596,
+      "step": 11047
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8939643221075262e-05,
+      "loss": 0.7451,
+      "step": 11048
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8936532038229197e-05,
+      "loss": 0.8555,
+      "step": 11049
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8933420881190373e-05,
+      "loss": 0.834,
+      "step": 11050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8930309750034283e-05,
+      "loss": 0.8799,
+      "step": 11051
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8927198644836437e-05,
+      "loss": 1.0,
+      "step": 11052
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.892408756567232e-05,
+      "loss": 0.8457,
+      "step": 11053
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8920976512617436e-05,
+      "loss": 0.9082,
+      "step": 11054
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8917865485747278e-05,
+      "loss": 0.8447,
+      "step": 11055
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8914754485137343e-05,
+      "loss": 1.0186,
+      "step": 11056
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8911643510863122e-05,
+      "loss": 0.8848,
+      "step": 11057
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8908532563000118e-05,
+      "loss": 1.0,
+      "step": 11058
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890542164162381e-05,
+      "loss": 0.9893,
+      "step": 11059
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890231074680971e-05,
+      "loss": 0.7432,
+      "step": 11060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889919987863329e-05,
+      "loss": 0.8789,
+      "step": 11061
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889608903717006e-05,
+      "loss": 0.9492,
+      "step": 11062
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892978222495498e-05,
+      "loss": 0.9395,
+      "step": 11063
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88898674346851e-05,
+      "loss": 0.7124,
+      "step": 11064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8886756673814354e-05,
+      "loss": 0.873,
+      "step": 11065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8883645939958753e-05,
+      "loss": 0.8789,
+      "step": 11066
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8880535233193777e-05,
+      "loss": 0.802,
+      "step": 11067
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8877424553594922e-05,
+      "loss": 0.9648,
+      "step": 11068
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8874313901237667e-05,
+      "loss": 0.8516,
+      "step": 11069
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887120327619751e-05,
+      "loss": 0.8906,
+      "step": 11070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8868092678549923e-05,
+      "loss": 0.7666,
+      "step": 11071
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8864982108370406e-05,
+      "loss": 0.9424,
+      "step": 11072
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8861871565734427e-05,
+      "loss": 1.0635,
+      "step": 11073
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8858761050717476e-05,
+      "loss": 0.9648,
+      "step": 11074
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8855650563395047e-05,
+      "loss": 0.9033,
+      "step": 11075
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.885254010384261e-05,
+      "loss": 1.0088,
+      "step": 11076
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8849429672135652e-05,
+      "loss": 0.8789,
+      "step": 11077
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8846319268349648e-05,
+      "loss": 0.8057,
+      "step": 11078
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8843208892560085e-05,
+      "loss": 0.8525,
+      "step": 11079
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8840098544842438e-05,
+      "loss": 0.8516,
+      "step": 11080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8836988225272193e-05,
+      "loss": 0.9102,
+      "step": 11081
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.883387793392482e-05,
+      "loss": 0.8623,
+      "step": 11082
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8830767670875792e-05,
+      "loss": 0.9893,
+      "step": 11083
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8827657436200607e-05,
+      "loss": 0.9395,
+      "step": 11084
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8824547229974718e-05,
+      "loss": 0.8105,
+      "step": 11085
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8821437052273622e-05,
+      "loss": 0.7832,
+      "step": 11086
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8818326903172773e-05,
+      "loss": 0.9893,
+      "step": 11087
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8815216782747662e-05,
+      "loss": 1.0049,
+      "step": 11088
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.881210669107375e-05,
+      "loss": 0.8335,
+      "step": 11089
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.880899662822652e-05,
+      "loss": 1.04,
+      "step": 11090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8805886594281437e-05,
+      "loss": 0.7988,
+      "step": 11091
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8802776589313974e-05,
+      "loss": 0.9795,
+      "step": 11092
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8799666613399603e-05,
+      "loss": 0.8926,
+      "step": 11093
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8796556666613796e-05,
+      "loss": 1.0439,
+      "step": 11094
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8793446749032012e-05,
+      "loss": 0.8389,
+      "step": 11095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8790336860729738e-05,
+      "loss": 1.0303,
+      "step": 11096
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.878722700178242e-05,
+      "loss": 0.8555,
+      "step": 11097
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8784117172265544e-05,
+      "loss": 0.9951,
+      "step": 11098
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8781007372254563e-05,
+      "loss": 0.7344,
+      "step": 11099
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8777897601824953e-05,
+      "loss": 0.7827,
+      "step": 11100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.877478786105217e-05,
+      "loss": 0.832,
+      "step": 11101
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8771678150011685e-05,
+      "loss": 1.1162,
+      "step": 11102
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8768568468778957e-05,
+      "loss": 0.9893,
+      "step": 11103
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.876545881742945e-05,
+      "loss": 0.873,
+      "step": 11104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8762349196038626e-05,
+      "loss": 0.8652,
+      "step": 11105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8759239604681952e-05,
+      "loss": 0.9531,
+      "step": 11106
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.875613004343488e-05,
+      "loss": 0.8594,
+      "step": 11107
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.875302051237288e-05,
+      "loss": 0.9678,
+      "step": 11108
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8749911011571397e-05,
+      "loss": 1.0137,
+      "step": 11109
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.87468015411059e-05,
+      "loss": 0.873,
+      "step": 11110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8743692101051843e-05,
+      "loss": 0.8389,
+      "step": 11111
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8740582691484686e-05,
+      "loss": 0.8662,
+      "step": 11112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8737473312479884e-05,
+      "loss": 0.9082,
+      "step": 11113
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8734363964112887e-05,
+      "loss": 0.7969,
+      "step": 11114
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.873125464645916e-05,
+      "loss": 0.8428,
+      "step": 11115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8728145359594147e-05,
+      "loss": 0.8545,
+      "step": 11116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.872503610359331e-05,
+      "loss": 0.917,
+      "step": 11117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8721926878532096e-05,
+      "loss": 0.8975,
+      "step": 11118
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.871881768448596e-05,
+      "loss": 0.8105,
+      "step": 11119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8715708521530345e-05,
+      "loss": 0.9629,
+      "step": 11120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8712599389740715e-05,
+      "loss": 0.9463,
+      "step": 11121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8709490289192505e-05,
+      "loss": 1.0107,
+      "step": 11122
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8706381219961174e-05,
+      "loss": 0.8955,
+      "step": 11123
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8703272182122164e-05,
+      "loss": 0.5947,
+      "step": 11124
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8700163175750927e-05,
+      "loss": 0.835,
+      "step": 11125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8697054200922905e-05,
+      "loss": 0.8584,
+      "step": 11126
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.869394525771355e-05,
+      "loss": 0.9404,
+      "step": 11127
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8690836346198292e-05,
+      "loss": 0.7231,
+      "step": 11128
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8687727466452595e-05,
+      "loss": 0.9131,
+      "step": 11129
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.868461861855189e-05,
+      "loss": 0.9082,
+      "step": 11130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8681509802571627e-05,
+      "loss": 0.9658,
+      "step": 11131
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8678401018587237e-05,
+      "loss": 0.8076,
+      "step": 11132
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8675292266674173e-05,
+      "loss": 0.9062,
+      "step": 11133
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8672183546907867e-05,
+      "loss": 0.916,
+      "step": 11134
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8669074859363764e-05,
+      "loss": 0.9424,
+      "step": 11135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8665966204117296e-05,
+      "loss": 1.0146,
+      "step": 11136
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8662857581243908e-05,
+      "loss": 0.8467,
+      "step": 11137
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8659748990819035e-05,
+      "loss": 1.085,
+      "step": 11138
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8656640432918112e-05,
+      "loss": 0.8018,
+      "step": 11139
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.865353190761657e-05,
+      "loss": 0.9736,
+      "step": 11140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8650423414989857e-05,
+      "loss": 0.9775,
+      "step": 11141
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.864731495511339e-05,
+      "loss": 1.0391,
+      "step": 11142
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.864420652806262e-05,
+      "loss": 1.0664,
+      "step": 11143
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8641098133912963e-05,
+      "loss": 0.9414,
+      "step": 11144
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.863798977273986e-05,
+      "loss": 0.894,
+      "step": 11145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.863488144461874e-05,
+      "loss": 1.1064,
+      "step": 11146
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8631773149625036e-05,
+      "loss": 0.6807,
+      "step": 11147
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8628664887834167e-05,
+      "loss": 0.7256,
+      "step": 11148
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8625556659321576e-05,
+      "loss": 0.7705,
+      "step": 11149
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8622448464162675e-05,
+      "loss": 0.8076,
+      "step": 11150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8619340302432903e-05,
+      "loss": 0.8232,
+      "step": 11151
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.861623217420768e-05,
+      "loss": 0.8457,
+      "step": 11152
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8613124079562434e-05,
+      "loss": 0.8525,
+      "step": 11153
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8610016018572587e-05,
+      "loss": 0.9033,
+      "step": 11154
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8606907991313567e-05,
+      "loss": 0.8945,
+      "step": 11155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8603799997860784e-05,
+      "loss": 0.708,
+      "step": 11156
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.860069203828968e-05,
+      "loss": 0.8564,
+      "step": 11157
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8597584112675656e-05,
+      "loss": 0.8926,
+      "step": 11158
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.859447622109415e-05,
+      "loss": 0.9971,
+      "step": 11159
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8591368363620564e-05,
+      "loss": 0.9941,
+      "step": 11160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8588260540330327e-05,
+      "loss": 0.9404,
+      "step": 11161
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.858515275129886e-05,
+      "loss": 0.6191,
+      "step": 11162
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8582044996601567e-05,
+      "loss": 0.8574,
+      "step": 11163
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.857893727631388e-05,
+      "loss": 0.8262,
+      "step": 11164
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.85758295905112e-05,
+      "loss": 0.7275,
+      "step": 11165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8572721939268943e-05,
+      "loss": 0.8369,
+      "step": 11166
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8569614322662538e-05,
+      "loss": 0.7666,
+      "step": 11167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.856650674076738e-05,
+      "loss": 0.7598,
+      "step": 11168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8563399193658894e-05,
+      "loss": 0.8916,
+      "step": 11169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8560291681412478e-05,
+      "loss": 0.834,
+      "step": 11170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8557184204103555e-05,
+      "loss": 0.9639,
+      "step": 11171
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.855407676180752e-05,
+      "loss": 0.8799,
+      "step": 11172
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.85509693545998e-05,
+      "loss": 0.8545,
+      "step": 11173
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8547861982555785e-05,
+      "loss": 0.9111,
+      "step": 11174
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8544754645750892e-05,
+      "loss": 0.8223,
+      "step": 11175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8541647344260523e-05,
+      "loss": 0.9502,
+      "step": 11176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8538540078160085e-05,
+      "loss": 0.9277,
+      "step": 11177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8535432847524978e-05,
+      "loss": 0.877,
+      "step": 11178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8532325652430614e-05,
+      "loss": 1.0449,
+      "step": 11179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8529218492952387e-05,
+      "loss": 1.0078,
+      "step": 11180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8526111369165705e-05,
+      "loss": 0.8867,
+      "step": 11181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.852300428114596e-05,
+      "loss": 0.8877,
+      "step": 11182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.851989722896856e-05,
+      "loss": 1.04,
+      "step": 11183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.85167902127089e-05,
+      "loss": 0.8555,
+      "step": 11184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8513683232442384e-05,
+      "loss": 1.084,
+      "step": 11185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.85105762882444e-05,
+      "loss": 0.7939,
+      "step": 11186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8507469380190356e-05,
+      "loss": 0.7754,
+      "step": 11187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8504362508355634e-05,
+      "loss": 0.9609,
+      "step": 11188
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.850125567281564e-05,
+      "loss": 0.8584,
+      "step": 11189
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8498148873645756e-05,
+      "loss": 1.0322,
+      "step": 11190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.849504211092139e-05,
+      "loss": 0.8125,
+      "step": 11191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.849193538471792e-05,
+      "loss": 0.8535,
+      "step": 11192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.848882869511075e-05,
+      "loss": 0.8262,
+      "step": 11193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8485722042175256e-05,
+      "loss": 0.8975,
+      "step": 11194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.848261542598684e-05,
+      "loss": 0.8325,
+      "step": 11195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8479508846620883e-05,
+      "loss": 0.9033,
+      "step": 11196
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8476402304152777e-05,
+      "loss": 0.8066,
+      "step": 11197
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.84732957986579e-05,
+      "loss": 0.7803,
+      "step": 11198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8470189330211652e-05,
+      "loss": 0.9355,
+      "step": 11199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8467082898889404e-05,
+      "loss": 0.6543,
+      "step": 11200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8463976504766546e-05,
+      "loss": 0.8281,
+      "step": 11201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8460870147918464e-05,
+      "loss": 0.9453,
+      "step": 11202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8457763828420533e-05,
+      "loss": 1.0498,
+      "step": 11203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8454657546348144e-05,
+      "loss": 0.8623,
+      "step": 11204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8451551301776667e-05,
+      "loss": 0.7627,
+      "step": 11205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8448445094781494e-05,
+      "loss": 0.9346,
+      "step": 11206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8445338925437985e-05,
+      "loss": 0.9609,
+      "step": 11207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8442232793821536e-05,
+      "loss": 0.8848,
+      "step": 11208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.843912670000751e-05,
+      "loss": 0.9375,
+      "step": 11209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8436020644071296e-05,
+      "loss": 0.8076,
+      "step": 11210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8432914626088255e-05,
+      "loss": 0.9883,
+      "step": 11211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8429808646133767e-05,
+      "loss": 1.0273,
+      "step": 11212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.842670270428321e-05,
+      "loss": 0.9395,
+      "step": 11213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.842359680061195e-05,
+      "loss": 0.9512,
+      "step": 11214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8420490935195355e-05,
+      "loss": 1.0322,
+      "step": 11215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8417385108108807e-05,
+      "loss": 1.0703,
+      "step": 11216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8414279319427664e-05,
+      "loss": 0.8701,
+      "step": 11217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.84111735692273e-05,
+      "loss": 0.9814,
+      "step": 11218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.840806785758308e-05,
+      "loss": 0.8066,
+      "step": 11219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8404962184570373e-05,
+      "loss": 0.9424,
+      "step": 11220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.840185655026454e-05,
+      "loss": 0.9043,
+      "step": 11221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8398750954740953e-05,
+      "loss": 0.9326,
+      "step": 11222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8395645398074965e-05,
+      "loss": 0.9766,
+      "step": 11223
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.839253988034195e-05,
+      "loss": 1.0693,
+      "step": 11224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.838943440161726e-05,
+      "loss": 0.8457,
+      "step": 11225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8386328961976267e-05,
+      "loss": 0.918,
+      "step": 11226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.838322356149432e-05,
+      "loss": 0.9297,
+      "step": 11227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8380118200246784e-05,
+      "loss": 0.8018,
+      "step": 11228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8377012878309012e-05,
+      "loss": 1.0879,
+      "step": 11229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.837390759575637e-05,
+      "loss": 0.918,
+      "step": 11230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8370802352664206e-05,
+      "loss": 0.9395,
+      "step": 11231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8367697149107877e-05,
+      "loss": 0.7764,
+      "step": 11232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8364591985162736e-05,
+      "loss": 0.8691,
+      "step": 11233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8361486860904144e-05,
+      "loss": 0.8008,
+      "step": 11234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.835838177640744e-05,
+      "loss": 0.875,
+      "step": 11235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.835527673174799e-05,
+      "loss": 0.9902,
+      "step": 11236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.835217172700113e-05,
+      "loss": 0.8896,
+      "step": 11237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.834906676224222e-05,
+      "loss": 0.7827,
+      "step": 11238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.83459618375466e-05,
+      "loss": 0.7695,
+      "step": 11239
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8342856952989627e-05,
+      "loss": 0.9531,
+      "step": 11240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8339752108646638e-05,
+      "loss": 0.9082,
+      "step": 11241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.833664730459299e-05,
+      "loss": 0.9355,
+      "step": 11242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.833354254090401e-05,
+      "loss": 0.8516,
+      "step": 11243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8330437817655057e-05,
+      "loss": 0.9121,
+      "step": 11244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.832733313492147e-05,
+      "loss": 0.9619,
+      "step": 11245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.832422849277858e-05,
+      "loss": 0.9268,
+      "step": 11246
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8321123891301748e-05,
+      "loss": 0.9756,
+      "step": 11247
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8318019330566292e-05,
+      "loss": 0.9619,
+      "step": 11248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8314914810647568e-05,
+      "loss": 0.9746,
+      "step": 11249
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8311810331620897e-05,
+      "loss": 0.9053,
+      "step": 11250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8308705893561626e-05,
+      "loss": 0.8877,
+      "step": 11251
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8305601496545095e-05,
+      "loss": 0.7246,
+      "step": 11252
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8302497140646627e-05,
+      "loss": 0.7266,
+      "step": 11253
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.829939282594157e-05,
+      "loss": 0.8096,
+      "step": 11254
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8296288552505243e-05,
+      "loss": 1.0,
+      "step": 11255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8293184320412983e-05,
+      "loss": 0.7539,
+      "step": 11256
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8290080129740116e-05,
+      "loss": 0.7871,
+      "step": 11257
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.828697598056198e-05,
+      "loss": 1.0283,
+      "step": 11258
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8283871872953897e-05,
+      "loss": 0.8867,
+      "step": 11259
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.82807678069912e-05,
+      "loss": 0.8916,
+      "step": 11260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.827766378274921e-05,
+      "loss": 0.8652,
+      "step": 11261
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8274559800303255e-05,
+      "loss": 0.8594,
+      "step": 11262
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8271455859728656e-05,
+      "loss": 0.8203,
+      "step": 11263
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8268351961100745e-05,
+      "loss": 0.8857,
+      "step": 11264
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8265248104494833e-05,
+      "loss": 0.8691,
+      "step": 11265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8262144289986256e-05,
+      "loss": 0.9189,
+      "step": 11266
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.825904051765032e-05,
+      "loss": 0.7021,
+      "step": 11267
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8255936787562355e-05,
+      "loss": 0.8613,
+      "step": 11268
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8252833099797672e-05,
+      "loss": 0.8809,
+      "step": 11269
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8249729454431592e-05,
+      "loss": 0.8887,
+      "step": 11270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.824662585153943e-05,
+      "loss": 0.8877,
+      "step": 11271
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8243522291196506e-05,
+      "loss": 0.8613,
+      "step": 11272
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8240418773478126e-05,
+      "loss": 0.873,
+      "step": 11273
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.823731529845961e-05,
+      "loss": 0.9775,
+      "step": 11274
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8234211866216264e-05,
+      "loss": 0.9365,
+      "step": 11275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.823110847682341e-05,
+      "loss": 0.7734,
+      "step": 11276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8228005130356344e-05,
+      "loss": 0.9854,
+      "step": 11277
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.822490182689039e-05,
+      "loss": 0.8516,
+      "step": 11278
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.822179856650084e-05,
+      "loss": 0.9043,
+      "step": 11279
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8218695349263016e-05,
+      "loss": 0.9297,
+      "step": 11280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.821559217525221e-05,
+      "loss": 1.0,
+      "step": 11281
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.821248904454374e-05,
+      "loss": 0.9443,
+      "step": 11282
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.82093859572129e-05,
+      "loss": 0.9229,
+      "step": 11283
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8206282913335003e-05,
+      "loss": 0.9229,
+      "step": 11284
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.820317991298534e-05,
+      "loss": 0.9658,
+      "step": 11285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.820007695623921e-05,
+      "loss": 0.8232,
+      "step": 11286
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8196974043171932e-05,
+      "loss": 0.9424,
+      "step": 11287
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8193871173858784e-05,
+      "loss": 0.9512,
+      "step": 11288
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.819076834837507e-05,
+      "loss": 0.8232,
+      "step": 11289
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.818766556679609e-05,
+      "loss": 0.9883,
+      "step": 11290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.818456282919714e-05,
+      "loss": 0.8779,
+      "step": 11291
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8181460135653505e-05,
+      "loss": 0.9258,
+      "step": 11292
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.817835748624049e-05,
+      "loss": 0.8857,
+      "step": 11293
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8175254881033374e-05,
+      "loss": 1.0996,
+      "step": 11294
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8172152320107463e-05,
+      "loss": 0.8457,
+      "step": 11295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8169049803538035e-05,
+      "loss": 0.8818,
+      "step": 11296
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8165947331400387e-05,
+      "loss": 0.917,
+      "step": 11297
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.81628449037698e-05,
+      "loss": 0.9766,
+      "step": 11298
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.815974252072157e-05,
+      "loss": 1.0098,
+      "step": 11299
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.815664018233097e-05,
+      "loss": 1.0127,
+      "step": 11300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.81535378886733e-05,
+      "loss": 0.793,
+      "step": 11301
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8150435639823827e-05,
+      "loss": 0.8555,
+      "step": 11302
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8147333435857847e-05,
+      "loss": 0.8076,
+      "step": 11303
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.814423127685063e-05,
+      "loss": 0.9141,
+      "step": 11304
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8141129162877467e-05,
+      "loss": 0.9473,
+      "step": 11305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8138027094013632e-05,
+      "loss": 1.0781,
+      "step": 11306
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8134925070334407e-05,
+      "loss": 0.8672,
+      "step": 11307
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8131823091915056e-05,
+      "loss": 1.041,
+      "step": 11308
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8128721158830874e-05,
+      "loss": 1.04,
+      "step": 11309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.812561927115712e-05,
+      "loss": 0.8555,
+      "step": 11310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.812251742896908e-05,
+      "loss": 0.7246,
+      "step": 11311
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8119415632342014e-05,
+      "loss": 1.0166,
+      "step": 11312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8116313881351206e-05,
+      "loss": 0.8955,
+      "step": 11313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8113212176071916e-05,
+      "loss": 0.7715,
+      "step": 11314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.811011051657942e-05,
+      "loss": 0.792,
+      "step": 11315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.810700890294898e-05,
+      "loss": 0.8188,
+      "step": 11316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.810390733525587e-05,
+      "loss": 0.8994,
+      "step": 11317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8100805813575354e-05,
+      "loss": 0.9453,
+      "step": 11318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8097704337982696e-05,
+      "loss": 0.9707,
+      "step": 11319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8094602908553153e-05,
+      "loss": 0.918,
+      "step": 11320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8091501525362e-05,
+      "loss": 0.8086,
+      "step": 11321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.808840018848449e-05,
+      "loss": 0.792,
+      "step": 11322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8085298897995886e-05,
+      "loss": 0.7744,
+      "step": 11323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8082197653971444e-05,
+      "loss": 0.5898,
+      "step": 11324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.807909645648643e-05,
+      "loss": 0.7969,
+      "step": 11325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.807599530561609e-05,
+      "loss": 0.8857,
+      "step": 11326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8072894201435695e-05,
+      "loss": 1.0322,
+      "step": 11327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8069793144020482e-05,
+      "loss": 0.8447,
+      "step": 11328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8066692133445717e-05,
+      "loss": 0.8125,
+      "step": 11329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8063591169786646e-05,
+      "loss": 0.9141,
+      "step": 11330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8060490253118524e-05,
+      "loss": 0.8828,
+      "step": 11331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8057389383516597e-05,
+      "loss": 0.8232,
+      "step": 11332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.805428856105611e-05,
+      "loss": 0.9312,
+      "step": 11333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8051187785812325e-05,
+      "loss": 0.9385,
+      "step": 11334
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8048087057860484e-05,
+      "loss": 0.7871,
+      "step": 11335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8044986377275826e-05,
+      "loss": 0.9023,
+      "step": 11336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8041885744133602e-05,
+      "loss": 0.8916,
+      "step": 11337
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.803878515850905e-05,
+      "loss": 0.7227,
+      "step": 11338
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8035684620477418e-05,
+      "loss": 1.0244,
+      "step": 11339
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.803258413011394e-05,
+      "loss": 0.7246,
+      "step": 11340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8029483687493862e-05,
+      "loss": 0.9062,
+      "step": 11341
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8026383292692415e-05,
+      "loss": 0.9766,
+      "step": 11342
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8023282945784847e-05,
+      "loss": 0.7949,
+      "step": 11343
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8020182646846385e-05,
+      "loss": 0.8467,
+      "step": 11344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.801708239595227e-05,
+      "loss": 0.7344,
+      "step": 11345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.801398219317773e-05,
+      "loss": 0.8418,
+      "step": 11346
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.801088203859801e-05,
+      "loss": 0.9746,
+      "step": 11347
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8007781932288323e-05,
+      "loss": 0.8594,
+      "step": 11348
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8004681874323914e-05,
+      "loss": 0.9521,
+      "step": 11349
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8001581864780007e-05,
+      "loss": 1.0234,
+      "step": 11350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7998481903731834e-05,
+      "loss": 0.7383,
+      "step": 11351
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7995381991254613e-05,
+      "loss": 0.8906,
+      "step": 11352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7992282127423582e-05,
+      "loss": 0.9297,
+      "step": 11353
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7989182312313954e-05,
+      "loss": 0.9355,
+      "step": 11354
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7986082546000966e-05,
+      "loss": 1.0049,
+      "step": 11355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7982982828559824e-05,
+      "loss": 0.8379,
+      "step": 11356
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7979883160065758e-05,
+      "loss": 0.9951,
+      "step": 11357
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7976783540593988e-05,
+      "loss": 1.0723,
+      "step": 11358
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7973683970219734e-05,
+      "loss": 0.8926,
+      "step": 11359
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7970584449018206e-05,
+      "loss": 0.8965,
+      "step": 11360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7967484977064627e-05,
+      "loss": 0.9805,
+      "step": 11361
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.796438555443421e-05,
+      "loss": 1.0264,
+      "step": 11362
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7961286181202172e-05,
+      "loss": 0.9688,
+      "step": 11363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.795818685744372e-05,
+      "loss": 0.9043,
+      "step": 11364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7955087583234068e-05,
+      "loss": 0.9355,
+      "step": 11365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7951988358648425e-05,
+      "loss": 1.1133,
+      "step": 11366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7948889183762002e-05,
+      "loss": 1.0898,
+      "step": 11367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7945790058650007e-05,
+      "loss": 0.9346,
+      "step": 11368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7942690983387653e-05,
+      "loss": 0.9062,
+      "step": 11369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.793959195805013e-05,
+      "loss": 0.8887,
+      "step": 11370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.793649298271265e-05,
+      "loss": 0.9561,
+      "step": 11371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7933394057450423e-05,
+      "loss": 1.0059,
+      "step": 11372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7930295182338638e-05,
+      "loss": 0.9492,
+      "step": 11373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7927196357452507e-05,
+      "loss": 1.0039,
+      "step": 11374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7924097582867222e-05,
+      "loss": 0.9033,
+      "step": 11375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7920998858657988e-05,
+      "loss": 0.877,
+      "step": 11376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.791790018489999e-05,
+      "loss": 0.8467,
+      "step": 11377
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7914801561668438e-05,
+      "loss": 0.7715,
+      "step": 11378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7911702989038514e-05,
+      "loss": 0.8301,
+      "step": 11379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7908604467085423e-05,
+      "loss": 0.7866,
+      "step": 11380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7905505995884348e-05,
+      "loss": 0.9463,
+      "step": 11381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7902407575510484e-05,
+      "loss": 0.8101,
+      "step": 11382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7899309206039017e-05,
+      "loss": 0.8828,
+      "step": 11383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.789621088754514e-05,
+      "loss": 0.8789,
+      "step": 11384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7893112620104035e-05,
+      "loss": 0.7339,
+      "step": 11385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7890014403790897e-05,
+      "loss": 1.0,
+      "step": 11386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7886916238680898e-05,
+      "loss": 0.8252,
+      "step": 11387
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.788381812484923e-05,
+      "loss": 0.957,
+      "step": 11388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7880720062371072e-05,
+      "loss": 0.9697,
+      "step": 11389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7877622051321604e-05,
+      "loss": 0.9902,
+      "step": 11390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.787452409177601e-05,
+      "loss": 0.7246,
+      "step": 11391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7871426183809464e-05,
+      "loss": 0.9951,
+      "step": 11392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7868328327497144e-05,
+      "loss": 0.7715,
+      "step": 11393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7865230522914226e-05,
+      "loss": 0.8594,
+      "step": 11394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.786213277013589e-05,
+      "loss": 0.8193,
+      "step": 11395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7859035069237302e-05,
+      "loss": 0.8916,
+      "step": 11396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.785593742029363e-05,
+      "loss": 0.9004,
+      "step": 11397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.785283982338006e-05,
+      "loss": 0.9023,
+      "step": 11398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7849742278571746e-05,
+      "loss": 0.9404,
+      "step": 11399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.784664478594387e-05,
+      "loss": 0.9834,
+      "step": 11400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7843547345571585e-05,
+      "loss": 0.7676,
+      "step": 11401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.784044995753007e-05,
+      "loss": 0.8936,
+      "step": 11402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7837352621894478e-05,
+      "loss": 0.9609,
+      "step": 11403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7834255338739985e-05,
+      "loss": 0.9355,
+      "step": 11404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7831158108141738e-05,
+      "loss": 1.0674,
+      "step": 11405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7828060930174913e-05,
+      "loss": 0.9258,
+      "step": 11406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7824963804914656e-05,
+      "loss": 1.0029,
+      "step": 11407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7821866732436135e-05,
+      "loss": 0.7085,
+      "step": 11408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.78187697128145e-05,
+      "loss": 0.9072,
+      "step": 11409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7815672746124912e-05,
+      "loss": 0.9443,
+      "step": 11410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7812575832442518e-05,
+      "loss": 0.8008,
+      "step": 11411
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.780947897184248e-05,
+      "loss": 0.9316,
+      "step": 11412
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7806382164399944e-05,
+      "loss": 0.7852,
+      "step": 11413
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7803285410190062e-05,
+      "loss": 0.7803,
+      "step": 11414
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.780018870928798e-05,
+      "loss": 0.9932,
+      "step": 11415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7797092061768842e-05,
+      "loss": 0.9512,
+      "step": 11416
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7793995467707808e-05,
+      "loss": 1.0059,
+      "step": 11417
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7790898927180022e-05,
+      "loss": 0.9141,
+      "step": 11418
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7787802440260612e-05,
+      "loss": 0.6958,
+      "step": 11419
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.778470600702474e-05,
+      "loss": 0.9561,
+      "step": 11420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7781609627547534e-05,
+      "loss": 0.9307,
+      "step": 11421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7778513301904143e-05,
+      "loss": 0.8296,
+      "step": 11422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7775417030169697e-05,
+      "loss": 1.0586,
+      "step": 11423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.777232081241934e-05,
+      "loss": 0.812,
+      "step": 11424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.77692246487282e-05,
+      "loss": 0.749,
+      "step": 11425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7766128539171422e-05,
+      "loss": 0.9736,
+      "step": 11426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7763032483824132e-05,
+      "loss": 0.9395,
+      "step": 11427
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.775993648276147e-05,
+      "loss": 0.8838,
+      "step": 11428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7756840536058555e-05,
+      "loss": 0.7998,
+      "step": 11429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.775374464379053e-05,
+      "loss": 1.0361,
+      "step": 11430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.775064880603251e-05,
+      "loss": 0.9795,
+      "step": 11431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7747553022859634e-05,
+      "loss": 0.8672,
+      "step": 11432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7744457294347017e-05,
+      "loss": 0.9297,
+      "step": 11433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7741361620569794e-05,
+      "loss": 0.9248,
+      "step": 11434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7738266001603077e-05,
+      "loss": 0.7451,
+      "step": 11435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7735170437521994e-05,
+      "loss": 0.8945,
+      "step": 11436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7732074928401664e-05,
+      "loss": 0.9717,
+      "step": 11437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.772897947431721e-05,
+      "loss": 0.8887,
+      "step": 11438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.772588407534374e-05,
+      "loss": 0.8965,
+      "step": 11439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7722788731556376e-05,
+      "loss": 0.8301,
+      "step": 11440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7719693443030232e-05,
+      "loss": 1.1367,
+      "step": 11441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7716598209840426e-05,
+      "loss": 0.8916,
+      "step": 11442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.771350303206206e-05,
+      "loss": 0.9414,
+      "step": 11443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7710407909770257e-05,
+      "loss": 0.8975,
+      "step": 11444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7707312843040117e-05,
+      "loss": 1.0742,
+      "step": 11445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7704217831946754e-05,
+      "loss": 0.9316,
+      "step": 11446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7701122876565272e-05,
+      "loss": 1.0859,
+      "step": 11447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.769802797697078e-05,
+      "loss": 0.9863,
+      "step": 11448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.769493313323837e-05,
+      "loss": 1.0859,
+      "step": 11449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7691838345443163e-05,
+      "loss": 0.9395,
+      "step": 11450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7688743613660246e-05,
+      "loss": 0.7681,
+      "step": 11451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7685648937964726e-05,
+      "loss": 0.7983,
+      "step": 11452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.76825543184317e-05,
+      "loss": 0.7158,
+      "step": 11453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.767945975513627e-05,
+      "loss": 0.9541,
+      "step": 11454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7676365248153518e-05,
+      "loss": 0.5728,
+      "step": 11455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7673270797558556e-05,
+      "loss": 0.7954,
+      "step": 11456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.767017640342646e-05,
+      "loss": 0.8984,
+      "step": 11457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7667082065832336e-05,
+      "loss": 0.7246,
+      "step": 11458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7663987784851273e-05,
+      "loss": 1.0059,
+      "step": 11459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.766089356055835e-05,
+      "loss": 1.0137,
+      "step": 11460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.765779939302867e-05,
+      "loss": 0.8662,
+      "step": 11461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7654705282337304e-05,
+      "loss": 0.8896,
+      "step": 11462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7651611228559344e-05,
+      "loss": 0.9082,
+      "step": 11463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7648517231769873e-05,
+      "loss": 0.7686,
+      "step": 11464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.764542329204398e-05,
+      "loss": 1.042,
+      "step": 11465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7642329409456728e-05,
+      "loss": 0.7383,
+      "step": 11466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7639235584083215e-05,
+      "loss": 1.0117,
+      "step": 11467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7636141815998508e-05,
+      "loss": 0.918,
+      "step": 11468
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.763304810527769e-05,
+      "loss": 0.7969,
+      "step": 11469
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7629954451995832e-05,
+      "loss": 0.9502,
+      "step": 11470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.762686085622801e-05,
+      "loss": 1.0781,
+      "step": 11471
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7623767318049293e-05,
+      "loss": 0.8638,
+      "step": 11472
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7620673837534762e-05,
+      "loss": 0.8457,
+      "step": 11473
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7617580414759472e-05,
+      "loss": 0.9141,
+      "step": 11474
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7614487049798507e-05,
+      "loss": 0.8516,
+      "step": 11475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7611393742726918e-05,
+      "loss": 0.8945,
+      "step": 11476
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7608300493619782e-05,
+      "loss": 1.0908,
+      "step": 11477
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7605207302552157e-05,
+      "loss": 0.9551,
+      "step": 11478
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.760211416959911e-05,
+      "loss": 1.0391,
+      "step": 11479
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.75990210948357e-05,
+      "loss": 0.8828,
+      "step": 11480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7595928078336987e-05,
+      "loss": 0.8198,
+      "step": 11481
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.759283512017803e-05,
+      "loss": 0.9336,
+      "step": 11482
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7589742220433884e-05,
+      "loss": 0.9141,
+      "step": 11483
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7586649379179604e-05,
+      "loss": 0.7241,
+      "step": 11484
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7583556596490253e-05,
+      "loss": 0.9502,
+      "step": 11485
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.758046387244087e-05,
+      "loss": 0.8906,
+      "step": 11486
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7577371207106518e-05,
+      "loss": 0.8633,
+      "step": 11487
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7574278600562235e-05,
+      "loss": 0.8574,
+      "step": 11488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7571186052883086e-05,
+      "loss": 0.8965,
+      "step": 11489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.75680935641441e-05,
+      "loss": 0.8848,
+      "step": 11490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.756500113442034e-05,
+      "loss": 0.9561,
+      "step": 11491
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.756190876378683e-05,
+      "loss": 0.8066,
+      "step": 11492
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7558816452318635e-05,
+      "loss": 0.834,
+      "step": 11493
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.755572420009078e-05,
+      "loss": 0.8945,
+      "step": 11494
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7552632007178312e-05,
+      "loss": 0.9932,
+      "step": 11495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7549539873656265e-05,
+      "loss": 1.0039,
+      "step": 11496
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7546447799599685e-05,
+      "loss": 0.918,
+      "step": 11497
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7543355785083592e-05,
+      "loss": 1.0137,
+      "step": 11498
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.754026383018304e-05,
+      "loss": 0.8115,
+      "step": 11499
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7537171934973035e-05,
+      "loss": 1.0332,
+      "step": 11500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7534080099528635e-05,
+      "loss": 0.9062,
+      "step": 11501
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7530988323924856e-05,
+      "loss": 0.9766,
+      "step": 11502
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7527896608236735e-05,
+      "loss": 0.8291,
+      "step": 11503
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.752480495253929e-05,
+      "loss": 0.9141,
+      "step": 11504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7521713356907552e-05,
+      "loss": 0.9941,
+      "step": 11505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7518621821416536e-05,
+      "loss": 1.0645,
+      "step": 11506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7515530346141275e-05,
+      "loss": 1.0537,
+      "step": 11507
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7512438931156786e-05,
+      "loss": 0.9756,
+      "step": 11508
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7509347576538092e-05,
+      "loss": 0.8525,
+      "step": 11509
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.75062562823602e-05,
+      "loss": 0.6855,
+      "step": 11510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.750316504869814e-05,
+      "loss": 0.8281,
+      "step": 11511
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.750007387562692e-05,
+      "loss": 0.7588,
+      "step": 11512
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7496982763221554e-05,
+      "loss": 0.6377,
+      "step": 11513
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7493891711557055e-05,
+      "loss": 0.7695,
+      "step": 11514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7490800720708438e-05,
+      "loss": 0.9395,
+      "step": 11515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.74877097907507e-05,
+      "loss": 0.8574,
+      "step": 11516
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7484618921758866e-05,
+      "loss": 0.9639,
+      "step": 11517
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7481528113807925e-05,
+      "loss": 0.833,
+      "step": 11518
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.74784373669729e-05,
+      "loss": 0.8564,
+      "step": 11519
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7475346681328774e-05,
+      "loss": 0.7666,
+      "step": 11520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7472256056950567e-05,
+      "loss": 1.0088,
+      "step": 11521
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7469165493913266e-05,
+      "loss": 0.8643,
+      "step": 11522
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7466074992291882e-05,
+      "loss": 0.8848,
+      "step": 11523
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.74629845521614e-05,
+      "loss": 0.8247,
+      "step": 11524
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7459894173596826e-05,
+      "loss": 0.8154,
+      "step": 11525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7456803856673148e-05,
+      "loss": 0.9678,
+      "step": 11526
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7453713601465364e-05,
+      "loss": 1.0801,
+      "step": 11527
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7450623408048458e-05,
+      "loss": 1.0898,
+      "step": 11528
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7447533276497428e-05,
+      "loss": 0.8652,
+      "step": 11529
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7444443206887255e-05,
+      "loss": 0.8291,
+      "step": 11530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7441353199292934e-05,
+      "loss": 1.0801,
+      "step": 11531
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7438263253789442e-05,
+      "loss": 0.9824,
+      "step": 11532
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.743517337045177e-05,
+      "loss": 0.9727,
+      "step": 11533
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7432083549354896e-05,
+      "loss": 1.1084,
+      "step": 11534
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7428993790573806e-05,
+      "loss": 0.8672,
+      "step": 11535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.742590409418347e-05,
+      "loss": 0.8154,
+      "step": 11536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7422814460258876e-05,
+      "loss": 0.9043,
+      "step": 11537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.741972488887499e-05,
+      "loss": 0.916,
+      "step": 11538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7416635380106802e-05,
+      "loss": 0.6021,
+      "step": 11539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7413545934029267e-05,
+      "loss": 0.9404,
+      "step": 11540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.741045655071737e-05,
+      "loss": 1.0059,
+      "step": 11541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7407367230246074e-05,
+      "loss": 0.8594,
+      "step": 11542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7404277972690352e-05,
+      "loss": 0.8623,
+      "step": 11543
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7401188778125173e-05,
+      "loss": 0.9775,
+      "step": 11544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7398099646625493e-05,
+      "loss": 0.9639,
+      "step": 11545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7395010578266292e-05,
+      "loss": 0.8262,
+      "step": 11546
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7391921573122515e-05,
+      "loss": 0.8174,
+      "step": 11547
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.738883263126914e-05,
+      "loss": 1.0068,
+      "step": 11548
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7385743752781107e-05,
+      "loss": 0.9395,
+      "step": 11549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.738265493773339e-05,
+      "loss": 0.7021,
+      "step": 11550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.737956618620094e-05,
+      "loss": 0.9727,
+      "step": 11551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7376477498258714e-05,
+      "loss": 0.9844,
+      "step": 11552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.737338887398166e-05,
+      "loss": 0.8252,
+      "step": 11553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7370300313444737e-05,
+      "loss": 0.939,
+      "step": 11554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7367211816722892e-05,
+      "loss": 0.8828,
+      "step": 11555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.736412338389107e-05,
+      "loss": 0.8955,
+      "step": 11556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7361035015024224e-05,
+      "loss": 0.875,
+      "step": 11557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.73579467101973e-05,
+      "loss": 0.9336,
+      "step": 11558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.735485846948523e-05,
+      "loss": 0.8115,
+      "step": 11559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7351770292962977e-05,
+      "loss": 1.0635,
+      "step": 11560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7348682180705463e-05,
+      "loss": 0.8223,
+      "step": 11561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7345594132787643e-05,
+      "loss": 1.0176,
+      "step": 11562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7342506149284436e-05,
+      "loss": 0.9072,
+      "step": 11563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.73394182302708e-05,
+      "loss": 0.8828,
+      "step": 11564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7336330375821652e-05,
+      "loss": 0.792,
+      "step": 11565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.733324258601194e-05,
+      "loss": 1.0029,
+      "step": 11566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7330154860916576e-05,
+      "loss": 0.9473,
+      "step": 11567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7327067200610513e-05,
+      "loss": 0.916,
+      "step": 11568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.732397960516866e-05,
+      "loss": 0.96,
+      "step": 11569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7320892074665958e-05,
+      "loss": 0.8652,
+      "step": 11570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7317804609177324e-05,
+      "loss": 0.8125,
+      "step": 11571
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7314717208777686e-05,
+      "loss": 0.9229,
+      "step": 11572
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.731162987354196e-05,
+      "loss": 0.8457,
+      "step": 11573
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7308542603545073e-05,
+      "loss": 1.0986,
+      "step": 11574
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7305455398861942e-05,
+      "loss": 0.7803,
+      "step": 11575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7302368259567485e-05,
+      "loss": 0.9385,
+      "step": 11576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7299281185736612e-05,
+      "loss": 0.9678,
+      "step": 11577
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.729619417744425e-05,
+      "loss": 0.7822,
+      "step": 11578
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.72931072347653e-05,
+      "loss": 1.0303,
+      "step": 11579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7290020357774676e-05,
+      "loss": 0.8276,
+      "step": 11580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7286933546547287e-05,
+      "loss": 0.8281,
+      "step": 11581
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7283846801158044e-05,
+      "loss": 0.9805,
+      "step": 11582
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7280760121681842e-05,
+      "loss": 0.7383,
+      "step": 11583
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7277673508193604e-05,
+      "loss": 0.8867,
+      "step": 11584
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7274586960768222e-05,
+      "loss": 0.8623,
+      "step": 11585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7271500479480603e-05,
+      "loss": 0.9141,
+      "step": 11586
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7268414064405638e-05,
+      "loss": 1.0195,
+      "step": 11587
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7265327715618238e-05,
+      "loss": 0.8047,
+      "step": 11588
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7262241433193284e-05,
+      "loss": 0.8047,
+      "step": 11589
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7259155217205685e-05,
+      "loss": 1.0029,
+      "step": 11590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7256069067730324e-05,
+      "loss": 1.0127,
+      "step": 11591
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.72529829848421e-05,
+      "loss": 0.9873,
+      "step": 11592
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7249896968615895e-05,
+      "loss": 0.9502,
+      "step": 11593
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.724681101912661e-05,
+      "loss": 0.9258,
+      "step": 11594
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7243725136449117e-05,
+      "loss": 0.7607,
+      "step": 11595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7240639320658317e-05,
+      "loss": 0.8398,
+      "step": 11596
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.723755357182908e-05,
+      "loss": 0.8408,
+      "step": 11597
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.72344678900363e-05,
+      "loss": 0.9414,
+      "step": 11598
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7231382275354846e-05,
+      "loss": 0.791,
+      "step": 11599
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7228296727859604e-05,
+      "loss": 0.9121,
+      "step": 11600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7225211247625443e-05,
+      "loss": 0.5913,
+      "step": 11601
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7222125834727253e-05,
+      "loss": 0.7979,
+      "step": 11602
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7219040489239896e-05,
+      "loss": 0.9639,
+      "step": 11603
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.721595521123825e-05,
+      "loss": 0.9551,
+      "step": 11604
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.721287000079718e-05,
+      "loss": 0.918,
+      "step": 11605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7209784857991563e-05,
+      "loss": 0.9814,
+      "step": 11606
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7206699782896256e-05,
+      "loss": 0.8604,
+      "step": 11607
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7203614775586136e-05,
+      "loss": 0.9033,
+      "step": 11608
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7200529836136056e-05,
+      "loss": 0.9307,
+      "step": 11609
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.719744496462089e-05,
+      "loss": 0.8989,
+      "step": 11610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7194360161115485e-05,
+      "loss": 0.7646,
+      "step": 11611
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7191275425694715e-05,
+      "loss": 0.6772,
+      "step": 11612
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7188190758433425e-05,
+      "loss": 0.9346,
+      "step": 11613
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.718510615940648e-05,
+      "loss": 1.1016,
+      "step": 11614
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7182021628688728e-05,
+      "loss": 0.8301,
+      "step": 11615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7178937166355028e-05,
+      "loss": 0.8945,
+      "step": 11616
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.717585277248022e-05,
+      "loss": 0.7505,
+      "step": 11617
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7172768447139164e-05,
+      "loss": 0.7285,
+      "step": 11618
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.71696841904067e-05,
+      "loss": 1.0342,
+      "step": 11619
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.716660000235768e-05,
+      "loss": 0.7566,
+      "step": 11620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7163515883066942e-05,
+      "loss": 0.9502,
+      "step": 11621
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7160431832609336e-05,
+      "loss": 0.6147,
+      "step": 11622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7157347851059696e-05,
+      "loss": 0.7007,
+      "step": 11623
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7154263938492866e-05,
+      "loss": 0.7612,
+      "step": 11624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7151180094983677e-05,
+      "loss": 1.002,
+      "step": 11625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7148096320606973e-05,
+      "loss": 0.9629,
+      "step": 11626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.714501261543758e-05,
+      "loss": 0.8828,
+      "step": 11627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.714192897955034e-05,
+      "loss": 0.9609,
+      "step": 11628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7138845413020074e-05,
+      "loss": 0.8643,
+      "step": 11629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7135761915921615e-05,
+      "loss": 0.9404,
+      "step": 11630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7132678488329794e-05,
+      "loss": 0.9365,
+      "step": 11631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.712959513031943e-05,
+      "loss": 0.877,
+      "step": 11632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7126511841965355e-05,
+      "loss": 0.7852,
+      "step": 11633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7123428623342384e-05,
+      "loss": 0.8467,
+      "step": 11634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7120345474525342e-05,
+      "loss": 0.9893,
+      "step": 11635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7117262395589044e-05,
+      "loss": 0.9141,
+      "step": 11636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7114179386608316e-05,
+      "loss": 0.9277,
+      "step": 11637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.711109644765796e-05,
+      "loss": 0.96,
+      "step": 11638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7108013578812804e-05,
+      "loss": 0.8213,
+      "step": 11639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.710493078014765e-05,
+      "loss": 0.9609,
+      "step": 11640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7101848051737312e-05,
+      "loss": 1.0508,
+      "step": 11641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7098765393656598e-05,
+      "loss": 0.7485,
+      "step": 11642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7095682805980317e-05,
+      "loss": 0.8164,
+      "step": 11643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.709260028878327e-05,
+      "loss": 0.8154,
+      "step": 11644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7089517842140268e-05,
+      "loss": 0.8184,
+      "step": 11645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7086435466126106e-05,
+      "loss": 0.8564,
+      "step": 11646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7083353160815586e-05,
+      "loss": 0.8369,
+      "step": 11647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.708027092628351e-05,
+      "loss": 0.8652,
+      "step": 11648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7077188762604674e-05,
+      "loss": 0.8574,
+      "step": 11649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7074106669853864e-05,
+      "loss": 0.9717,
+      "step": 11650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7071024648105886e-05,
+      "loss": 0.9297,
+      "step": 11651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.706794269743552e-05,
+      "loss": 0.9805,
+      "step": 11652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.706486081791757e-05,
+      "loss": 0.9727,
+      "step": 11653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.706177900962681e-05,
+      "loss": 0.8994,
+      "step": 11654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7058697272638034e-05,
+      "loss": 0.9824,
+      "step": 11655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7055615607026024e-05,
+      "loss": 0.9648,
+      "step": 11656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.705253401286557e-05,
+      "loss": 1.0273,
+      "step": 11657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.704945249023144e-05,
+      "loss": 0.8975,
+      "step": 11658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7046371039198422e-05,
+      "loss": 0.7568,
+      "step": 11659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7043289659841295e-05,
+      "loss": 0.8994,
+      "step": 11660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.704020835223483e-05,
+      "loss": 0.915,
+      "step": 11661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7037127116453807e-05,
+      "loss": 0.9863,
+      "step": 11662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7034045952572995e-05,
+      "loss": 0.9902,
+      "step": 11663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7030964860667164e-05,
+      "loss": 0.8843,
+      "step": 11664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7027883840811087e-05,
+      "loss": 1.0918,
+      "step": 11665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7024802893079517e-05,
+      "loss": 0.9902,
+      "step": 11666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7021722017547236e-05,
+      "loss": 1.0127,
+      "step": 11667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.701864121428901e-05,
+      "loss": 0.8926,
+      "step": 11668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.701556048337959e-05,
+      "loss": 0.9414,
+      "step": 11669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7012479824893738e-05,
+      "loss": 0.998,
+      "step": 11670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7009399238906223e-05,
+      "loss": 0.9697,
+      "step": 11671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7006318725491782e-05,
+      "loss": 0.7783,
+      "step": 11672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7003238284725193e-05,
+      "loss": 1.0127,
+      "step": 11673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.700015791668119e-05,
+      "loss": 0.8066,
+      "step": 11674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6997077621434535e-05,
+      "loss": 1.0977,
+      "step": 11675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.699399739905997e-05,
+      "loss": 0.9707,
+      "step": 11676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6990917249632255e-05,
+      "loss": 0.7651,
+      "step": 11677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6987837173226123e-05,
+      "loss": 0.9834,
+      "step": 11678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.698475716991633e-05,
+      "loss": 0.9131,
+      "step": 11679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6981677239777606e-05,
+      "loss": 0.9023,
+      "step": 11680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.697859738288471e-05,
+      "loss": 1.0371,
+      "step": 11681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6975517599312358e-05,
+      "loss": 0.9453,
+      "step": 11682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6972437889135304e-05,
+      "loss": 0.8525,
+      "step": 11683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.696935825242828e-05,
+      "loss": 0.9404,
+      "step": 11684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6966278689266017e-05,
+      "loss": 0.9111,
+      "step": 11685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6963199199723245e-05,
+      "loss": 0.9941,
+      "step": 11686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6960119783874704e-05,
+      "loss": 0.7695,
+      "step": 11687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.695704044179511e-05,
+      "loss": 0.9336,
+      "step": 11688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6953961173559203e-05,
+      "loss": 0.7085,
+      "step": 11689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.695088197924169e-05,
+      "loss": 0.8711,
+      "step": 11690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6947802858917314e-05,
+      "loss": 1.0107,
+      "step": 11691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.694472381266078e-05,
+      "loss": 1.0332,
+      "step": 11692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6941644840546817e-05,
+      "loss": 0.7939,
+      "step": 11693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6938565942650134e-05,
+      "loss": 0.8008,
+      "step": 11694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.693548711904546e-05,
+      "loss": 0.9658,
+      "step": 11695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.69324083698075e-05,
+      "loss": 0.8916,
+      "step": 11696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6929329695010964e-05,
+      "loss": 0.9717,
+      "step": 11697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6926251094730566e-05,
+      "loss": 0.9629,
+      "step": 11698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6923172569041018e-05,
+      "loss": 0.8145,
+      "step": 11699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6920094118017017e-05,
+      "loss": 0.8262,
+      "step": 11700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691701574173328e-05,
+      "loss": 1.0752,
+      "step": 11701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.69139374402645e-05,
+      "loss": 1.0029,
+      "step": 11702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6910859213685388e-05,
+      "loss": 0.7222,
+      "step": 11703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.690778106207063e-05,
+      "loss": 0.7969,
+      "step": 11704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.690470298549494e-05,
+      "loss": 1.0225,
+      "step": 11705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6901624984032998e-05,
+      "loss": 0.8311,
+      "step": 11706
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.689854705775951e-05,
+      "loss": 0.8711,
+      "step": 11707
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6895469206749165e-05,
+      "loss": 0.9863,
+      "step": 11708
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6892391431076652e-05,
+      "loss": 0.8945,
+      "step": 11709
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6889313730816656e-05,
+      "loss": 0.7993,
+      "step": 11710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6886236106043873e-05,
+      "loss": 0.7852,
+      "step": 11711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6883158556832974e-05,
+      "loss": 0.9717,
+      "step": 11712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.688008108325866e-05,
+      "loss": 0.8984,
+      "step": 11713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6877003685395597e-05,
+      "loss": 0.8369,
+      "step": 11714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6873926363318468e-05,
+      "loss": 0.9092,
+      "step": 11715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6870849117101958e-05,
+      "loss": 0.8701,
+      "step": 11716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6867771946820735e-05,
+      "loss": 0.8506,
+      "step": 11717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6864694852549478e-05,
+      "loss": 0.8242,
+      "step": 11718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6861617834362853e-05,
+      "loss": 0.9004,
+      "step": 11719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6858540892335536e-05,
+      "loss": 0.8164,
+      "step": 11720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.685546402654219e-05,
+      "loss": 0.9619,
+      "step": 11721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6852387237057487e-05,
+      "loss": 0.7773,
+      "step": 11722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6849310523956088e-05,
+      "loss": 0.6953,
+      "step": 11723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.684623388731266e-05,
+      "loss": 0.9678,
+      "step": 11724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6843157327201852e-05,
+      "loss": 0.9609,
+      "step": 11725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6840080843698345e-05,
+      "loss": 0.9512,
+      "step": 11726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.683700443687677e-05,
+      "loss": 0.9619,
+      "step": 11727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6833928106811805e-05,
+      "loss": 0.8848,
+      "step": 11728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6830851853578084e-05,
+      "loss": 0.9512,
+      "step": 11729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.682777567725028e-05,
+      "loss": 0.917,
+      "step": 11730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.682469957790302e-05,
+      "loss": 0.8066,
+      "step": 11731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6821623555610968e-05,
+      "loss": 0.8711,
+      "step": 11732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6818547610448765e-05,
+      "loss": 0.7539,
+      "step": 11733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6815471742491056e-05,
+      "loss": 0.6787,
+      "step": 11734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.681239595181248e-05,
+      "loss": 0.7666,
+      "step": 11735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6809320238487683e-05,
+      "loss": 0.8037,
+      "step": 11736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6806244602591296e-05,
+      "loss": 0.8203,
+      "step": 11737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6803169044197963e-05,
+      "loss": 0.9238,
+      "step": 11738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6800093563382318e-05,
+      "loss": 1.0088,
+      "step": 11739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.679701816021899e-05,
+      "loss": 1.0381,
+      "step": 11740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6793942834782608e-05,
+      "loss": 0.8965,
+      "step": 11741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6790867587147808e-05,
+      "loss": 0.875,
+      "step": 11742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6787792417389214e-05,
+      "loss": 0.9199,
+      "step": 11743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.678471732558145e-05,
+      "loss": 0.9062,
+      "step": 11744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.678164231179914e-05,
+      "loss": 0.9971,
+      "step": 11745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.677856737611691e-05,
+      "loss": 0.7754,
+      "step": 11746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.677549251860937e-05,
+      "loss": 1.0273,
+      "step": 11747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6772417739351152e-05,
+      "loss": 0.8018,
+      "step": 11748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6769343038416854e-05,
+      "loss": 0.9072,
+      "step": 11749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6766268415881097e-05,
+      "loss": 0.8516,
+      "step": 11750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6763193871818507e-05,
+      "loss": 0.8975,
+      "step": 11751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6760119406303674e-05,
+      "loss": 0.8008,
+      "step": 11752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6757045019411222e-05,
+      "loss": 0.9316,
+      "step": 11753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6753970711215745e-05,
+      "loss": 0.7998,
+      "step": 11754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6750896481791854e-05,
+      "loss": 0.835,
+      "step": 11755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.674782233121415e-05,
+      "loss": 0.958,
+      "step": 11756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6744748259557235e-05,
+      "loss": 0.958,
+      "step": 11757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6741674266895706e-05,
+      "loss": 0.8818,
+      "step": 11758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.673860035330416e-05,
+      "loss": 1.04,
+      "step": 11759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6735526518857195e-05,
+      "loss": 0.9424,
+      "step": 11760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6732452763629396e-05,
+      "loss": 0.8545,
+      "step": 11761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6729379087695363e-05,
+      "loss": 0.8789,
+      "step": 11762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6726305491129676e-05,
+      "loss": 0.7588,
+      "step": 11763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6723231974006934e-05,
+      "loss": 0.8701,
+      "step": 11764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6720158536401708e-05,
+      "loss": 0.8652,
+      "step": 11765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6717085178388593e-05,
+      "loss": 0.6836,
+      "step": 11766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6714011900042163e-05,
+      "loss": 0.8906,
+      "step": 11767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6710938701437006e-05,
+      "loss": 0.9307,
+      "step": 11768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.670786558264769e-05,
+      "loss": 0.6436,
+      "step": 11769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6704792543748797e-05,
+      "loss": 0.8506,
+      "step": 11770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6701719584814893e-05,
+      "loss": 0.8613,
+      "step": 11771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6698646705920557e-05,
+      "loss": 0.8564,
+      "step": 11772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6695573907140356e-05,
+      "loss": 0.9727,
+      "step": 11773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6692501188548866e-05,
+      "loss": 0.6968,
+      "step": 11774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6689428550220635e-05,
+      "loss": 0.8799,
+      "step": 11775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6686355992230244e-05,
+      "loss": 0.8203,
+      "step": 11776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.668328351465224e-05,
+      "loss": 0.9629,
+      "step": 11777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.66802111175612e-05,
+      "loss": 1.0781,
+      "step": 11778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6677138801031666e-05,
+      "loss": 0.9678,
+      "step": 11779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6674066565138204e-05,
+      "loss": 0.9688,
+      "step": 11780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6670994409955365e-05,
+      "loss": 0.8018,
+      "step": 11781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6667922335557706e-05,
+      "loss": 0.9648,
+      "step": 11782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6664850342019765e-05,
+      "loss": 0.9043,
+      "step": 11783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6661778429416102e-05,
+      "loss": 0.8828,
+      "step": 11784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6658706597821258e-05,
+      "loss": 0.9463,
+      "step": 11785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.665563484730978e-05,
+      "loss": 0.9189,
+      "step": 11786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6652563177956204e-05,
+      "loss": 0.9043,
+      "step": 11787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.664949158983508e-05,
+      "loss": 0.7729,
+      "step": 11788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.664642008302094e-05,
+      "loss": 0.9365,
+      "step": 11789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6643348657588325e-05,
+      "loss": 0.8589,
+      "step": 11790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.664027731361176e-05,
+      "loss": 0.9082,
+      "step": 11791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6637206051165795e-05,
+      "loss": 0.7622,
+      "step": 11792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6634134870324937e-05,
+      "loss": 0.8662,
+      "step": 11793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6631063771163735e-05,
+      "loss": 0.957,
+      "step": 11794
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6627992753756706e-05,
+      "loss": 0.9082,
+      "step": 11795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.662492181817838e-05,
+      "loss": 0.7383,
+      "step": 11796
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6621850964503265e-05,
+      "loss": 1.0596,
+      "step": 11797
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.66187801928059e-05,
+      "loss": 0.6846,
+      "step": 11798
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6615709503160792e-05,
+      "loss": 0.7559,
+      "step": 11799
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.661263889564247e-05,
+      "loss": 0.8184,
+      "step": 11800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6609568370325427e-05,
+      "loss": 0.7949,
+      "step": 11801
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6606497927284196e-05,
+      "loss": 0.9902,
+      "step": 11802
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6603427566593283e-05,
+      "loss": 0.9131,
+      "step": 11803
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6600357288327186e-05,
+      "loss": 0.6914,
+      "step": 11804
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.659728709256043e-05,
+      "loss": 0.9893,
+      "step": 11805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6594216979367498e-05,
+      "loss": 0.8994,
+      "step": 11806
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6591146948822913e-05,
+      "loss": 0.8584,
+      "step": 11807
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.658807700100116e-05,
+      "loss": 0.8877,
+      "step": 11808
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.658500713597675e-05,
+      "loss": 0.9033,
+      "step": 11809
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.658193735382417e-05,
+      "loss": 0.6748,
+      "step": 11810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6578867654617922e-05,
+      "loss": 0.9844,
+      "step": 11811
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.657579803843249e-05,
+      "loss": 0.9238,
+      "step": 11812
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6572728505342375e-05,
+      "loss": 0.9141,
+      "step": 11813
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6569659055422053e-05,
+      "loss": 0.9043,
+      "step": 11814
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6566589688746025e-05,
+      "loss": 0.7627,
+      "step": 11815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.656352040538876e-05,
+      "loss": 0.9629,
+      "step": 11816
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6560451205424756e-05,
+      "loss": 0.8877,
+      "step": 11817
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.655738208892848e-05,
+      "loss": 0.8594,
+      "step": 11818
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.655431305597442e-05,
+      "loss": 0.8379,
+      "step": 11819
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6551244106637046e-05,
+      "loss": 0.958,
+      "step": 11820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6548175240990838e-05,
+      "loss": 0.9121,
+      "step": 11821
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6545106459110263e-05,
+      "loss": 0.7031,
+      "step": 11822
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6542037761069796e-05,
+      "loss": 0.7729,
+      "step": 11823
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.65389691469439e-05,
+      "loss": 0.9482,
+      "step": 11824
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6535900616807046e-05,
+      "loss": 0.9414,
+      "step": 11825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6532832170733693e-05,
+      "loss": 0.9775,
+      "step": 11826
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6529763808798315e-05,
+      "loss": 0.9473,
+      "step": 11827
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6526695531075355e-05,
+      "loss": 0.9463,
+      "step": 11828
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.652362733763929e-05,
+      "loss": 1.0137,
+      "step": 11829
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6520559228564553e-05,
+      "loss": 0.9082,
+      "step": 11830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6517491203925623e-05,
+      "loss": 0.7393,
+      "step": 11831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.651442326379693e-05,
+      "loss": 0.9883,
+      "step": 11832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6511355408252933e-05,
+      "loss": 0.8818,
+      "step": 11833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6508287637368085e-05,
+      "loss": 0.9053,
+      "step": 11834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6505219951216825e-05,
+      "loss": 0.9492,
+      "step": 11835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6502152349873604e-05,
+      "loss": 1.0469,
+      "step": 11836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6499084833412856e-05,
+      "loss": 1.001,
+      "step": 11837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6496017401909025e-05,
+      "loss": 0.9658,
+      "step": 11838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6492950055436544e-05,
+      "loss": 0.8916,
+      "step": 11839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6489882794069848e-05,
+      "loss": 0.9492,
+      "step": 11840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6486815617883383e-05,
+      "loss": 1.0547,
+      "step": 11841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6483748526951564e-05,
+      "loss": 0.9746,
+      "step": 11842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.648068152134883e-05,
+      "loss": 0.9092,
+      "step": 11843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6477614601149604e-05,
+      "loss": 1.0146,
+      "step": 11844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6474547766428318e-05,
+      "loss": 0.9404,
+      "step": 11845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.647148101725938e-05,
+      "loss": 0.9053,
+      "step": 11846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.646841435371723e-05,
+      "loss": 0.9814,
+      "step": 11847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6465347775876272e-05,
+      "loss": 0.9307,
+      "step": 11848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6462281283810933e-05,
+      "loss": 0.9707,
+      "step": 11849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.645921487759562e-05,
+      "loss": 0.9697,
+      "step": 11850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6456148557304753e-05,
+      "loss": 0.8242,
+      "step": 11851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.645308232301273e-05,
+      "loss": 0.8662,
+      "step": 11852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6450016174793977e-05,
+      "loss": 0.8721,
+      "step": 11853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6446950112722884e-05,
+      "loss": 0.9951,
+      "step": 11854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.644388413687387e-05,
+      "loss": 0.9941,
+      "step": 11855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.644081824732132e-05,
+      "loss": 0.8818,
+      "step": 11856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6437752444139655e-05,
+      "loss": 0.959,
+      "step": 11857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6434686727403254e-05,
+      "loss": 0.8535,
+      "step": 11858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6431621097186522e-05,
+      "loss": 0.9443,
+      "step": 11859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6428555553563855e-05,
+      "loss": 0.8643,
+      "step": 11860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.642549009660964e-05,
+      "loss": 1.1289,
+      "step": 11861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6422424726398264e-05,
+      "loss": 0.9053,
+      "step": 11862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6419359443004125e-05,
+      "loss": 1.0391,
+      "step": 11863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6416294246501598e-05,
+      "loss": 0.8535,
+      "step": 11864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6413229136965075e-05,
+      "loss": 1.0303,
+      "step": 11865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6410164114468926e-05,
+      "loss": 0.9043,
+      "step": 11866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6407099179087544e-05,
+      "loss": 0.874,
+      "step": 11867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.640403433089529e-05,
+      "loss": 0.9238,
+      "step": 11868
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6400969569966555e-05,
+      "loss": 0.9697,
+      "step": 11869
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.63979048963757e-05,
+      "loss": 0.9404,
+      "step": 11870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6394840310197103e-05,
+      "loss": 0.8564,
+      "step": 11871
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6391775811505125e-05,
+      "loss": 0.8916,
+      "step": 11872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.638871140037414e-05,
+      "loss": 0.917,
+      "step": 11873
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6385647076878506e-05,
+      "loss": 1.0322,
+      "step": 11874
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6382582841092592e-05,
+      "loss": 0.9404,
+      "step": 11875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.637951869309075e-05,
+      "loss": 0.9629,
+      "step": 11876
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6376454632947344e-05,
+      "loss": 0.8047,
+      "step": 11877
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.637339066073673e-05,
+      "loss": 1.0059,
+      "step": 11878
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6370326776533256e-05,
+      "loss": 0.9307,
+      "step": 11879
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6367262980411273e-05,
+      "loss": 0.9521,
+      "step": 11880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.636419927244514e-05,
+      "loss": 0.5884,
+      "step": 11881
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6361135652709194e-05,
+      "loss": 0.9482,
+      "step": 11882
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.635807212127779e-05,
+      "loss": 0.8193,
+      "step": 11883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6355008678225258e-05,
+      "loss": 0.6777,
+      "step": 11884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6351945323625954e-05,
+      "loss": 0.7126,
+      "step": 11885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.63488820575542e-05,
+      "loss": 0.7852,
+      "step": 11886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6345818880084343e-05,
+      "loss": 0.9316,
+      "step": 11887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6342755791290718e-05,
+      "loss": 0.9268,
+      "step": 11888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6339692791247652e-05,
+      "loss": 0.9404,
+      "step": 11889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6336629880029483e-05,
+      "loss": 0.8135,
+      "step": 11890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6333567057710525e-05,
+      "loss": 0.9424,
+      "step": 11891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6330504324365118e-05,
+      "loss": 0.8994,
+      "step": 11892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6327441680067576e-05,
+      "loss": 0.9766,
+      "step": 11893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6324379124892225e-05,
+      "loss": 1.0605,
+      "step": 11894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6321316658913384e-05,
+      "loss": 0.8701,
+      "step": 11895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6318254282205367e-05,
+      "loss": 0.9473,
+      "step": 11896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.631519199484249e-05,
+      "loss": 0.792,
+      "step": 11897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.631212979689907e-05,
+      "loss": 0.9043,
+      "step": 11898
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6309067688449412e-05,
+      "loss": 0.8623,
+      "step": 11899
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6306005669567832e-05,
+      "loss": 0.9932,
+      "step": 11900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6302943740328623e-05,
+      "loss": 1.0361,
+      "step": 11901
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.62998819008061e-05,
+      "loss": 0.9121,
+      "step": 11902
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.629682015107456e-05,
+      "loss": 0.8496,
+      "step": 11903
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.629375849120831e-05,
+      "loss": 0.8799,
+      "step": 11904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6290696921281635e-05,
+      "loss": 0.8398,
+      "step": 11905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.628763544136884e-05,
+      "loss": 0.9258,
+      "step": 11906
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6284574051544213e-05,
+      "loss": 0.9199,
+      "step": 11907
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6281512751882053e-05,
+      "loss": 0.8145,
+      "step": 11908
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6278451542456636e-05,
+      "loss": 0.8291,
+      "step": 11909
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6275390423342262e-05,
+      "loss": 0.9756,
+      "step": 11910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.62723293946132e-05,
+      "loss": 0.9561,
+      "step": 11911
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.626926845634375e-05,
+      "loss": 1.0186,
+      "step": 11912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6266207608608175e-05,
+      "loss": 0.9746,
+      "step": 11913
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6263146851480767e-05,
+      "loss": 1.0225,
+      "step": 11914
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.626008618503579e-05,
+      "loss": 0.8643,
+      "step": 11915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.625702560934752e-05,
+      "loss": 1.0127,
+      "step": 11916
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6253965124490236e-05,
+      "loss": 0.8936,
+      "step": 11917
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.62509047305382e-05,
+      "loss": 0.9307,
+      "step": 11918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6247844427565685e-05,
+      "loss": 0.8203,
+      "step": 11919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.624478421564695e-05,
+      "loss": 1.0859,
+      "step": 11920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6241724094856255e-05,
+      "loss": 0.8916,
+      "step": 11921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6238664065267865e-05,
+      "loss": 0.8506,
+      "step": 11922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6235604126956038e-05,
+      "loss": 0.9961,
+      "step": 11923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6232544279995023e-05,
+      "loss": 0.7803,
+      "step": 11924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6229484524459085e-05,
+      "loss": 0.6404,
+      "step": 11925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6226424860422465e-05,
+      "loss": 0.8032,
+      "step": 11926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6223365287959413e-05,
+      "loss": 0.8652,
+      "step": 11927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.622030580714419e-05,
+      "loss": 0.835,
+      "step": 11928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6217246418051014e-05,
+      "loss": 0.9932,
+      "step": 11929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6214187120754158e-05,
+      "loss": 0.9971,
+      "step": 11930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6211127915327837e-05,
+      "loss": 1.0,
+      "step": 11931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6208068801846305e-05,
+      "loss": 0.8823,
+      "step": 11932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6205009780383783e-05,
+      "loss": 0.7832,
+      "step": 11933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6201950851014522e-05,
+      "loss": 0.957,
+      "step": 11934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.619889201381274e-05,
+      "loss": 1.0635,
+      "step": 11935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.619583326885267e-05,
+      "loss": 0.917,
+      "step": 11936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6192774616208538e-05,
+      "loss": 0.9131,
+      "step": 11937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6189716055954575e-05,
+      "loss": 0.9658,
+      "step": 11938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6186657588164992e-05,
+      "loss": 0.7358,
+      "step": 11939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.618359921291402e-05,
+      "loss": 0.8877,
+      "step": 11940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6180540930275864e-05,
+      "loss": 0.9668,
+      "step": 11941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6177482740324753e-05,
+      "loss": 0.8047,
+      "step": 11942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.617442464313489e-05,
+      "loss": 0.8643,
+      "step": 11943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6171366638780494e-05,
+      "loss": 0.8076,
+      "step": 11944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6168308727335768e-05,
+      "loss": 0.8203,
+      "step": 11945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.616525090887492e-05,
+      "loss": 1.0801,
+      "step": 11946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6162193183472154e-05,
+      "loss": 0.834,
+      "step": 11947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6159135551201678e-05,
+      "loss": 0.7812,
+      "step": 11948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.615607801213768e-05,
+      "loss": 0.8887,
+      "step": 11949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.615302056635437e-05,
+      "loss": 0.9541,
+      "step": 11950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6149963213925935e-05,
+      "loss": 0.9326,
+      "step": 11951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.614690595492657e-05,
+      "loss": 0.9092,
+      "step": 11952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6143848789430466e-05,
+      "loss": 0.7427,
+      "step": 11953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6140791717511813e-05,
+      "loss": 0.7373,
+      "step": 11954
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.613773473924479e-05,
+      "loss": 1.0,
+      "step": 11955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6134677854703593e-05,
+      "loss": 1.0176,
+      "step": 11956
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6131621063962393e-05,
+      "loss": 0.9023,
+      "step": 11957
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6128564367095377e-05,
+      "loss": 0.8965,
+      "step": 11958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6125507764176712e-05,
+      "loss": 0.874,
+      "step": 11959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6122451255280588e-05,
+      "loss": 0.9736,
+      "step": 11960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6119394840481158e-05,
+      "loss": 0.9004,
+      "step": 11961
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.611633851985261e-05,
+      "loss": 0.7568,
+      "step": 11962
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6113282293469104e-05,
+      "loss": 0.9277,
+      "step": 11963
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6110226161404806e-05,
+      "loss": 0.9844,
+      "step": 11964
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6107170123733873e-05,
+      "loss": 0.9697,
+      "step": 11965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6104114180530482e-05,
+      "loss": 0.9092,
+      "step": 11966
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.610105833186878e-05,
+      "loss": 0.9697,
+      "step": 11967
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6098002577822925e-05,
+      "loss": 0.8101,
+      "step": 11968
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.609494691846707e-05,
+      "loss": 0.6699,
+      "step": 11969
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6091891353875376e-05,
+      "loss": 1.0605,
+      "step": 11970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6088835884121975e-05,
+      "loss": 0.9873,
+      "step": 11971
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.608578050928103e-05,
+      "loss": 0.8271,
+      "step": 11972
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6082725229426683e-05,
+      "loss": 0.9004,
+      "step": 11973
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.607967004463307e-05,
+      "loss": 0.791,
+      "step": 11974
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.607661495497434e-05,
+      "loss": 0.9424,
+      "step": 11975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6073559960524624e-05,
+      "loss": 0.8428,
+      "step": 11976
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6070505061358062e-05,
+      "loss": 0.9883,
+      "step": 11977
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6067450257548782e-05,
+      "loss": 0.8643,
+      "step": 11978
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6064395549170926e-05,
+      "loss": 0.9131,
+      "step": 11979
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.606134093629861e-05,
+      "loss": 0.8721,
+      "step": 11980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.605828641900597e-05,
+      "loss": 0.957,
+      "step": 11981
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6055231997367123e-05,
+      "loss": 0.9404,
+      "step": 11982
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6052177671456196e-05,
+      "loss": 0.8906,
+      "step": 11983
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6049123441347308e-05,
+      "loss": 0.8857,
+      "step": 11984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6046069307114574e-05,
+      "loss": 0.7583,
+      "step": 11985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.604301526883211e-05,
+      "loss": 0.9863,
+      "step": 11986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.603996132657403e-05,
+      "loss": 1.0176,
+      "step": 11987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.603690748041444e-05,
+      "loss": 0.9512,
+      "step": 11988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.603385373042745e-05,
+      "loss": 0.7676,
+      "step": 11989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6030800076687164e-05,
+      "loss": 0.9424,
+      "step": 11990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6027746519267692e-05,
+      "loss": 0.834,
+      "step": 11991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6024693058243127e-05,
+      "loss": 0.9033,
+      "step": 11992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6021639693687576e-05,
+      "loss": 0.7812,
+      "step": 11993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.601858642567512e-05,
+      "loss": 0.9424,
+      "step": 11994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.601553325427987e-05,
+      "loss": 0.8311,
+      "step": 11995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6012480179575906e-05,
+      "loss": 0.8926,
+      "step": 11996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6009427201637325e-05,
+      "loss": 0.874,
+      "step": 11997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6006374320538208e-05,
+      "loss": 1.0283,
+      "step": 11998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6003321536352643e-05,
+      "loss": 0.9463,
+      "step": 11999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6000268849154703e-05,
+      "loss": 0.9199,
+      "step": 12000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.599721625901848e-05,
+      "loss": 0.9268,
+      "step": 12001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.599416376601805e-05,
+      "loss": 0.8975,
+      "step": 12002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.599111137022748e-05,
+      "loss": 0.8818,
+      "step": 12003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5988059071720853e-05,
+      "loss": 0.8193,
+      "step": 12004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5985006870572227e-05,
+      "loss": 0.8555,
+      "step": 12005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.598195476685568e-05,
+      "loss": 0.9238,
+      "step": 12006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5978902760645273e-05,
+      "loss": 0.918,
+      "step": 12007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5975850852015074e-05,
+      "loss": 0.9482,
+      "step": 12008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5972799041039134e-05,
+      "loss": 0.9326,
+      "step": 12009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5969747327791524e-05,
+      "loss": 0.9043,
+      "step": 12010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5966695712346293e-05,
+      "loss": 0.7578,
+      "step": 12011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.596364419477749e-05,
+      "loss": 0.917,
+      "step": 12012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.596059277515918e-05,
+      "loss": 0.7065,
+      "step": 12013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5957541453565394e-05,
+      "loss": 1.1035,
+      "step": 12014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.59544902300702e-05,
+      "loss": 0.7627,
+      "step": 12015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5951439104747622e-05,
+      "loss": 0.8105,
+      "step": 12016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5948388077671714e-05,
+      "loss": 1.0117,
+      "step": 12017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.594533714891651e-05,
+      "loss": 0.9678,
+      "step": 12018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.594228631855605e-05,
+      "loss": 0.9561,
+      "step": 12019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5939235586664368e-05,
+      "loss": 0.6787,
+      "step": 12020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.59361849533155e-05,
+      "loss": 0.8223,
+      "step": 12021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5933134418583463e-05,
+      "loss": 0.8271,
+      "step": 12022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.59300839825423e-05,
+      "loss": 0.8896,
+      "step": 12023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.592703364526603e-05,
+      "loss": 0.9131,
+      "step": 12024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5923983406828673e-05,
+      "loss": 0.8887,
+      "step": 12025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.592093326730425e-05,
+      "loss": 1.0293,
+      "step": 12026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5917883226766782e-05,
+      "loss": 0.8506,
+      "step": 12027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5914833285290284e-05,
+      "loss": 0.958,
+      "step": 12028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.591178344294877e-05,
+      "loss": 1.0352,
+      "step": 12029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5908733699816242e-05,
+      "loss": 0.9092,
+      "step": 12030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5905684055966725e-05,
+      "loss": 0.835,
+      "step": 12031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5902634511474207e-05,
+      "loss": 0.8467,
+      "step": 12032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5899585066412704e-05,
+      "loss": 0.8662,
+      "step": 12033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5896535720856207e-05,
+      "loss": 0.9102,
+      "step": 12034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5893486474878727e-05,
+      "loss": 0.832,
+      "step": 12035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5890437328554245e-05,
+      "loss": 0.9521,
+      "step": 12036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5887388281956774e-05,
+      "loss": 0.8813,
+      "step": 12037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5884339335160285e-05,
+      "loss": 0.8252,
+      "step": 12038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5881290488238782e-05,
+      "loss": 0.9355,
+      "step": 12039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5878241741266246e-05,
+      "loss": 1.0039,
+      "step": 12040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5875193094316662e-05,
+      "loss": 0.9795,
+      "step": 12041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5872144547464005e-05,
+      "loss": 0.832,
+      "step": 12042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5869096100782266e-05,
+      "loss": 0.8545,
+      "step": 12043
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5866047754345414e-05,
+      "loss": 0.9229,
+      "step": 12044
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5862999508227427e-05,
+      "loss": 0.8691,
+      "step": 12045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.585995136250227e-05,
+      "loss": 0.9199,
+      "step": 12046
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5856903317243928e-05,
+      "loss": 1.0293,
+      "step": 12047
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.585385537252635e-05,
+      "loss": 0.8818,
+      "step": 12048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5850807528423513e-05,
+      "loss": 1.0029,
+      "step": 12049
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5847759785009372e-05,
+      "loss": 1.0254,
+      "step": 12050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.584471214235789e-05,
+      "loss": 0.9463,
+      "step": 12051
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5841664600543025e-05,
+      "loss": 0.9805,
+      "step": 12052
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5838617159638735e-05,
+      "loss": 0.9746,
+      "step": 12053
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.583556981971896e-05,
+      "loss": 1.0098,
+      "step": 12054
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5832522580857666e-05,
+      "loss": 0.8975,
+      "step": 12055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5829475443128786e-05,
+      "loss": 0.873,
+      "step": 12056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.582642840660628e-05,
+      "loss": 0.79,
+      "step": 12057
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5823381471364074e-05,
+      "loss": 0.8662,
+      "step": 12058
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5820334637476117e-05,
+      "loss": 0.7378,
+      "step": 12059
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5817287905016353e-05,
+      "loss": 0.9404,
+      "step": 12060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5814241274058707e-05,
+      "loss": 0.9492,
+      "step": 12061
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5811194744677116e-05,
+      "loss": 0.8047,
+      "step": 12062
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.580814831694551e-05,
+      "loss": 0.791,
+      "step": 12063
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5805101990937816e-05,
+      "loss": 0.9258,
+      "step": 12064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.580205576672796e-05,
+      "loss": 0.8711,
+      "step": 12065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5799009644389866e-05,
+      "loss": 0.937,
+      "step": 12066
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.579596362399745e-05,
+      "loss": 0.835,
+      "step": 12067
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5792917705624638e-05,
+      "loss": 0.7378,
+      "step": 12068
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5789871889345336e-05,
+      "loss": 0.8848,
+      "step": 12069
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5786826175233465e-05,
+      "loss": 0.8496,
+      "step": 12070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.578378056336293e-05,
+      "loss": 0.9268,
+      "step": 12071
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5780735053807645e-05,
+      "loss": 0.7832,
+      "step": 12072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5777689646641506e-05,
+      "loss": 0.8516,
+      "step": 12073
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.577464434193843e-05,
+      "loss": 1.042,
+      "step": 12074
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.57715991397723e-05,
+      "loss": 0.7764,
+      "step": 12075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.576855404021703e-05,
+      "loss": 1.0,
+      "step": 12076
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5765509043346506e-05,
+      "loss": 0.9189,
+      "step": 12077
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5762464149234627e-05,
+      "loss": 0.8096,
+      "step": 12078
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5759419357955275e-05,
+      "loss": 0.9346,
+      "step": 12079
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5756374669582353e-05,
+      "loss": 0.9385,
+      "step": 12080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5753330084189734e-05,
+      "loss": 0.998,
+      "step": 12081
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5750285601851305e-05,
+      "loss": 0.7939,
+      "step": 12082
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.574724122264094e-05,
+      "loss": 1.1035,
+      "step": 12083
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5744196946632526e-05,
+      "loss": 0.9854,
+      "step": 12084
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5741152773899943e-05,
+      "loss": 1.0352,
+      "step": 12085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.573810870451705e-05,
+      "loss": 0.9717,
+      "step": 12086
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5735064738557733e-05,
+      "loss": 0.8291,
+      "step": 12087
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5732020876095843e-05,
+      "loss": 0.8496,
+      "step": 12088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5728977117205265e-05,
+      "loss": 0.8965,
+      "step": 12089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5725933461959845e-05,
+      "loss": 0.9609,
+      "step": 12090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5722889910433453e-05,
+      "loss": 0.7734,
+      "step": 12091
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.571984646269994e-05,
+      "loss": 0.7402,
+      "step": 12092
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5716803118833173e-05,
+      "loss": 0.8916,
+      "step": 12093
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.571375987890699e-05,
+      "loss": 0.875,
+      "step": 12094
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5710716742995257e-05,
+      "loss": 0.9658,
+      "step": 12095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.570767371117181e-05,
+      "loss": 0.9727,
+      "step": 12096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5704630783510498e-05,
+      "loss": 0.8389,
+      "step": 12097
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5701587960085164e-05,
+      "loss": 0.7422,
+      "step": 12098
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.569854524096965e-05,
+      "loss": 0.8662,
+      "step": 12099
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5695502626237795e-05,
+      "loss": 0.8262,
+      "step": 12100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.569246011596343e-05,
+      "loss": 0.9482,
+      "step": 12101
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.568941771022039e-05,
+      "loss": 0.8032,
+      "step": 12102
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5686375409082505e-05,
+      "loss": 0.9023,
+      "step": 12103
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.568333321262361e-05,
+      "loss": 0.8486,
+      "step": 12104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5680291120917512e-05,
+      "loss": 1.043,
+      "step": 12105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.567724913403805e-05,
+      "loss": 0.8008,
+      "step": 12106
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.567420725205904e-05,
+      "loss": 0.8301,
+      "step": 12107
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.56711654750543e-05,
+      "loss": 0.8965,
+      "step": 12108
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5668123803097636e-05,
+      "loss": 0.9883,
+      "step": 12109
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5665082236262874e-05,
+      "loss": 0.8555,
+      "step": 12110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5662040774623814e-05,
+      "loss": 0.7246,
+      "step": 12111
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.565899941825427e-05,
+      "loss": 0.9355,
+      "step": 12112
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.565595816722804e-05,
+      "loss": 0.8496,
+      "step": 12113
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5652917021618934e-05,
+      "loss": 0.8418,
+      "step": 12114
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5649875981500745e-05,
+      "loss": 1.1133,
+      "step": 12115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5646835046947275e-05,
+      "loss": 0.959,
+      "step": 12116
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5643794218032313e-05,
+      "loss": 0.873,
+      "step": 12117
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5640753494829657e-05,
+      "loss": 0.9404,
+      "step": 12118
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5637712877413092e-05,
+      "loss": 0.7852,
+      "step": 12119
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5634672365856408e-05,
+      "loss": 0.707,
+      "step": 12120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5631631960233384e-05,
+      "loss": 0.9248,
+      "step": 12121
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5628591660617808e-05,
+      "loss": 1.0283,
+      "step": 12122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5625551467083455e-05,
+      "loss": 0.9805,
+      "step": 12123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5622511379704106e-05,
+      "loss": 0.834,
+      "step": 12124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.561947139855353e-05,
+      "loss": 0.8174,
+      "step": 12125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5616431523705502e-05,
+      "loss": 0.8291,
+      "step": 12126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5613391755233787e-05,
+      "loss": 0.9062,
+      "step": 12127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5610352093212158e-05,
+      "loss": 1.1113,
+      "step": 12128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.560731253771437e-05,
+      "loss": 0.9365,
+      "step": 12129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.560427308881419e-05,
+      "loss": 0.7998,
+      "step": 12130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5601233746585373e-05,
+      "loss": 0.9766,
+      "step": 12131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.559819451110168e-05,
+      "loss": 0.8584,
+      "step": 12132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5595155382436856e-05,
+      "loss": 0.8354,
+      "step": 12133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5592116360664664e-05,
+      "loss": 0.7241,
+      "step": 12134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5589077445858834e-05,
+      "loss": 0.8926,
+      "step": 12135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.558603863809313e-05,
+      "loss": 0.9014,
+      "step": 12136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5582999937441287e-05,
+      "loss": 1.0234,
+      "step": 12137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5579961343977047e-05,
+      "loss": 0.9219,
+      "step": 12138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.557692285777414e-05,
+      "loss": 0.7183,
+      "step": 12139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5573884478906312e-05,
+      "loss": 0.8955,
+      "step": 12140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.557084620744729e-05,
+      "loss": 0.873,
+      "step": 12141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.556780804347081e-05,
+      "loss": 1.0234,
+      "step": 12142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5564769987050586e-05,
+      "loss": 0.9375,
+      "step": 12143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.556173203826035e-05,
+      "loss": 0.873,
+      "step": 12144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.555869419717383e-05,
+      "loss": 0.9707,
+      "step": 12145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.555565646386474e-05,
+      "loss": 0.7998,
+      "step": 12146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5552618838406798e-05,
+      "loss": 0.9473,
+      "step": 12147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5549581320873715e-05,
+      "loss": 0.9258,
+      "step": 12148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5546543911339206e-05,
+      "loss": 0.8916,
+      "step": 12149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.554350660987698e-05,
+      "loss": 0.9932,
+      "step": 12150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5540469416560744e-05,
+      "loss": 0.8887,
+      "step": 12151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5537432331464195e-05,
+      "loss": 0.9014,
+      "step": 12152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5534395354661045e-05,
+      "loss": 0.8037,
+      "step": 12153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5531358486224983e-05,
+      "loss": 0.8975,
+      "step": 12154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5528321726229714e-05,
+      "loss": 0.624,
+      "step": 12155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.552528507474892e-05,
+      "loss": 0.9434,
+      "step": 12156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.55222485318563e-05,
+      "loss": 0.9072,
+      "step": 12157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.551921209762554e-05,
+      "loss": 0.9043,
+      "step": 12158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5516175772130326e-05,
+      "loss": 0.877,
+      "step": 12159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5513139555444334e-05,
+      "loss": 0.8037,
+      "step": 12160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5510103447641257e-05,
+      "loss": 0.8975,
+      "step": 12161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5507067448794758e-05,
+      "loss": 0.916,
+      "step": 12162
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5504031558978523e-05,
+      "loss": 0.9307,
+      "step": 12163
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5500995778266217e-05,
+      "loss": 0.8945,
+      "step": 12164
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.549796010673152e-05,
+      "loss": 0.7617,
+      "step": 12165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5494924544448078e-05,
+      "loss": 0.9805,
+      "step": 12166
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5491889091489573e-05,
+      "loss": 0.9473,
+      "step": 12167
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5488853747929667e-05,
+      "loss": 0.7939,
+      "step": 12168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.548581851384201e-05,
+      "loss": 0.9082,
+      "step": 12169
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5482783389300266e-05,
+      "loss": 0.8115,
+      "step": 12170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.547974837437808e-05,
+      "loss": 0.9287,
+      "step": 12171
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5476713469149113e-05,
+      "loss": 0.9697,
+      "step": 12172
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5473678673687e-05,
+      "loss": 0.9092,
+      "step": 12173
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.54706439880654e-05,
+      "loss": 0.8721,
+      "step": 12174
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5467609412357948e-05,
+      "loss": 0.8135,
+      "step": 12175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5464574946638286e-05,
+      "loss": 0.7686,
+      "step": 12176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.546154059098005e-05,
+      "loss": 0.9688,
+      "step": 12177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5458506345456883e-05,
+      "loss": 0.6353,
+      "step": 12178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5455472210142404e-05,
+      "loss": 0.8174,
+      "step": 12179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5452438185110254e-05,
+      "loss": 0.7695,
+      "step": 12180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.544940427043405e-05,
+      "loss": 0.7656,
+      "step": 12181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5446370466187432e-05,
+      "loss": 0.9893,
+      "step": 12182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5443336772444002e-05,
+      "loss": 1.0244,
+      "step": 12183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.544030318927739e-05,
+      "loss": 1.0713,
+      "step": 12184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.543726971676121e-05,
+      "loss": 0.9248,
+      "step": 12185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5434236354969076e-05,
+      "loss": 0.9131,
+      "step": 12186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5431203103974596e-05,
+      "loss": 0.8877,
+      "step": 12187
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5428169963851382e-05,
+      "loss": 0.9014,
+      "step": 12188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5425136934673038e-05,
+      "loss": 0.9746,
+      "step": 12189
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5422104016513165e-05,
+      "loss": 1.0088,
+      "step": 12190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.541907120944537e-05,
+      "loss": 0.7432,
+      "step": 12191
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5416038513543235e-05,
+      "loss": 0.8779,
+      "step": 12192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5413005928880376e-05,
+      "loss": 0.8623,
+      "step": 12193
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5409973455530364e-05,
+      "loss": 0.9355,
+      "step": 12194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5406941093566802e-05,
+      "loss": 0.8125,
+      "step": 12195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5403908843063268e-05,
+      "loss": 0.9873,
+      "step": 12196
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5400876704093354e-05,
+      "loss": 0.7578,
+      "step": 12197
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5397844676730635e-05,
+      "loss": 0.9521,
+      "step": 12198
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.539481276104869e-05,
+      "loss": 0.915,
+      "step": 12199
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5391780957121097e-05,
+      "loss": 0.8037,
+      "step": 12200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.538874926502143e-05,
+      "loss": 0.8379,
+      "step": 12201
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5385717684823255e-05,
+      "loss": 0.7705,
+      "step": 12202
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5382686216600144e-05,
+      "loss": 0.8721,
+      "step": 12203
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5379654860425655e-05,
+      "loss": 0.9209,
+      "step": 12204
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5376623616373357e-05,
+      "loss": 1.0645,
+      "step": 12205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5373592484516808e-05,
+      "loss": 0.8555,
+      "step": 12206
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5370561464929563e-05,
+      "loss": 0.9316,
+      "step": 12207
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5367530557685175e-05,
+      "loss": 0.6631,
+      "step": 12208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.53644997628572e-05,
+      "loss": 0.9473,
+      "step": 12209
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.536146908051918e-05,
+      "loss": 0.9346,
+      "step": 12210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5358438510744668e-05,
+      "loss": 0.9199,
+      "step": 12211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5355408053607198e-05,
+      "loss": 0.957,
+      "step": 12212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.535237770918032e-05,
+      "loss": 0.8809,
+      "step": 12213
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5349347477537563e-05,
+      "loss": 0.8159,
+      "step": 12214
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5346317358752466e-05,
+      "loss": 0.9219,
+      "step": 12215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.534328735289856e-05,
+      "loss": 0.8359,
+      "step": 12216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5340257460049383e-05,
+      "loss": 1.0273,
+      "step": 12217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.533722768027844e-05,
+      "loss": 0.9512,
+      "step": 12218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.533419801365928e-05,
+      "loss": 0.6685,
+      "step": 12219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5331168460265407e-05,
+      "loss": 0.9258,
+      "step": 12220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5328139020170348e-05,
+      "loss": 0.9238,
+      "step": 12221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5325109693447613e-05,
+      "loss": 0.8604,
+      "step": 12222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5322080480170722e-05,
+      "loss": 0.8203,
+      "step": 12223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5319051380413172e-05,
+      "loss": 0.9961,
+      "step": 12224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5316022394248487e-05,
+      "loss": 0.7378,
+      "step": 12225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5312993521750156e-05,
+      "loss": 1.042,
+      "step": 12226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5309964762991696e-05,
+      "loss": 0.7168,
+      "step": 12227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.530693611804659e-05,
+      "loss": 0.8271,
+      "step": 12228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.530390758698835e-05,
+      "loss": 0.6553,
+      "step": 12229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.530087916989046e-05,
+      "loss": 0.8984,
+      "step": 12230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5297850866826408e-05,
+      "loss": 0.7251,
+      "step": 12231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5294822677869695e-05,
+      "loss": 1.0947,
+      "step": 12232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5291794603093793e-05,
+      "loss": 1.0205,
+      "step": 12233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.528876664257219e-05,
+      "loss": 0.8525,
+      "step": 12234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5285738796378367e-05,
+      "loss": 1.0039,
+      "step": 12235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.52827110645858e-05,
+      "loss": 0.9707,
+      "step": 12236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5279683447267957e-05,
+      "loss": 0.6733,
+      "step": 12237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.527665594449832e-05,
+      "loss": 0.9902,
+      "step": 12238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.527362855635035e-05,
+      "loss": 0.8467,
+      "step": 12239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.527060128289752e-05,
+      "loss": 1.0039,
+      "step": 12240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5267574124213284e-05,
+      "loss": 0.8232,
+      "step": 12241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.526454708037111e-05,
+      "loss": 0.876,
+      "step": 12242
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5261520151444444e-05,
+      "loss": 0.7998,
+      "step": 12243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5258493337506756e-05,
+      "loss": 0.9814,
+      "step": 12244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5255466638631487e-05,
+      "loss": 0.793,
+      "step": 12245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.525244005489209e-05,
+      "loss": 1.0469,
+      "step": 12246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.524941358636201e-05,
+      "loss": 0.8516,
+      "step": 12247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5246387233114695e-05,
+      "loss": 0.873,
+      "step": 12248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5243360995223577e-05,
+      "loss": 0.8213,
+      "step": 12249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5240334872762095e-05,
+      "loss": 0.8379,
+      "step": 12250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5237308865803698e-05,
+      "loss": 0.7803,
+      "step": 12251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5234282974421801e-05,
+      "loss": 0.7764,
+      "step": 12252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5231257198689848e-05,
+      "loss": 0.832,
+      "step": 12253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5228231538681253e-05,
+      "loss": 0.7686,
+      "step": 12254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5225205994469447e-05,
+      "loss": 0.8594,
+      "step": 12255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5222180566127848e-05,
+      "loss": 0.8271,
+      "step": 12256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5219155253729878e-05,
+      "loss": 0.9971,
+      "step": 12257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5216130057348947e-05,
+      "loss": 0.7583,
+      "step": 12258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5213104977058476e-05,
+      "loss": 0.8994,
+      "step": 12259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5210080012931863e-05,
+      "loss": 0.9336,
+      "step": 12260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5207055165042527e-05,
+      "loss": 0.876,
+      "step": 12261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5204030433463863e-05,
+      "loss": 0.9141,
+      "step": 12262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5201005818269278e-05,
+      "loss": 0.8032,
+      "step": 12263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5197981319532167e-05,
+      "loss": 0.8213,
+      "step": 12264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5194956937325928e-05,
+      "loss": 0.9082,
+      "step": 12265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5191932671723953e-05,
+      "loss": 0.9932,
+      "step": 12266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5188908522799633e-05,
+      "loss": 0.8477,
+      "step": 12267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.518588449062635e-05,
+      "loss": 0.9111,
+      "step": 12268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5182860575277494e-05,
+      "loss": 1.082,
+      "step": 12269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5179836776826449e-05,
+      "loss": 0.9375,
+      "step": 12270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5176813095346585e-05,
+      "loss": 0.9463,
+      "step": 12271
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5173789530911287e-05,
+      "loss": 0.9746,
+      "step": 12272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.517076608359392e-05,
+      "loss": 1.0049,
+      "step": 12273
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5167742753467862e-05,
+      "loss": 0.7979,
+      "step": 12274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5164719540606472e-05,
+      "loss": 0.8477,
+      "step": 12275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5161696445083123e-05,
+      "loss": 0.9541,
+      "step": 12276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5158673466971168e-05,
+      "loss": 1.1289,
+      "step": 12277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5155650606343974e-05,
+      "loss": 0.8555,
+      "step": 12278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5152627863274889e-05,
+      "loss": 0.957,
+      "step": 12279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5149605237837275e-05,
+      "loss": 0.8379,
+      "step": 12280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5146582730104473e-05,
+      "loss": 0.877,
+      "step": 12281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5143560340149838e-05,
+      "loss": 0.668,
+      "step": 12282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5140538068046708e-05,
+      "loss": 1.0254,
+      "step": 12283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5137515913868432e-05,
+      "loss": 0.7725,
+      "step": 12284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5134493877688342e-05,
+      "loss": 0.9922,
+      "step": 12285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5131471959579778e-05,
+      "loss": 0.8496,
+      "step": 12286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5128450159616068e-05,
+      "loss": 0.8555,
+      "step": 12287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5125428477870552e-05,
+      "loss": 0.9922,
+      "step": 12288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5122406914416545e-05,
+      "loss": 0.8467,
+      "step": 12289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5119385469327383e-05,
+      "loss": 1.0537,
+      "step": 12290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5116364142676379e-05,
+      "loss": 0.9268,
+      "step": 12291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5113342934536856e-05,
+      "loss": 0.9727,
+      "step": 12292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.511032184498213e-05,
+      "loss": 0.9893,
+      "step": 12293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5107300874085513e-05,
+      "loss": 0.958,
+      "step": 12294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.510428002192031e-05,
+      "loss": 0.9746,
+      "step": 12295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5101259288559839e-05,
+      "loss": 0.8428,
+      "step": 12296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5098238674077394e-05,
+      "loss": 0.915,
+      "step": 12297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5095218178546285e-05,
+      "loss": 0.9219,
+      "step": 12298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5092197802039803e-05,
+      "loss": 0.7842,
+      "step": 12299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.508917754463125e-05,
+      "loss": 0.874,
+      "step": 12300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5086157406393911e-05,
+      "loss": 0.9961,
+      "step": 12301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5083137387401085e-05,
+      "loss": 0.7544,
+      "step": 12302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5080117487726052e-05,
+      "loss": 0.793,
+      "step": 12303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5077097707442104e-05,
+      "loss": 0.8516,
+      "step": 12304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.507407804662251e-05,
+      "loss": 0.8574,
+      "step": 12305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5071058505340558e-05,
+      "loss": 1.1035,
+      "step": 12306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5068039083669522e-05,
+      "loss": 0.7275,
+      "step": 12307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5065019781682672e-05,
+      "loss": 0.8682,
+      "step": 12308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5062000599453276e-05,
+      "loss": 0.6265,
+      "step": 12309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5058981537054607e-05,
+      "loss": 0.8955,
+      "step": 12310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5055962594559923e-05,
+      "loss": 0.8633,
+      "step": 12311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.505294377204249e-05,
+      "loss": 0.9619,
+      "step": 12312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5049925069575557e-05,
+      "loss": 0.9355,
+      "step": 12313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.504690648723239e-05,
+      "loss": 0.9561,
+      "step": 12314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5043888025086233e-05,
+      "loss": 1.002,
+      "step": 12315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5040869683210337e-05,
+      "loss": 0.832,
+      "step": 12316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5037851461677955e-05,
+      "loss": 0.9004,
+      "step": 12317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.503483336056232e-05,
+      "loss": 0.8135,
+      "step": 12318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5031815379936683e-05,
+      "loss": 0.8838,
+      "step": 12319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.502879751987427e-05,
+      "loss": 0.8066,
+      "step": 12320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5025779780448326e-05,
+      "loss": 1.0029,
+      "step": 12321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5022762161732072e-05,
+      "loss": 0.9453,
+      "step": 12322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.501974466379875e-05,
+      "loss": 0.9463,
+      "step": 12323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5016727286721573e-05,
+      "loss": 1.0156,
+      "step": 12324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5013710030573773e-05,
+      "loss": 0.915,
+      "step": 12325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5010692895428563e-05,
+      "loss": 1.0,
+      "step": 12326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5007675881359168e-05,
+      "loss": 0.8477,
+      "step": 12327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5004658988438791e-05,
+      "loss": 0.7559,
+      "step": 12328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5001642216740658e-05,
+      "loss": 0.8408,
+      "step": 12329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4998625566337962e-05,
+      "loss": 0.9111,
+      "step": 12330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4995609037303918e-05,
+      "loss": 0.8936,
+      "step": 12331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4992592629711724e-05,
+      "loss": 0.8682,
+      "step": 12332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4989576343634572e-05,
+      "loss": 0.7373,
+      "step": 12333
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4986560179145681e-05,
+      "loss": 1.0332,
+      "step": 12334
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4983544136318222e-05,
+      "loss": 0.9707,
+      "step": 12335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.49805282152254e-05,
+      "loss": 0.9258,
+      "step": 12336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4977512415940392e-05,
+      "loss": 0.9316,
+      "step": 12337
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4974496738536393e-05,
+      "loss": 0.9199,
+      "step": 12338
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4971481183086572e-05,
+      "loss": 0.833,
+      "step": 12339
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4968465749664118e-05,
+      "loss": 0.9229,
+      "step": 12340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4965450438342202e-05,
+      "loss": 0.832,
+      "step": 12341
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4962435249193998e-05,
+      "loss": 0.7266,
+      "step": 12342
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4959420182292674e-05,
+      "loss": 0.9297,
+      "step": 12343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4956405237711403e-05,
+      "loss": 0.9238,
+      "step": 12344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4953390415523336e-05,
+      "loss": 0.8789,
+      "step": 12345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.495037571580165e-05,
+      "loss": 0.9434,
+      "step": 12346
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4947361138619492e-05,
+      "loss": 0.7783,
+      "step": 12347
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4944346684050021e-05,
+      "loss": 0.9326,
+      "step": 12348
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4941332352166385e-05,
+      "loss": 0.7471,
+      "step": 12349
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4938318143041742e-05,
+      "loss": 1.0479,
+      "step": 12350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4935304056749225e-05,
+      "loss": 0.9766,
+      "step": 12351
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4932290093361988e-05,
+      "loss": 0.8618,
+      "step": 12352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4929276252953161e-05,
+      "loss": 0.7529,
+      "step": 12353
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4926262535595894e-05,
+      "loss": 0.834,
+      "step": 12354
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4923248941363308e-05,
+      "loss": 0.9092,
+      "step": 12355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4920235470328541e-05,
+      "loss": 0.7549,
+      "step": 12356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4917222122564726e-05,
+      "loss": 0.9082,
+      "step": 12357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4914208898144976e-05,
+      "loss": 0.9248,
+      "step": 12358
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4911195797142424e-05,
+      "loss": 0.8525,
+      "step": 12359
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.490818281963018e-05,
+      "loss": 0.8848,
+      "step": 12360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4905169965681365e-05,
+      "loss": 0.9912,
+      "step": 12361
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4902157235369092e-05,
+      "loss": 0.8672,
+      "step": 12362
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4899144628766473e-05,
+      "loss": 0.8125,
+      "step": 12363
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4896132145946611e-05,
+      "loss": 0.9346,
+      "step": 12364
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4893119786982613e-05,
+      "loss": 0.8115,
+      "step": 12365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4890107551947574e-05,
+      "loss": 0.8477,
+      "step": 12366
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4887095440914605e-05,
+      "loss": 0.7383,
+      "step": 12367
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4884083453956788e-05,
+      "loss": 0.916,
+      "step": 12368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4881071591147222e-05,
+      "loss": 0.9463,
+      "step": 12369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4878059852558996e-05,
+      "loss": 0.874,
+      "step": 12370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4875048238265191e-05,
+      "loss": 0.9365,
+      "step": 12371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4872036748338894e-05,
+      "loss": 0.9531,
+      "step": 12372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4869025382853186e-05,
+      "loss": 0.7666,
+      "step": 12373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.486601414188114e-05,
+      "loss": 0.8174,
+      "step": 12374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4863003025495832e-05,
+      "loss": 0.7412,
+      "step": 12375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4859992033770334e-05,
+      "loss": 0.8506,
+      "step": 12376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4856981166777718e-05,
+      "loss": 0.8135,
+      "step": 12377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4853970424591035e-05,
+      "loss": 0.9795,
+      "step": 12378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4850959807283363e-05,
+      "loss": 1.0615,
+      "step": 12379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.484794931492775e-05,
+      "loss": 0.8428,
+      "step": 12380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4844938947597259e-05,
+      "loss": 0.8672,
+      "step": 12381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4841928705364934e-05,
+      "loss": 0.7783,
+      "step": 12382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4838918588303837e-05,
+      "loss": 0.9463,
+      "step": 12383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4835908596486998e-05,
+      "loss": 0.8105,
+      "step": 12384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.483289872998748e-05,
+      "loss": 0.7881,
+      "step": 12385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4829888988878307e-05,
+      "loss": 0.8682,
+      "step": 12386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4826879373232528e-05,
+      "loss": 0.7095,
+      "step": 12387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.482386988312317e-05,
+      "loss": 0.9092,
+      "step": 12388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4820860518623272e-05,
+      "loss": 0.9727,
+      "step": 12389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4817851279805852e-05,
+      "loss": 0.7891,
+      "step": 12390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4814842166743946e-05,
+      "loss": 1.0332,
+      "step": 12391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4811833179510569e-05,
+      "loss": 0.7422,
+      "step": 12392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4808824318178743e-05,
+      "loss": 0.8682,
+      "step": 12393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4805815582821483e-05,
+      "loss": 1.0459,
+      "step": 12394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4802806973511807e-05,
+      "loss": 0.9951,
+      "step": 12395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4799798490322719e-05,
+      "loss": 0.7666,
+      "step": 12396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4796790133327231e-05,
+      "loss": 0.9453,
+      "step": 12397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.479378190259834e-05,
+      "loss": 0.79,
+      "step": 12398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4790773798209052e-05,
+      "loss": 0.876,
+      "step": 12399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4787765820232364e-05,
+      "loss": 0.7314,
+      "step": 12400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4784757968741272e-05,
+      "loss": 0.7451,
+      "step": 12401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4781750243808767e-05,
+      "loss": 0.9639,
+      "step": 12402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4778742645507835e-05,
+      "loss": 1.0645,
+      "step": 12403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.477573517391147e-05,
+      "loss": 0.957,
+      "step": 12404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4772727829092644e-05,
+      "loss": 0.832,
+      "step": 12405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4769720611124342e-05,
+      "loss": 0.8447,
+      "step": 12406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4766713520079537e-05,
+      "loss": 0.9873,
+      "step": 12407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4763706556031206e-05,
+      "loss": 0.8818,
+      "step": 12408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4760699719052313e-05,
+      "loss": 0.8438,
+      "step": 12409
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4757693009215839e-05,
+      "loss": 1.0361,
+      "step": 12410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4754686426594732e-05,
+      "loss": 0.9229,
+      "step": 12411
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4751679971261961e-05,
+      "loss": 0.8398,
+      "step": 12412
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.474867364329048e-05,
+      "loss": 0.9629,
+      "step": 12413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4745667442753249e-05,
+      "loss": 1.0059,
+      "step": 12414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4742661369723214e-05,
+      "loss": 1.0469,
+      "step": 12415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4739655424273322e-05,
+      "loss": 0.9189,
+      "step": 12416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4736649606476532e-05,
+      "loss": 1.0264,
+      "step": 12417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.473364391640577e-05,
+      "loss": 0.8623,
+      "step": 12418
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4730638354133989e-05,
+      "loss": 0.8613,
+      "step": 12419
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4727632919734114e-05,
+      "loss": 0.8086,
+      "step": 12420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4724627613279086e-05,
+      "loss": 0.7603,
+      "step": 12421
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.472162243484183e-05,
+      "loss": 1.0762,
+      "step": 12422
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4718617384495275e-05,
+      "loss": 0.8672,
+      "step": 12423
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.471561246231234e-05,
+      "loss": 0.7495,
+      "step": 12424
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4712607668365956e-05,
+      "loss": 0.8857,
+      "step": 12425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4709603002729029e-05,
+      "loss": 0.7119,
+      "step": 12426
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4706598465474485e-05,
+      "loss": 0.9199,
+      "step": 12427
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4703594056675221e-05,
+      "loss": 0.8125,
+      "step": 12428
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.470058977640416e-05,
+      "loss": 0.9199,
+      "step": 12429
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4697585624734194e-05,
+      "loss": 0.8672,
+      "step": 12430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4694581601738236e-05,
+      "loss": 0.8691,
+      "step": 12431
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4691577707489177e-05,
+      "loss": 0.8164,
+      "step": 12432
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4688573942059918e-05,
+      "loss": 0.8555,
+      "step": 12433
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4685570305523346e-05,
+      "loss": 0.9082,
+      "step": 12434
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4682566797952356e-05,
+      "loss": 0.8047,
+      "step": 12435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4679563419419829e-05,
+      "loss": 0.8564,
+      "step": 12436
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4676560169998655e-05,
+      "loss": 0.9893,
+      "step": 12437
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4673557049761707e-05,
+      "loss": 0.9072,
+      "step": 12438
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4670554058781868e-05,
+      "loss": 0.7617,
+      "step": 12439
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4667551197132003e-05,
+      "loss": 0.9336,
+      "step": 12440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4664548464884991e-05,
+      "loss": 0.9248,
+      "step": 12441
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4661545862113698e-05,
+      "loss": 0.9648,
+      "step": 12442
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4658543388890984e-05,
+      "loss": 0.9111,
+      "step": 12443
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4655541045289718e-05,
+      "loss": 0.8838,
+      "step": 12444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.465253883138275e-05,
+      "loss": 0.7607,
+      "step": 12445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4649536747242941e-05,
+      "loss": 0.7329,
+      "step": 12446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4646534792943137e-05,
+      "loss": 0.8867,
+      "step": 12447
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4643532968556194e-05,
+      "loss": 0.7568,
+      "step": 12448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4640531274154946e-05,
+      "loss": 0.7109,
+      "step": 12449
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4637529709812249e-05,
+      "loss": 0.8389,
+      "step": 12450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4634528275600931e-05,
+      "loss": 0.8154,
+      "step": 12451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4631526971593836e-05,
+      "loss": 0.9541,
+      "step": 12452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4628525797863788e-05,
+      "loss": 0.9727,
+      "step": 12453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.462552475448363e-05,
+      "loss": 0.9697,
+      "step": 12454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4622523841526173e-05,
+      "loss": 0.7661,
+      "step": 12455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4619523059064254e-05,
+      "loss": 0.8555,
+      "step": 12456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.461652240717068e-05,
+      "loss": 0.9033,
+      "step": 12457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4613521885918282e-05,
+      "loss": 0.9551,
+      "step": 12458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4610521495379861e-05,
+      "loss": 0.8867,
+      "step": 12459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.460752123562824e-05,
+      "loss": 1.0186,
+      "step": 12460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4604521106736212e-05,
+      "loss": 0.9121,
+      "step": 12461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4601521108776594e-05,
+      "loss": 1.002,
+      "step": 12462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.459852124182218e-05,
+      "loss": 0.8916,
+      "step": 12463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4595521505945773e-05,
+      "loss": 0.8223,
+      "step": 12464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4592521901220163e-05,
+      "loss": 0.9824,
+      "step": 12465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4589522427718145e-05,
+      "loss": 1.0254,
+      "step": 12466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4586523085512502e-05,
+      "loss": 0.9336,
+      "step": 12467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4583523874676032e-05,
+      "loss": 0.8545,
+      "step": 12468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4580524795281499e-05,
+      "loss": 0.8223,
+      "step": 12469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4577525847401697e-05,
+      "loss": 0.9375,
+      "step": 12470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4574527031109394e-05,
+      "loss": 0.7881,
+      "step": 12471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4571528346477364e-05,
+      "loss": 0.9531,
+      "step": 12472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4568529793578374e-05,
+      "loss": 0.9004,
+      "step": 12473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4565531372485198e-05,
+      "loss": 0.8848,
+      "step": 12474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4562533083270586e-05,
+      "loss": 0.9082,
+      "step": 12475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4559534926007312e-05,
+      "loss": 0.7666,
+      "step": 12476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4556536900768122e-05,
+      "loss": 0.9307,
+      "step": 12477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4553539007625777e-05,
+      "loss": 0.7432,
+      "step": 12478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4550541246653016e-05,
+      "loss": 0.8496,
+      "step": 12479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4547543617922597e-05,
+      "loss": 0.9531,
+      "step": 12480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4544546121507258e-05,
+      "loss": 0.9258,
+      "step": 12481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4541548757479743e-05,
+      "loss": 0.8057,
+      "step": 12482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4538551525912783e-05,
+      "loss": 0.8652,
+      "step": 12483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.453555442687912e-05,
+      "loss": 1.0576,
+      "step": 12484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4532557460451475e-05,
+      "loss": 0.8291,
+      "step": 12485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4529560626702587e-05,
+      "loss": 0.8926,
+      "step": 12486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4526563925705172e-05,
+      "loss": 1.0264,
+      "step": 12487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4523567357531954e-05,
+      "loss": 0.9648,
+      "step": 12488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.452057092225565e-05,
+      "loss": 0.9512,
+      "step": 12489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4517574619948976e-05,
+      "loss": 0.8936,
+      "step": 12490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4514578450684645e-05,
+      "loss": 1.0469,
+      "step": 12491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4511582414535359e-05,
+      "loss": 0.9717,
+      "step": 12492
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.450858651157383e-05,
+      "loss": 0.8018,
+      "step": 12493
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4505590741872753e-05,
+      "loss": 0.9727,
+      "step": 12494
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4502595105504835e-05,
+      "loss": 0.8252,
+      "step": 12495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4499599602542762e-05,
+      "loss": 1.0,
+      "step": 12496
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4496604233059235e-05,
+      "loss": 0.9287,
+      "step": 12497
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4493608997126933e-05,
+      "loss": 0.8945,
+      "step": 12498
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4490613894818552e-05,
+      "loss": 0.8799,
+      "step": 12499
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4487618926206763e-05,
+      "loss": 1.0078,
+      "step": 12500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4484624091364254e-05,
+      "loss": 1.0322,
+      "step": 12501
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.44816293903637e-05,
+      "loss": 0.8828,
+      "step": 12502
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4478634823277771e-05,
+      "loss": 0.9385,
+      "step": 12503
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4475640390179139e-05,
+      "loss": 0.876,
+      "step": 12504
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4472646091140466e-05,
+      "loss": 0.9033,
+      "step": 12505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4469651926234418e-05,
+      "loss": 0.918,
+      "step": 12506
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4466657895533653e-05,
+      "loss": 0.835,
+      "step": 12507
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.446366399911083e-05,
+      "loss": 0.7549,
+      "step": 12508
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4460670237038596e-05,
+      "loss": 0.8223,
+      "step": 12509
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.445767660938961e-05,
+      "loss": 1.0029,
+      "step": 12510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4454683116236508e-05,
+      "loss": 0.6382,
+      "step": 12511
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4451689757651943e-05,
+      "loss": 0.8838,
+      "step": 12512
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4448696533708547e-05,
+      "loss": 0.8945,
+      "step": 12513
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4445703444478963e-05,
+      "loss": 0.9033,
+      "step": 12514
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4442710490035818e-05,
+      "loss": 0.8457,
+      "step": 12515
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4439717670451751e-05,
+      "loss": 0.9023,
+      "step": 12516
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4436724985799381e-05,
+      "loss": 0.9287,
+      "step": 12517
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4433732436151337e-05,
+      "loss": 0.8936,
+      "step": 12518
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4430740021580234e-05,
+      "loss": 0.9258,
+      "step": 12519
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4427747742158694e-05,
+      "loss": 0.9004,
+      "step": 12520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4424755597959328e-05,
+      "loss": 0.9209,
+      "step": 12521
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4421763589054747e-05,
+      "loss": 0.8965,
+      "step": 12522
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4418771715517559e-05,
+      "loss": 0.9375,
+      "step": 12523
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.441577997742037e-05,
+      "loss": 0.9268,
+      "step": 12524
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4412788374835773e-05,
+      "loss": 0.7207,
+      "step": 12525
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4409796907836375e-05,
+      "loss": 0.9609,
+      "step": 12526
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4406805576494765e-05,
+      "loss": 0.8945,
+      "step": 12527
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4403814380883533e-05,
+      "loss": 0.8936,
+      "step": 12528
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4400823321075273e-05,
+      "loss": 0.9023,
+      "step": 12529
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4397832397142557e-05,
+      "loss": 0.8818,
+      "step": 12530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.439484160915798e-05,
+      "loss": 0.9238,
+      "step": 12531
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4391850957194109e-05,
+      "loss": 0.8945,
+      "step": 12532
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4388860441323525e-05,
+      "loss": 0.9932,
+      "step": 12533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4385870061618793e-05,
+      "loss": 0.9199,
+      "step": 12534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4382879818152486e-05,
+      "loss": 0.9219,
+      "step": 12535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4379889710997166e-05,
+      "loss": 0.7871,
+      "step": 12536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4376899740225398e-05,
+      "loss": 0.8389,
+      "step": 12537
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.437390990590973e-05,
+      "loss": 0.8662,
+      "step": 12538
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4370920208122729e-05,
+      "loss": 1.042,
+      "step": 12539
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4367930646936937e-05,
+      "loss": 0.8594,
+      "step": 12540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4364941222424907e-05,
+      "loss": 0.9854,
+      "step": 12541
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4361951934659178e-05,
+      "loss": 0.7925,
+      "step": 12542
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4358962783712298e-05,
+      "loss": 1.0049,
+      "step": 12543
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4355973769656796e-05,
+      "loss": 0.9355,
+      "step": 12544
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4352984892565221e-05,
+      "loss": 0.9541,
+      "step": 12545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4349996152510087e-05,
+      "loss": 0.9277,
+      "step": 12546
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4347007549563934e-05,
+      "loss": 0.8652,
+      "step": 12547
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.434401908379928e-05,
+      "loss": 0.8887,
+      "step": 12548
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4341030755288655e-05,
+      "loss": 0.958,
+      "step": 12549
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4338042564104567e-05,
+      "loss": 0.9863,
+      "step": 12550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4335054510319538e-05,
+      "loss": 0.6687,
+      "step": 12551
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.433206659400607e-05,
+      "loss": 0.8027,
+      "step": 12552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4329078815236684e-05,
+      "loss": 0.9883,
+      "step": 12553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4326091174083872e-05,
+      "loss": 0.9814,
+      "step": 12554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4323103670620145e-05,
+      "loss": 0.7207,
+      "step": 12555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4320116304917992e-05,
+      "loss": 0.8574,
+      "step": 12556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4317129077049918e-05,
+      "loss": 0.9033,
+      "step": 12557
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4314141987088403e-05,
+      "loss": 0.8228,
+      "step": 12558
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4311155035105946e-05,
+      "loss": 0.7925,
+      "step": 12559
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4308168221175021e-05,
+      "loss": 0.9678,
+      "step": 12560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4305181545368119e-05,
+      "loss": 0.8135,
+      "step": 12561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4302195007757708e-05,
+      "loss": 0.8086,
+      "step": 12562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.429920860841627e-05,
+      "loss": 0.708,
+      "step": 12563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4296222347416273e-05,
+      "loss": 0.9268,
+      "step": 12564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4293236224830187e-05,
+      "loss": 1.0566,
+      "step": 12565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4290250240730475e-05,
+      "loss": 0.9043,
+      "step": 12566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4287264395189597e-05,
+      "loss": 0.9023,
+      "step": 12567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4284278688280009e-05,
+      "loss": 0.8857,
+      "step": 12568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4281293120074173e-05,
+      "loss": 0.9814,
+      "step": 12569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4278307690644528e-05,
+      "loss": 0.8311,
+      "step": 12570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4275322400063538e-05,
+      "loss": 1.0146,
+      "step": 12571
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.427233724840363e-05,
+      "loss": 0.8857,
+      "step": 12572
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4269352235737253e-05,
+      "loss": 0.9775,
+      "step": 12573
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4266367362136845e-05,
+      "loss": 0.9697,
+      "step": 12574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4263382627674838e-05,
+      "loss": 0.915,
+      "step": 12575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4260398032423668e-05,
+      "loss": 0.9775,
+      "step": 12576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4257413576455754e-05,
+      "loss": 0.7622,
+      "step": 12577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.425442925984353e-05,
+      "loss": 0.8096,
+      "step": 12578
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4251445082659404e-05,
+      "loss": 0.9248,
+      "step": 12579
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4248461044975805e-05,
+      "loss": 0.9814,
+      "step": 12580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4245477146865137e-05,
+      "loss": 0.9727,
+      "step": 12581
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4242493388399821e-05,
+      "loss": 0.8281,
+      "step": 12582
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4239509769652248e-05,
+      "loss": 0.9619,
+      "step": 12583
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4236526290694836e-05,
+      "loss": 0.8818,
+      "step": 12584
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4233542951599986e-05,
+      "loss": 0.9482,
+      "step": 12585
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4230559752440087e-05,
+      "loss": 0.8809,
+      "step": 12586
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.422757669328754e-05,
+      "loss": 0.9121,
+      "step": 12587
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4224593774214723e-05,
+      "loss": 0.8545,
+      "step": 12588
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4221610995294037e-05,
+      "loss": 1.0684,
+      "step": 12589
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4218628356597855e-05,
+      "loss": 0.9951,
+      "step": 12590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4215645858198564e-05,
+      "loss": 1.0078,
+      "step": 12591
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4212663500168531e-05,
+      "loss": 0.8564,
+      "step": 12592
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4209681282580139e-05,
+      "loss": 0.958,
+      "step": 12593
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4206699205505751e-05,
+      "loss": 0.8486,
+      "step": 12594
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4203717269017738e-05,
+      "loss": 0.8262,
+      "step": 12595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4200735473188458e-05,
+      "loss": 0.9727,
+      "step": 12596
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4197753818090278e-05,
+      "loss": 0.874,
+      "step": 12597
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4194772303795546e-05,
+      "loss": 0.9258,
+      "step": 12598
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4191790930376618e-05,
+      "loss": 0.9531,
+      "step": 12599
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4188809697905841e-05,
+      "loss": 0.9424,
+      "step": 12600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4185828606455565e-05,
+      "loss": 0.6929,
+      "step": 12601
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4182847656098129e-05,
+      "loss": 0.6001,
+      "step": 12602
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4179866846905874e-05,
+      "loss": 0.8438,
+      "step": 12603
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4176886178951131e-05,
+      "loss": 0.9697,
+      "step": 12604
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.417390565230624e-05,
+      "loss": 1.001,
+      "step": 12605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.417092526704352e-05,
+      "loss": 0.8916,
+      "step": 12606
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4167945023235303e-05,
+      "loss": 0.9316,
+      "step": 12607
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4164964920953908e-05,
+      "loss": 1.0225,
+      "step": 12608
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4161984960271658e-05,
+      "loss": 0.8838,
+      "step": 12609
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.415900514126086e-05,
+      "loss": 0.7949,
+      "step": 12610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.415602546399383e-05,
+      "loss": 0.7837,
+      "step": 12611
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4153045928542875e-05,
+      "loss": 0.9238,
+      "step": 12612
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.41500665349803e-05,
+      "loss": 0.9692,
+      "step": 12613
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.414708728337841e-05,
+      "loss": 0.8174,
+      "step": 12614
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4144108173809495e-05,
+      "loss": 0.9258,
+      "step": 12615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4141129206345854e-05,
+      "loss": 0.749,
+      "step": 12616
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4138150381059777e-05,
+      "loss": 0.8574,
+      "step": 12617
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.413517169802355e-05,
+      "loss": 0.9707,
+      "step": 12618
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4132193157309457e-05,
+      "loss": 0.7539,
+      "step": 12619
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4129214758989785e-05,
+      "loss": 0.8506,
+      "step": 12620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.41262365031368e-05,
+      "loss": 0.8594,
+      "step": 12621
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4123258389822785e-05,
+      "loss": 0.8887,
+      "step": 12622
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4120280419120003e-05,
+      "loss": 0.8535,
+      "step": 12623
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4117302591100726e-05,
+      "loss": 0.959,
+      "step": 12624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4114324905837214e-05,
+      "loss": 0.8887,
+      "step": 12625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4111347363401726e-05,
+      "loss": 0.8252,
+      "step": 12626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.410836996386652e-05,
+      "loss": 0.6396,
+      "step": 12627
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4105392707303851e-05,
+      "loss": 0.9756,
+      "step": 12628
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4102415593785961e-05,
+      "loss": 0.7891,
+      "step": 12629
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4099438623385104e-05,
+      "loss": 0.8018,
+      "step": 12630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4096461796173518e-05,
+      "loss": 0.7979,
+      "step": 12631
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4093485112223441e-05,
+      "loss": 0.9053,
+      "step": 12632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4090508571607112e-05,
+      "loss": 0.9404,
+      "step": 12633
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4087532174396761e-05,
+      "loss": 0.9111,
+      "step": 12634
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.408455592066461e-05,
+      "loss": 0.8936,
+      "step": 12635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.40815798104829e-05,
+      "loss": 1.0059,
+      "step": 12636
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4078603843923832e-05,
+      "loss": 0.8662,
+      "step": 12637
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4075628021059644e-05,
+      "loss": 0.813,
+      "step": 12638
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4072652341962534e-05,
+      "loss": 0.7632,
+      "step": 12639
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4069676806704724e-05,
+      "loss": 0.7041,
+      "step": 12640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4066701415358414e-05,
+      "loss": 0.8428,
+      "step": 12641
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4063726167995817e-05,
+      "loss": 0.9229,
+      "step": 12642
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4060751064689124e-05,
+      "loss": 0.9922,
+      "step": 12643
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.405777610551054e-05,
+      "loss": 0.8916,
+      "step": 12644
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4054801290532246e-05,
+      "loss": 0.7178,
+      "step": 12645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4051826619826447e-05,
+      "loss": 0.9854,
+      "step": 12646
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.404885209346532e-05,
+      "loss": 0.8223,
+      "step": 12647
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4045877711521056e-05,
+      "loss": 0.9951,
+      "step": 12648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4042903474065823e-05,
+      "loss": 0.9463,
+      "step": 12649
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4039929381171811e-05,
+      "loss": 0.6875,
+      "step": 12650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4036955432911177e-05,
+      "loss": 0.9785,
+      "step": 12651
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4033981629356104e-05,
+      "loss": 0.7246,
+      "step": 12652
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4031007970578748e-05,
+      "loss": 0.9062,
+      "step": 12653
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4028034456651276e-05,
+      "loss": 0.9521,
+      "step": 12654
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4025061087645842e-05,
+      "loss": 0.9795,
+      "step": 12655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4022087863634609e-05,
+      "loss": 0.8252,
+      "step": 12656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4019114784689717e-05,
+      "loss": 0.873,
+      "step": 12657
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4016141850883325e-05,
+      "loss": 0.5928,
+      "step": 12658
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4013169062287567e-05,
+      "loss": 1.0322,
+      "step": 12659
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.401019641897459e-05,
+      "loss": 0.8291,
+      "step": 12660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4007223921016537e-05,
+      "loss": 0.9336,
+      "step": 12661
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4004251568485529e-05,
+      "loss": 0.9482,
+      "step": 12662
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4001279361453707e-05,
+      "loss": 0.8418,
+      "step": 12663
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.399830729999319e-05,
+      "loss": 0.8613,
+      "step": 12664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3995335384176107e-05,
+      "loss": 1.0254,
+      "step": 12665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3992363614074569e-05,
+      "loss": 0.8516,
+      "step": 12666
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3989391989760698e-05,
+      "loss": 0.8818,
+      "step": 12667
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3986420511306617e-05,
+      "loss": 0.9258,
+      "step": 12668
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3983449178784416e-05,
+      "loss": 0.916,
+      "step": 12669
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3980477992266217e-05,
+      "loss": 0.8525,
+      "step": 12670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3977506951824109e-05,
+      "loss": 0.8018,
+      "step": 12671
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.39745360575302e-05,
+      "loss": 0.9775,
+      "step": 12672
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3971565309456577e-05,
+      "loss": 0.9404,
+      "step": 12673
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3968594707675335e-05,
+      "loss": 0.8252,
+      "step": 12674
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3965624252258563e-05,
+      "loss": 1.0293,
+      "step": 12675
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3962653943278346e-05,
+      "loss": 0.9541,
+      "step": 12676
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3959683780806756e-05,
+      "loss": 0.6562,
+      "step": 12677
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3956713764915883e-05,
+      "loss": 0.9043,
+      "step": 12678
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3953743895677791e-05,
+      "loss": 0.9004,
+      "step": 12679
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3950774173164559e-05,
+      "loss": 0.7603,
+      "step": 12680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.394780459744824e-05,
+      "loss": 0.8271,
+      "step": 12681
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.394483516860091e-05,
+      "loss": 0.9023,
+      "step": 12682
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3941865886694617e-05,
+      "loss": 0.7373,
+      "step": 12683
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.393889675180143e-05,
+      "loss": 0.8433,
+      "step": 12684
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3935927763993387e-05,
+      "loss": 0.9248,
+      "step": 12685
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3932958923342547e-05,
+      "loss": 0.7979,
+      "step": 12686
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.392999022992095e-05,
+      "loss": 0.9355,
+      "step": 12687
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3927021683800642e-05,
+      "loss": 0.7822,
+      "step": 12688
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3924053285053654e-05,
+      "loss": 0.8447,
+      "step": 12689
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3921085033752029e-05,
+      "loss": 0.8389,
+      "step": 12690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3918116929967785e-05,
+      "loss": 0.833,
+      "step": 12691
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3915148973772965e-05,
+      "loss": 1.0303,
+      "step": 12692
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3912181165239577e-05,
+      "loss": 0.9434,
+      "step": 12693
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3909213504439656e-05,
+      "loss": 0.8945,
+      "step": 12694
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3906245991445203e-05,
+      "loss": 0.7549,
+      "step": 12695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3903278626328244e-05,
+      "loss": 0.7788,
+      "step": 12696
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3900311409160781e-05,
+      "loss": 0.9014,
+      "step": 12697
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.389734434001482e-05,
+      "loss": 0.998,
+      "step": 12698
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.389437741896237e-05,
+      "loss": 1.0557,
+      "step": 12699
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3891410646075419e-05,
+      "loss": 0.9424,
+      "step": 12700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3888444021425969e-05,
+      "loss": 1.0107,
+      "step": 12701
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3885477545086008e-05,
+      "loss": 0.8281,
+      "step": 12702
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3882511217127528e-05,
+      "loss": 0.9922,
+      "step": 12703
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3879545037622504e-05,
+      "loss": 0.6367,
+      "step": 12704
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.387657900664293e-05,
+      "loss": 0.9473,
+      "step": 12705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.387361312426077e-05,
+      "loss": 0.9629,
+      "step": 12706
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3870647390548004e-05,
+      "loss": 0.9668,
+      "step": 12707
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3867681805576596e-05,
+      "loss": 0.9189,
+      "step": 12708
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3864716369418526e-05,
+      "loss": 0.6484,
+      "step": 12709
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3861751082145736e-05,
+      "loss": 0.9131,
+      "step": 12710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3858785943830201e-05,
+      "loss": 0.8467,
+      "step": 12711
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3855820954543866e-05,
+      "loss": 0.9385,
+      "step": 12712
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3852856114358693e-05,
+      "loss": 0.9707,
+      "step": 12713
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.384989142334662e-05,
+      "loss": 0.8545,
+      "step": 12714
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3846926881579598e-05,
+      "loss": 0.8828,
+      "step": 12715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3843962489129561e-05,
+      "loss": 0.9512,
+      "step": 12716
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3840998246068456e-05,
+      "loss": 0.8838,
+      "step": 12717
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3838034152468206e-05,
+      "loss": 0.8818,
+      "step": 12718
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3835070208400744e-05,
+      "loss": 0.8652,
+      "step": 12719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3832106413938e-05,
+      "loss": 0.9316,
+      "step": 12720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3829142769151897e-05,
+      "loss": 0.8447,
+      "step": 12721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3826179274114347e-05,
+      "loss": 0.9092,
+      "step": 12722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3823215928897273e-05,
+      "loss": 0.8564,
+      "step": 12723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3820252733572578e-05,
+      "loss": 0.8379,
+      "step": 12724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3817289688212181e-05,
+      "loss": 0.917,
+      "step": 12725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3814326792887976e-05,
+      "loss": 0.9072,
+      "step": 12726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3811364047671871e-05,
+      "loss": 0.7959,
+      "step": 12727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.380840145263576e-05,
+      "loss": 0.7275,
+      "step": 12728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3805439007851538e-05,
+      "loss": 0.7686,
+      "step": 12729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.380247671339109e-05,
+      "loss": 0.9453,
+      "step": 12730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3799514569326311e-05,
+      "loss": 0.8535,
+      "step": 12731
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3796552575729072e-05,
+      "loss": 0.8809,
+      "step": 12732
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3793590732671265e-05,
+      "loss": 0.8203,
+      "step": 12733
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3790629040224755e-05,
+      "loss": 0.583,
+      "step": 12734
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3787667498461418e-05,
+      "loss": 0.917,
+      "step": 12735
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.378470610745312e-05,
+      "loss": 1.0518,
+      "step": 12736
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3781744867271726e-05,
+      "loss": 0.7783,
+      "step": 12737
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3778783777989096e-05,
+      "loss": 0.8545,
+      "step": 12738
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3775822839677093e-05,
+      "loss": 0.8643,
+      "step": 12739
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3772862052407559e-05,
+      "loss": 0.8838,
+      "step": 12740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3769901416252354e-05,
+      "loss": 0.957,
+      "step": 12741
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3766940931283315e-05,
+      "loss": 0.8057,
+      "step": 12742
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3763980597572296e-05,
+      "loss": 0.7422,
+      "step": 12743
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.376102041519112e-05,
+      "loss": 0.9717,
+      "step": 12744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3758060384211631e-05,
+      "loss": 0.8174,
+      "step": 12745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3755100504705667e-05,
+      "loss": 0.707,
+      "step": 12746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.375214077674504e-05,
+      "loss": 0.8604,
+      "step": 12747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3749181200401588e-05,
+      "loss": 1.0059,
+      "step": 12748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3746221775747123e-05,
+      "loss": 0.9219,
+      "step": 12749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3743262502853458e-05,
+      "loss": 0.8252,
+      "step": 12750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.374030338179242e-05,
+      "loss": 0.874,
+      "step": 12751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3737344412635805e-05,
+      "loss": 0.8428,
+      "step": 12752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3734385595455428e-05,
+      "loss": 0.8057,
+      "step": 12753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3731426930323083e-05,
+      "loss": 0.998,
+      "step": 12754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3728468417310575e-05,
+      "loss": 0.9951,
+      "step": 12755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3725510056489688e-05,
+      "loss": 0.9346,
+      "step": 12756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3722551847932226e-05,
+      "loss": 0.8428,
+      "step": 12757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3719593791709963e-05,
+      "loss": 0.9355,
+      "step": 12758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3716635887894697e-05,
+      "loss": 1.0205,
+      "step": 12759
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3713678136558191e-05,
+      "loss": 0.8721,
+      "step": 12760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3710720537772235e-05,
+      "loss": 0.8164,
+      "step": 12761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.370776309160859e-05,
+      "loss": 0.9102,
+      "step": 12762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3704805798139034e-05,
+      "loss": 0.9287,
+      "step": 12763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3701848657435326e-05,
+      "loss": 0.8945,
+      "step": 12764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3698891669569232e-05,
+      "loss": 0.7676,
+      "step": 12765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3695934834612502e-05,
+      "loss": 0.9424,
+      "step": 12766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3692978152636898e-05,
+      "loss": 0.6289,
+      "step": 12767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3690021623714161e-05,
+      "loss": 0.9414,
+      "step": 12768
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3687065247916046e-05,
+      "loss": 0.9482,
+      "step": 12769
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3684109025314292e-05,
+      "loss": 1.0752,
+      "step": 12770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3681152955980636e-05,
+      "loss": 0.9814,
+      "step": 12771
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3678197039986813e-05,
+      "loss": 0.8086,
+      "step": 12772
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.367524127740456e-05,
+      "loss": 0.792,
+      "step": 12773
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3672285668305599e-05,
+      "loss": 0.9004,
+      "step": 12774
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3669330212761658e-05,
+      "loss": 0.7705,
+      "step": 12775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3666374910844451e-05,
+      "loss": 0.9062,
+      "step": 12776
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.36634197626257e-05,
+      "loss": 0.9072,
+      "step": 12777
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3660464768177115e-05,
+      "loss": 0.8877,
+      "step": 12778
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3657509927570409e-05,
+      "loss": 0.9004,
+      "step": 12779
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3654555240877281e-05,
+      "loss": 0.8213,
+      "step": 12780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3651600708169439e-05,
+      "loss": 0.8154,
+      "step": 12781
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3648646329518575e-05,
+      "loss": 1.0518,
+      "step": 12782
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.364569210499639e-05,
+      "loss": 0.8096,
+      "step": 12783
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3642738034674566e-05,
+      "loss": 0.9844,
+      "step": 12784
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3639784118624793e-05,
+      "loss": 0.9463,
+      "step": 12785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.363683035691876e-05,
+      "loss": 0.8369,
+      "step": 12786
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3633876749628137e-05,
+      "loss": 0.9648,
+      "step": 12787
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3630923296824605e-05,
+      "loss": 0.8086,
+      "step": 12788
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3627969998579831e-05,
+      "loss": 0.7148,
+      "step": 12789
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.362501685496549e-05,
+      "loss": 0.8193,
+      "step": 12790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3622063866053238e-05,
+      "loss": 0.9268,
+      "step": 12791
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3619111031914743e-05,
+      "loss": 0.9355,
+      "step": 12792
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3616158352621653e-05,
+      "loss": 0.9219,
+      "step": 12793
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.361320582824563e-05,
+      "loss": 0.8301,
+      "step": 12794
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3610253458858316e-05,
+      "loss": 1.1123,
+      "step": 12795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3607301244531362e-05,
+      "loss": 0.9102,
+      "step": 12796
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3604349185336401e-05,
+      "loss": 0.8545,
+      "step": 12797
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3601397281345082e-05,
+      "loss": 0.8594,
+      "step": 12798
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.359844553262903e-05,
+      "loss": 0.7407,
+      "step": 12799
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3595493939259881e-05,
+      "loss": 0.834,
+      "step": 12800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3592542501309256e-05,
+      "loss": 0.7295,
+      "step": 12801
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3589591218848783e-05,
+      "loss": 1.0547,
+      "step": 12802
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3586640091950076e-05,
+      "loss": 0.9404,
+      "step": 12803
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.358368912068476e-05,
+      "loss": 0.7168,
+      "step": 12804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3580738305124433e-05,
+      "loss": 0.833,
+      "step": 12805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3577787645340713e-05,
+      "loss": 0.9551,
+      "step": 12806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3574837141405195e-05,
+      "loss": 0.9316,
+      "step": 12807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3571886793389486e-05,
+      "loss": 0.8984,
+      "step": 12808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.356893660136518e-05,
+      "loss": 0.8584,
+      "step": 12809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3565986565403873e-05,
+      "loss": 0.96,
+      "step": 12810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3563036685577145e-05,
+      "loss": 0.8828,
+      "step": 12811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3560086961956591e-05,
+      "loss": 0.9746,
+      "step": 12812
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3557137394613782e-05,
+      "loss": 0.9033,
+      "step": 12813
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3554187983620306e-05,
+      "loss": 0.8506,
+      "step": 12814
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3551238729047726e-05,
+      "loss": 0.9717,
+      "step": 12815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3548289630967621e-05,
+      "loss": 0.9502,
+      "step": 12816
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3545340689451551e-05,
+      "loss": 0.8096,
+      "step": 12817
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3542391904571082e-05,
+      "loss": 0.793,
+      "step": 12818
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3539443276397766e-05,
+      "loss": 0.7637,
+      "step": 12819
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.353649480500317e-05,
+      "loss": 0.9219,
+      "step": 12820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3533546490458827e-05,
+      "loss": 0.7344,
+      "step": 12821
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.35305983328363e-05,
+      "loss": 0.8926,
+      "step": 12822
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.352765033220712e-05,
+      "loss": 0.9297,
+      "step": 12823
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3524702488642838e-05,
+      "loss": 0.7354,
+      "step": 12824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.352175480221498e-05,
+      "loss": 0.8833,
+      "step": 12825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3518807272995083e-05,
+      "loss": 0.9375,
+      "step": 12826
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.351585990105467e-05,
+      "loss": 0.958,
+      "step": 12827
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3512912686465272e-05,
+      "loss": 0.9277,
+      "step": 12828
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.35099656292984e-05,
+      "loss": 0.8496,
+      "step": 12829
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3507018729625579e-05,
+      "loss": 0.9551,
+      "step": 12830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3504071987518319e-05,
+      "loss": 0.7881,
+      "step": 12831
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3501125403048126e-05,
+      "loss": 1.0469,
+      "step": 12832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.34981789762865e-05,
+      "loss": 0.8623,
+      "step": 12833
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3495232707304959e-05,
+      "loss": 0.8467,
+      "step": 12834
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3492286596174988e-05,
+      "loss": 0.8857,
+      "step": 12835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3489340642968088e-05,
+      "loss": 0.8428,
+      "step": 12836
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3486394847755734e-05,
+      "loss": 0.7861,
+      "step": 12837
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3483449210609427e-05,
+      "loss": 0.6826,
+      "step": 12838
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3480503731600641e-05,
+      "loss": 0.7979,
+      "step": 12839
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3477558410800859e-05,
+      "loss": 0.8418,
+      "step": 12840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3474613248281552e-05,
+      "loss": 0.915,
+      "step": 12841
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3471668244114192e-05,
+      "loss": 0.9365,
+      "step": 12842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3468723398370239e-05,
+      "loss": 1.0352,
+      "step": 12843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3465778711121167e-05,
+      "loss": 0.7363,
+      "step": 12844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3462834182438424e-05,
+      "loss": 0.8184,
+      "step": 12845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3459889812393479e-05,
+      "loss": 0.8877,
+      "step": 12846
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3456945601057766e-05,
+      "loss": 0.8145,
+      "step": 12847
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3454001548502746e-05,
+      "loss": 0.7754,
+      "step": 12848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3451057654799851e-05,
+      "loss": 0.8306,
+      "step": 12849
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3448113920020536e-05,
+      "loss": 0.9365,
+      "step": 12850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.344517034423622e-05,
+      "loss": 0.7944,
+      "step": 12851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3442226927518348e-05,
+      "loss": 0.8848,
+      "step": 12852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3439283669938336e-05,
+      "loss": 0.8818,
+      "step": 12853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3436340571567623e-05,
+      "loss": 0.8047,
+      "step": 12854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3433397632477614e-05,
+      "loss": 0.8516,
+      "step": 12855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.343045485273974e-05,
+      "loss": 0.7832,
+      "step": 12856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.34275122324254e-05,
+      "loss": 0.8408,
+      "step": 12857
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3424569771606014e-05,
+      "loss": 0.8936,
+      "step": 12858
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3421627470352978e-05,
+      "loss": 1.0186,
+      "step": 12859
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.34186853287377e-05,
+      "loss": 0.8184,
+      "step": 12860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.341574334683157e-05,
+      "loss": 0.6479,
+      "step": 12861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3412801524705992e-05,
+      "loss": 0.7432,
+      "step": 12862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3409859862432341e-05,
+      "loss": 0.8975,
+      "step": 12863
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.340691836008202e-05,
+      "loss": 0.8203,
+      "step": 12864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.340397701772639e-05,
+      "loss": 0.8379,
+      "step": 12865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3401035835436846e-05,
+      "loss": 0.833,
+      "step": 12866
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3398094813284754e-05,
+      "loss": 0.8965,
+      "step": 12867
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3395153951341487e-05,
+      "loss": 0.8896,
+      "step": 12868
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3392213249678405e-05,
+      "loss": 0.7446,
+      "step": 12869
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3389272708366878e-05,
+      "loss": 0.6436,
+      "step": 12870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3386332327478262e-05,
+      "loss": 0.875,
+      "step": 12871
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3383392107083909e-05,
+      "loss": 0.7124,
+      "step": 12872
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.338045204725517e-05,
+      "loss": 0.8271,
+      "step": 12873
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.337751214806339e-05,
+      "loss": 0.9307,
+      "step": 12874
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.337457240957992e-05,
+      "loss": 1.0137,
+      "step": 12875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3371632831876089e-05,
+      "loss": 0.7612,
+      "step": 12876
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3368693415023239e-05,
+      "loss": 0.7812,
+      "step": 12877
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3365754159092692e-05,
+      "loss": 0.8145,
+      "step": 12878
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.336281506415579e-05,
+      "loss": 0.8672,
+      "step": 12879
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3359876130283837e-05,
+      "loss": 0.8867,
+      "step": 12880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3356937357548169e-05,
+      "loss": 0.9043,
+      "step": 12881
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3353998746020093e-05,
+      "loss": 0.8848,
+      "step": 12882
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3351060295770924e-05,
+      "loss": 0.8008,
+      "step": 12883
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3348122006871963e-05,
+      "loss": 0.8809,
+      "step": 12884
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3345183879394523e-05,
+      "loss": 0.7432,
+      "step": 12885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3342245913409897e-05,
+      "loss": 0.8809,
+      "step": 12886
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3339308108989386e-05,
+      "loss": 0.958,
+      "step": 12887
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3336370466204274e-05,
+      "loss": 0.96,
+      "step": 12888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3333432985125858e-05,
+      "loss": 1.0654,
+      "step": 12889
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3330495665825414e-05,
+      "loss": 0.7617,
+      "step": 12890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.332755850837423e-05,
+      "loss": 0.9189,
+      "step": 12891
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3324621512843574e-05,
+      "loss": 0.8662,
+      "step": 12892
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3321684679304724e-05,
+      "loss": 0.8301,
+      "step": 12893
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3318748007828947e-05,
+      "loss": 0.9531,
+      "step": 12894
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3315811498487508e-05,
+      "loss": 0.9238,
+      "step": 12895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3312875151351663e-05,
+      "loss": 0.8311,
+      "step": 12896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3309938966492677e-05,
+      "loss": 0.9326,
+      "step": 12897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3307002943981793e-05,
+      "loss": 0.7686,
+      "step": 12898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3304067083890268e-05,
+      "loss": 1.0859,
+      "step": 12899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3301131386289339e-05,
+      "loss": 0.8936,
+      "step": 12900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3298195851250255e-05,
+      "loss": 0.9453,
+      "step": 12901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3295260478844242e-05,
+      "loss": 0.9395,
+      "step": 12902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3292325269142547e-05,
+      "loss": 0.8652,
+      "step": 12903
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3289390222216384e-05,
+      "loss": 0.8242,
+      "step": 12904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.328645533813699e-05,
+      "loss": 0.8818,
+      "step": 12905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3283520616975579e-05,
+      "loss": 0.9365,
+      "step": 12906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3280586058803371e-05,
+      "loss": 0.9395,
+      "step": 12907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3277651663691576e-05,
+      "loss": 0.8906,
+      "step": 12908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3274717431711407e-05,
+      "loss": 0.689,
+      "step": 12909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3271783362934067e-05,
+      "loss": 0.959,
+      "step": 12910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3268849457430761e-05,
+      "loss": 0.9014,
+      "step": 12911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3265915715272678e-05,
+      "loss": 0.9316,
+      "step": 12912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3262982136531022e-05,
+      "loss": 0.8877,
+      "step": 12913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3260048721276972e-05,
+      "loss": 0.7939,
+      "step": 12914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3257115469581723e-05,
+      "loss": 0.9307,
+      "step": 12915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3254182381516436e-05,
+      "loss": 0.9258,
+      "step": 12916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3251249457152318e-05,
+      "loss": 0.8301,
+      "step": 12917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3248316696560526e-05,
+      "loss": 0.9238,
+      "step": 12918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3245384099812233e-05,
+      "loss": 0.7646,
+      "step": 12919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.32424516669786e-05,
+      "loss": 0.7852,
+      "step": 12920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3239519398130796e-05,
+      "loss": 0.6479,
+      "step": 12921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3236587293339974e-05,
+      "loss": 1.0439,
+      "step": 12922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3233655352677286e-05,
+      "loss": 0.9189,
+      "step": 12923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3230723576213881e-05,
+      "loss": 0.9609,
+      "step": 12924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3227791964020912e-05,
+      "loss": 0.9561,
+      "step": 12925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3224860516169513e-05,
+      "loss": 0.8252,
+      "step": 12926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3221929232730827e-05,
+      "loss": 0.9033,
+      "step": 12927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.321899811377598e-05,
+      "loss": 0.8896,
+      "step": 12928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3216067159376112e-05,
+      "loss": 0.8389,
+      "step": 12929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.321313636960234e-05,
+      "loss": 0.9316,
+      "step": 12930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3210205744525792e-05,
+      "loss": 0.8711,
+      "step": 12931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3207275284217578e-05,
+      "loss": 0.9209,
+      "step": 12932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.320434498874882e-05,
+      "loss": 0.9893,
+      "step": 12933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.320141485819062e-05,
+      "loss": 0.9756,
+      "step": 12934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3198484892614091e-05,
+      "loss": 0.8525,
+      "step": 12935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3195555092090326e-05,
+      "loss": 0.9102,
+      "step": 12936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3192625456690436e-05,
+      "loss": 0.9883,
+      "step": 12937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.31896959864855e-05,
+      "loss": 0.8652,
+      "step": 12938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3186766681546619e-05,
+      "loss": 0.8721,
+      "step": 12939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3183837541944868e-05,
+      "loss": 0.8252,
+      "step": 12940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3180908567751337e-05,
+      "loss": 0.8213,
+      "step": 12941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.31779797590371e-05,
+      "loss": 1.0732,
+      "step": 12942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3175051115873233e-05,
+      "loss": 0.7896,
+      "step": 12943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.31721226383308e-05,
+      "loss": 1.0498,
+      "step": 12944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3169194326480877e-05,
+      "loss": 0.8711,
+      "step": 12945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3166266180394514e-05,
+      "loss": 0.8574,
+      "step": 12946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3163338200142776e-05,
+      "loss": 0.9121,
+      "step": 12947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3160410385796712e-05,
+      "loss": 0.8975,
+      "step": 12948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3157482737427377e-05,
+      "loss": 0.6226,
+      "step": 12949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3154555255105808e-05,
+      "loss": 0.7979,
+      "step": 12950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3151627938903056e-05,
+      "loss": 0.8662,
+      "step": 12951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3148700788890148e-05,
+      "loss": 1.0244,
+      "step": 12952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3145773805138128e-05,
+      "loss": 1.0098,
+      "step": 12953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3142846987718016e-05,
+      "loss": 0.9766,
+      "step": 12954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3139920336700844e-05,
+      "loss": 0.9756,
+      "step": 12955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3136993852157631e-05,
+      "loss": 0.9326,
+      "step": 12956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3134067534159392e-05,
+      "loss": 0.9697,
+      "step": 12957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3131141382777144e-05,
+      "loss": 0.7578,
+      "step": 12958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3128215398081889e-05,
+      "loss": 0.7573,
+      "step": 12959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3125289580144644e-05,
+      "loss": 0.9756,
+      "step": 12960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.31223639290364e-05,
+      "loss": 0.9121,
+      "step": 12961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3119438444828159e-05,
+      "loss": 0.8936,
+      "step": 12962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3116513127590907e-05,
+      "loss": 0.9229,
+      "step": 12963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3113587977395645e-05,
+      "loss": 0.8994,
+      "step": 12964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3110662994313343e-05,
+      "loss": 0.8682,
+      "step": 12965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3107738178414997e-05,
+      "loss": 0.8535,
+      "step": 12966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3104813529771567e-05,
+      "loss": 0.7832,
+      "step": 12967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3101889048454044e-05,
+      "loss": 0.8486,
+      "step": 12968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3098964734533381e-05,
+      "loss": 0.7715,
+      "step": 12969
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3096040588080553e-05,
+      "loss": 0.8691,
+      "step": 12970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.309311660916651e-05,
+      "loss": 0.8535,
+      "step": 12971
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.309019279786222e-05,
+      "loss": 0.8901,
+      "step": 12972
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3087269154238626e-05,
+      "loss": 0.8828,
+      "step": 12973
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3084345678366686e-05,
+      "loss": 0.9502,
+      "step": 12974
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3081422370317331e-05,
+      "loss": 0.96,
+      "step": 12975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3078499230161512e-05,
+      "loss": 0.8408,
+      "step": 12976
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3075576257970159e-05,
+      "loss": 0.8877,
+      "step": 12977
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3072653453814211e-05,
+      "loss": 0.7979,
+      "step": 12978
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3069730817764586e-05,
+      "loss": 0.9102,
+      "step": 12979
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3066808349892214e-05,
+      "loss": 0.9629,
+      "step": 12980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3063886050268013e-05,
+      "loss": 0.832,
+      "step": 12981
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3060963918962903e-05,
+      "loss": 0.8311,
+      "step": 12982
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3058041956047788e-05,
+      "loss": 0.9102,
+      "step": 12983
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3055120161593582e-05,
+      "loss": 0.9248,
+      "step": 12984
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3052198535671181e-05,
+      "loss": 0.9141,
+      "step": 12985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3049277078351494e-05,
+      "loss": 0.9131,
+      "step": 12986
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3046355789705408e-05,
+      "loss": 0.9307,
+      "step": 12987
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3043434669803818e-05,
+      "loss": 0.7822,
+      "step": 12988
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3040513718717609e-05,
+      "loss": 0.9199,
+      "step": 12989
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3037592936517669e-05,
+      "loss": 0.9209,
+      "step": 12990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3034672323274867e-05,
+      "loss": 0.6592,
+      "step": 12991
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3031751879060087e-05,
+      "loss": 0.7666,
+      "step": 12992
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3028831603944193e-05,
+      "loss": 0.8296,
+      "step": 12993
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3025911497998058e-05,
+      "loss": 0.9131,
+      "step": 12994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3022991561292538e-05,
+      "loss": 0.8525,
+      "step": 12995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3020071793898496e-05,
+      "loss": 0.9609,
+      "step": 12996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3017152195886782e-05,
+      "loss": 1.0117,
+      "step": 12997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3014232767328254e-05,
+      "loss": 0.9629,
+      "step": 12998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3011313508293746e-05,
+      "loss": 1.0029,
+      "step": 12999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3008394418854102e-05,
+      "loss": 0.9014,
+      "step": 13000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3005475499080169e-05,
+      "loss": 0.8184,
+      "step": 13001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3002556749042778e-05,
+      "loss": 0.8037,
+      "step": 13002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2999638168812754e-05,
+      "loss": 0.793,
+      "step": 13003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2996719758460925e-05,
+      "loss": 0.8086,
+      "step": 13004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.299380151805811e-05,
+      "loss": 1.0703,
+      "step": 13005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2990883447675134e-05,
+      "loss": 0.8438,
+      "step": 13006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2987965547382797e-05,
+      "loss": 0.9072,
+      "step": 13007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2985047817251916e-05,
+      "loss": 1.0684,
+      "step": 13008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2982130257353296e-05,
+      "loss": 0.918,
+      "step": 13009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2979212867757735e-05,
+      "loss": 0.8799,
+      "step": 13010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2976295648536029e-05,
+      "loss": 1.0303,
+      "step": 13011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2973378599758975e-05,
+      "loss": 0.7217,
+      "step": 13012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2970461721497354e-05,
+      "loss": 1.0029,
+      "step": 13013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2967545013821955e-05,
+      "loss": 0.6162,
+      "step": 13014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2964628476803558e-05,
+      "loss": 0.9922,
+      "step": 13015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2961712110512937e-05,
+      "loss": 0.8003,
+      "step": 13016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2958795915020863e-05,
+      "loss": 0.7764,
+      "step": 13017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2955879890398108e-05,
+      "loss": 0.8252,
+      "step": 13018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2952964036715427e-05,
+      "loss": 0.8877,
+      "step": 13019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.295004835404359e-05,
+      "loss": 0.9092,
+      "step": 13020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2947132842453341e-05,
+      "loss": 0.957,
+      "step": 13021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2944217502015439e-05,
+      "loss": 0.8955,
+      "step": 13022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2941302332800625e-05,
+      "loss": 0.8086,
+      "step": 13023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2938387334879652e-05,
+      "loss": 0.6943,
+      "step": 13024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2935472508323242e-05,
+      "loss": 0.9043,
+      "step": 13025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2932557853202145e-05,
+      "loss": 0.8193,
+      "step": 13026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2929643369587077e-05,
+      "loss": 0.9551,
+      "step": 13027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.292672905754878e-05,
+      "loss": 0.959,
+      "step": 13028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.292381491715796e-05,
+      "loss": 0.8643,
+      "step": 13029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2920900948485345e-05,
+      "loss": 0.7021,
+      "step": 13030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2917987151601644e-05,
+      "loss": 0.9375,
+      "step": 13031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.291507352657757e-05,
+      "loss": 0.8965,
+      "step": 13032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2912160073483819e-05,
+      "loss": 0.9141,
+      "step": 13033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2909246792391105e-05,
+      "loss": 0.7686,
+      "step": 13034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2906333683370114e-05,
+      "loss": 0.8369,
+      "step": 13035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2903420746491548e-05,
+      "loss": 0.8857,
+      "step": 13036
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2900507981826085e-05,
+      "loss": 0.9922,
+      "step": 13037
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2897595389444417e-05,
+      "loss": 0.793,
+      "step": 13038
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2894682969417219e-05,
+      "loss": 0.8252,
+      "step": 13039
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2891770721815173e-05,
+      "loss": 0.8145,
+      "step": 13040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2888858646708943e-05,
+      "loss": 0.8057,
+      "step": 13041
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.28859467441692e-05,
+      "loss": 0.9834,
+      "step": 13042
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2883035014266614e-05,
+      "loss": 0.873,
+      "step": 13043
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2880123457071831e-05,
+      "loss": 0.7119,
+      "step": 13044
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2877212072655519e-05,
+      "loss": 0.7686,
+      "step": 13045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2874300861088319e-05,
+      "loss": 0.9316,
+      "step": 13046
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2871389822440882e-05,
+      "loss": 0.7705,
+      "step": 13047
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2868478956783847e-05,
+      "loss": 0.9619,
+      "step": 13048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.286556826418786e-05,
+      "loss": 0.8506,
+      "step": 13049
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2862657744723546e-05,
+      "loss": 0.8281,
+      "step": 13050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2859747398461542e-05,
+      "loss": 0.9092,
+      "step": 13051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2856837225472464e-05,
+      "loss": 0.8926,
+      "step": 13052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2853927225826945e-05,
+      "loss": 0.7812,
+      "step": 13053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2851017399595592e-05,
+      "loss": 0.7559,
+      "step": 13054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2848107746849028e-05,
+      "loss": 0.7681,
+      "step": 13055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2845198267657847e-05,
+      "loss": 0.9072,
+      "step": 13056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.284228896209267e-05,
+      "loss": 0.7617,
+      "step": 13057
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2839379830224089e-05,
+      "loss": 0.9141,
+      "step": 13058
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.28364708721227e-05,
+      "loss": 0.8135,
+      "step": 13059
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2833562087859093e-05,
+      "loss": 1.002,
+      "step": 13060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.283065347750386e-05,
+      "loss": 0.7476,
+      "step": 13061
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2827745041127584e-05,
+      "loss": 0.6865,
+      "step": 13062
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2824836778800845e-05,
+      "loss": 0.9229,
+      "step": 13063
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.282192869059421e-05,
+      "loss": 0.731,
+      "step": 13064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2819020776578261e-05,
+      "loss": 0.9355,
+      "step": 13065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2816113036823556e-05,
+      "loss": 0.7979,
+      "step": 13066
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2813205471400662e-05,
+      "loss": 0.9473,
+      "step": 13067
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2810298080380135e-05,
+      "loss": 0.8838,
+      "step": 13068
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2807390863832533e-05,
+      "loss": 0.9951,
+      "step": 13069
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2804483821828395e-05,
+      "loss": 0.8105,
+      "step": 13070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2801576954438279e-05,
+      "loss": 0.8721,
+      "step": 13071
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.279867026173272e-05,
+      "loss": 0.9453,
+      "step": 13072
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2795763743782254e-05,
+      "loss": 0.9229,
+      "step": 13073
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2792857400657414e-05,
+      "loss": 0.9033,
+      "step": 13074
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2789951232428733e-05,
+      "loss": 0.8018,
+      "step": 13075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2787045239166727e-05,
+      "loss": 0.9336,
+      "step": 13076
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2784139420941926e-05,
+      "loss": 0.998,
+      "step": 13077
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2781233777824835e-05,
+      "loss": 0.8867,
+      "step": 13078
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2778328309885973e-05,
+      "loss": 0.8594,
+      "step": 13079
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2775423017195842e-05,
+      "loss": 0.8984,
+      "step": 13080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2772517899824952e-05,
+      "loss": 0.7676,
+      "step": 13081
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2769612957843793e-05,
+      "loss": 0.8154,
+      "step": 13082
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.276670819132286e-05,
+      "loss": 1.0439,
+      "step": 13083
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2763803600332651e-05,
+      "loss": 0.9453,
+      "step": 13084
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2760899184943653e-05,
+      "loss": 0.8047,
+      "step": 13085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2757994945226337e-05,
+      "loss": 0.7109,
+      "step": 13086
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2755090881251188e-05,
+      "loss": 0.7744,
+      "step": 13087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2752186993088676e-05,
+      "loss": 0.9365,
+      "step": 13088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2749283280809273e-05,
+      "loss": 0.8945,
+      "step": 13089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2746379744483437e-05,
+      "loss": 0.7612,
+      "step": 13090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2743476384181636e-05,
+      "loss": 0.8916,
+      "step": 13091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.274057319997432e-05,
+      "loss": 0.8027,
+      "step": 13092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2737670191931946e-05,
+      "loss": 0.8184,
+      "step": 13093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2734767360124955e-05,
+      "loss": 0.9678,
+      "step": 13094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2731864704623797e-05,
+      "loss": 0.8857,
+      "step": 13095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2728962225498906e-05,
+      "loss": 0.9707,
+      "step": 13096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2726059922820722e-05,
+      "loss": 0.8994,
+      "step": 13097
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2723157796659669e-05,
+      "loss": 0.8584,
+      "step": 13098
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2720255847086176e-05,
+      "loss": 0.8428,
+      "step": 13099
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2717354074170662e-05,
+      "loss": 0.8096,
+      "step": 13100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2714452477983552e-05,
+      "loss": 0.8394,
+      "step": 13101
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.271155105859525e-05,
+      "loss": 0.832,
+      "step": 13102
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2708649816076175e-05,
+      "loss": 0.9863,
+      "step": 13103
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2705748750496718e-05,
+      "loss": 0.998,
+      "step": 13104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2702847861927295e-05,
+      "loss": 0.9844,
+      "step": 13105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2699947150438288e-05,
+      "loss": 0.9434,
+      "step": 13106
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.26970466161001e-05,
+      "loss": 0.9629,
+      "step": 13107
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2694146258983109e-05,
+      "loss": 0.7866,
+      "step": 13108
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2691246079157706e-05,
+      "loss": 0.7871,
+      "step": 13109
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2688346076694264e-05,
+      "loss": 0.8965,
+      "step": 13110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2685446251663163e-05,
+      "loss": 0.8418,
+      "step": 13111
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2682546604134767e-05,
+      "loss": 1.0029,
+      "step": 13112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2679647134179445e-05,
+      "loss": 0.9297,
+      "step": 13113
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2676747841867561e-05,
+      "loss": 0.9658,
+      "step": 13114
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2673848727269468e-05,
+      "loss": 1.0371,
+      "step": 13115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.267094979045552e-05,
+      "loss": 0.8477,
+      "step": 13116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2668051031496069e-05,
+      "loss": 1.0,
+      "step": 13117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2665152450461455e-05,
+      "loss": 0.8486,
+      "step": 13118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2662254047422024e-05,
+      "loss": 0.9609,
+      "step": 13119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2659355822448104e-05,
+      "loss": 0.707,
+      "step": 13120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2656457775610034e-05,
+      "loss": 0.7793,
+      "step": 13121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.265355990697813e-05,
+      "loss": 0.9219,
+      "step": 13122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2650662216622728e-05,
+      "loss": 0.7769,
+      "step": 13123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2647764704614138e-05,
+      "loss": 0.876,
+      "step": 13124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.264486737102268e-05,
+      "loss": 0.8398,
+      "step": 13125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2641970215918656e-05,
+      "loss": 0.9541,
+      "step": 13126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2639073239372373e-05,
+      "loss": 0.9756,
+      "step": 13127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.263617644145414e-05,
+      "loss": 0.8662,
+      "step": 13128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2633279822234247e-05,
+      "loss": 0.9707,
+      "step": 13129
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2630383381782991e-05,
+      "loss": 1.0039,
+      "step": 13130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2627487120170651e-05,
+      "loss": 0.9111,
+      "step": 13131
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2624591037467521e-05,
+      "loss": 0.8672,
+      "step": 13132
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2621695133743874e-05,
+      "loss": 0.8125,
+      "step": 13133
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.261879940906999e-05,
+      "loss": 0.6807,
+      "step": 13134
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2615903863516134e-05,
+      "loss": 0.9189,
+      "step": 13135
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2613008497152581e-05,
+      "loss": 0.7812,
+      "step": 13136
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.261011331004958e-05,
+      "loss": 0.8604,
+      "step": 13137
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2607218302277401e-05,
+      "loss": 0.9355,
+      "step": 13138
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2604323473906291e-05,
+      "loss": 0.9092,
+      "step": 13139
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.26014288250065e-05,
+      "loss": 0.9814,
+      "step": 13140
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2598534355648273e-05,
+      "loss": 0.8711,
+      "step": 13141
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2595640065901853e-05,
+      "loss": 0.7979,
+      "step": 13142
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2592745955837469e-05,
+      "loss": 0.8662,
+      "step": 13143
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2589852025525361e-05,
+      "loss": 0.7935,
+      "step": 13144
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2586958275035745e-05,
+      "loss": 0.8301,
+      "step": 13145
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2584064704438859e-05,
+      "loss": 1.1309,
+      "step": 13146
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2581171313804907e-05,
+      "loss": 0.916,
+      "step": 13147
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2578278103204113e-05,
+      "loss": 0.8535,
+      "step": 13148
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2575385072706681e-05,
+      "loss": 0.9609,
+      "step": 13149
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.257249222238282e-05,
+      "loss": 0.8149,
+      "step": 13150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2569599552302725e-05,
+      "loss": 0.8408,
+      "step": 13151
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2566707062536602e-05,
+      "loss": 0.7666,
+      "step": 13152
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2563814753154632e-05,
+      "loss": 0.792,
+      "step": 13153
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2560922624227016e-05,
+      "loss": 0.9629,
+      "step": 13154
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2558030675823925e-05,
+      "loss": 0.8018,
+      "step": 13155
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.255513890801555e-05,
+      "loss": 1.001,
+      "step": 13156
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2552247320872052e-05,
+      "loss": 0.918,
+      "step": 13157
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2549355914463613e-05,
+      "loss": 1.0586,
+      "step": 13158
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.254646468886039e-05,
+      "loss": 0.8369,
+      "step": 13159
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2543573644132553e-05,
+      "loss": 0.8135,
+      "step": 13160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2540682780350252e-05,
+      "loss": 0.7568,
+      "step": 13161
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2537792097583647e-05,
+      "loss": 0.8145,
+      "step": 13162
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2534901595902878e-05,
+      "loss": 0.875,
+      "step": 13163
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2532011275378095e-05,
+      "loss": 0.9824,
+      "step": 13164
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2529121136079434e-05,
+      "loss": 0.9443,
+      "step": 13165
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2526231178077029e-05,
+      "loss": 0.9287,
+      "step": 13166
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2523341401441016e-05,
+      "loss": 0.8223,
+      "step": 13167
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2520451806241525e-05,
+      "loss": 0.7949,
+      "step": 13168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2517562392548666e-05,
+      "loss": 0.9639,
+      "step": 13169
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2514673160432567e-05,
+      "loss": 0.917,
+      "step": 13170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2511784109963334e-05,
+      "loss": 0.8408,
+      "step": 13171
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2508895241211081e-05,
+      "loss": 0.9551,
+      "step": 13172
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2506006554245906e-05,
+      "loss": 0.793,
+      "step": 13173
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2503118049137916e-05,
+      "loss": 1.0674,
+      "step": 13174
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2500229725957203e-05,
+      "loss": 0.8848,
+      "step": 13175
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2497341584773861e-05,
+      "loss": 0.8477,
+      "step": 13176
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.249445362565797e-05,
+      "loss": 0.8379,
+      "step": 13177
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2491565848679618e-05,
+      "loss": 0.7617,
+      "step": 13178
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2488678253908881e-05,
+      "loss": 0.8174,
+      "step": 13179
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2485790841415834e-05,
+      "loss": 0.8955,
+      "step": 13180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2482903611270544e-05,
+      "loss": 0.7202,
+      "step": 13181
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2480016563543078e-05,
+      "loss": 0.8643,
+      "step": 13182
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.247712969830349e-05,
+      "loss": 0.9355,
+      "step": 13183
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2474243015621848e-05,
+      "loss": 0.9658,
+      "step": 13184
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2471356515568188e-05,
+      "loss": 0.9434,
+      "step": 13185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2468470198212568e-05,
+      "loss": 0.9385,
+      "step": 13186
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2465584063625025e-05,
+      "loss": 1.0273,
+      "step": 13187
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2462698111875601e-05,
+      "loss": 0.9033,
+      "step": 13188
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2459812343034324e-05,
+      "loss": 0.8984,
+      "step": 13189
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2456926757171232e-05,
+      "loss": 0.834,
+      "step": 13190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2454041354356339e-05,
+      "loss": 0.9814,
+      "step": 13191
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2451156134659675e-05,
+      "loss": 0.9609,
+      "step": 13192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2448271098151245e-05,
+      "loss": 0.8154,
+      "step": 13193
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2445386244901074e-05,
+      "loss": 0.9043,
+      "step": 13194
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2442501574979156e-05,
+      "loss": 0.7559,
+      "step": 13195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2439617088455502e-05,
+      "loss": 0.7549,
+      "step": 13196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2436732785400106e-05,
+      "loss": 1.0029,
+      "step": 13197
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2433848665882962e-05,
+      "loss": 0.9121,
+      "step": 13198
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2430964729974058e-05,
+      "loss": 0.8848,
+      "step": 13199
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2428080977743386e-05,
+      "loss": 0.8516,
+      "step": 13200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2425197409260914e-05,
+      "loss": 0.9004,
+      "step": 13201
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2422314024596629e-05,
+      "loss": 0.7988,
+      "step": 13202
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2419430823820493e-05,
+      "loss": 0.7749,
+      "step": 13203
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2416547807002481e-05,
+      "loss": 0.8223,
+      "step": 13204
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2413664974212552e-05,
+      "loss": 0.8359,
+      "step": 13205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2410782325520661e-05,
+      "loss": 0.9326,
+      "step": 13206
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2407899860996764e-05,
+      "loss": 0.8564,
+      "step": 13207
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2405017580710812e-05,
+      "loss": 0.9023,
+      "step": 13208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2402135484732743e-05,
+      "loss": 0.7607,
+      "step": 13209
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2399253573132506e-05,
+      "loss": 0.9287,
+      "step": 13210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2396371845980029e-05,
+      "loss": 0.9561,
+      "step": 13211
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2393490303345248e-05,
+      "loss": 0.8975,
+      "step": 13212
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2390608945298085e-05,
+      "loss": 0.8721,
+      "step": 13213
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2387727771908462e-05,
+      "loss": 0.6597,
+      "step": 13214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2384846783246304e-05,
+      "loss": 0.8281,
+      "step": 13215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2381965979381519e-05,
+      "loss": 0.8867,
+      "step": 13216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2379085360384013e-05,
+      "loss": 0.8809,
+      "step": 13217
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2376204926323694e-05,
+      "loss": 0.8447,
+      "step": 13218
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2373324677270465e-05,
+      "loss": 0.9414,
+      "step": 13219
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.237044461329421e-05,
+      "loss": 0.8779,
+      "step": 13220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2367564734464832e-05,
+      "loss": 0.959,
+      "step": 13221
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2364685040852207e-05,
+      "loss": 0.8252,
+      "step": 13222
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2361805532526225e-05,
+      "loss": 0.7607,
+      "step": 13223
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2358926209556757e-05,
+      "loss": 0.9307,
+      "step": 13224
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2356047072013681e-05,
+      "loss": 0.8379,
+      "step": 13225
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.235316811996686e-05,
+      "loss": 0.8857,
+      "step": 13226
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2350289353486162e-05,
+      "loss": 1.0508,
+      "step": 13227
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2347410772641442e-05,
+      "loss": 0.8096,
+      "step": 13228
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2344532377502563e-05,
+      "loss": 0.792,
+      "step": 13229
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2341654168139364e-05,
+      "loss": 0.8398,
+      "step": 13230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2338776144621698e-05,
+      "loss": 0.874,
+      "step": 13231
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.23358983070194e-05,
+      "loss": 0.9619,
+      "step": 13232
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2333020655402317e-05,
+      "loss": 0.7256,
+      "step": 13233
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2330143189840268e-05,
+      "loss": 0.8447,
+      "step": 13234
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2327265910403094e-05,
+      "loss": 1.0303,
+      "step": 13235
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2324388817160608e-05,
+      "loss": 0.9941,
+      "step": 13236
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2321511910182635e-05,
+      "loss": 0.9287,
+      "step": 13237
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2318635189538984e-05,
+      "loss": 0.9111,
+      "step": 13238
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2315758655299467e-05,
+      "loss": 0.9424,
+      "step": 13239
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2312882307533888e-05,
+      "loss": 0.9219,
+      "step": 13240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2310006146312052e-05,
+      "loss": 0.8149,
+      "step": 13241
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2307130171703748e-05,
+      "loss": 0.8633,
+      "step": 13242
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2304254383778776e-05,
+      "loss": 0.9609,
+      "step": 13243
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.230137878260691e-05,
+      "loss": 0.7354,
+      "step": 13244
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2298503368257946e-05,
+      "loss": 1.0566,
+      "step": 13245
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2295628140801653e-05,
+      "loss": 0.876,
+      "step": 13246
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2292753100307808e-05,
+      "loss": 0.9072,
+      "step": 13247
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2289878246846176e-05,
+      "loss": 0.8447,
+      "step": 13248
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2287003580486532e-05,
+      "loss": 0.9092,
+      "step": 13249
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2284129101298616e-05,
+      "loss": 0.8838,
+      "step": 13250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2281254809352202e-05,
+      "loss": 0.9297,
+      "step": 13251
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2278380704717034e-05,
+      "loss": 0.9287,
+      "step": 13252
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.227550678746286e-05,
+      "loss": 0.9082,
+      "step": 13253
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2272633057659413e-05,
+      "loss": 0.8896,
+      "step": 13254
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2269759515376443e-05,
+      "loss": 0.8281,
+      "step": 13255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2266886160683672e-05,
+      "loss": 0.9561,
+      "step": 13256
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2264012993650836e-05,
+      "loss": 0.9854,
+      "step": 13257
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2261140014347648e-05,
+      "loss": 0.7017,
+      "step": 13258
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2258267222843839e-05,
+      "loss": 0.8613,
+      "step": 13259
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2255394619209111e-05,
+      "loss": 0.9004,
+      "step": 13260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2252522203513184e-05,
+      "loss": 1.0137,
+      "step": 13261
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2249649975825755e-05,
+      "loss": 0.7812,
+      "step": 13262
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2246777936216531e-05,
+      "loss": 0.6685,
+      "step": 13263
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2243906084755204e-05,
+      "loss": 0.9131,
+      "step": 13264
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2241034421511469e-05,
+      "loss": 0.9834,
+      "step": 13265
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2238162946555004e-05,
+      "loss": 0.8359,
+      "step": 13266
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2235291659955504e-05,
+      "loss": 0.7593,
+      "step": 13267
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2232420561782633e-05,
+      "loss": 0.8384,
+      "step": 13268
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2229549652106081e-05,
+      "loss": 0.8721,
+      "step": 13269
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2226678930995499e-05,
+      "loss": 1.0244,
+      "step": 13270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2223808398520565e-05,
+      "loss": 0.8975,
+      "step": 13271
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2220938054750926e-05,
+      "loss": 0.8906,
+      "step": 13272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2218067899756248e-05,
+      "loss": 1.0391,
+      "step": 13273
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.221519793360617e-05,
+      "loss": 0.9131,
+      "step": 13274
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2212328156370352e-05,
+      "loss": 0.8584,
+      "step": 13275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.220945856811842e-05,
+      "loss": 0.833,
+      "step": 13276
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2206589168920022e-05,
+      "loss": 0.8291,
+      "step": 13277
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2203719958844783e-05,
+      "loss": 0.9893,
+      "step": 13278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2200850937962336e-05,
+      "loss": 0.7969,
+      "step": 13279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2197982106342295e-05,
+      "loss": 0.8545,
+      "step": 13280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2195113464054286e-05,
+      "loss": 0.959,
+      "step": 13281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2192245011167917e-05,
+      "loss": 0.9785,
+      "step": 13282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2189376747752805e-05,
+      "loss": 0.8906,
+      "step": 13283
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2186508673878545e-05,
+      "loss": 0.9678,
+      "step": 13284
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2183640789614746e-05,
+      "loss": 0.8867,
+      "step": 13285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2180773095030991e-05,
+      "loss": 0.876,
+      "step": 13286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2177905590196884e-05,
+      "loss": 1.0137,
+      "step": 13287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2175038275181999e-05,
+      "loss": 1.0498,
+      "step": 13288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2172171150055924e-05,
+      "loss": 0.8379,
+      "step": 13289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2169304214888235e-05,
+      "loss": 0.7539,
+      "step": 13290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2166437469748506e-05,
+      "loss": 0.7969,
+      "step": 13291
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2163570914706297e-05,
+      "loss": 0.8496,
+      "step": 13292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2160704549831181e-05,
+      "loss": 0.9004,
+      "step": 13293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2157838375192705e-05,
+      "loss": 0.8867,
+      "step": 13294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2154972390860436e-05,
+      "loss": 0.8545,
+      "step": 13295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.215210659690391e-05,
+      "loss": 1.0078,
+      "step": 13296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.214924099339268e-05,
+      "loss": 0.7344,
+      "step": 13297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2146375580396282e-05,
+      "loss": 1.0352,
+      "step": 13298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2143510357984247e-05,
+      "loss": 0.9287,
+      "step": 13299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2140645326226119e-05,
+      "loss": 0.8359,
+      "step": 13300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2137780485191409e-05,
+      "loss": 0.8691,
+      "step": 13301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2134915834949647e-05,
+      "loss": 0.8965,
+      "step": 13302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2132051375570346e-05,
+      "loss": 0.79,
+      "step": 13303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2129187107123024e-05,
+      "loss": 1.0391,
+      "step": 13304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2126323029677179e-05,
+      "loss": 0.9453,
+      "step": 13305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2123459143302321e-05,
+      "loss": 0.9385,
+      "step": 13306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2120595448067944e-05,
+      "loss": 0.7803,
+      "step": 13307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2117731944043545e-05,
+      "loss": 0.999,
+      "step": 13308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2114868631298608e-05,
+      "loss": 0.9189,
+      "step": 13309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2112005509902622e-05,
+      "loss": 0.8428,
+      "step": 13310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2109142579925062e-05,
+      "loss": 0.8867,
+      "step": 13311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2106279841435408e-05,
+      "loss": 0.9629,
+      "step": 13312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2103417294503125e-05,
+      "loss": 0.9941,
+      "step": 13313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2100554939197686e-05,
+      "loss": 0.8154,
+      "step": 13314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2097692775588543e-05,
+      "loss": 0.8037,
+      "step": 13315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.209483080374516e-05,
+      "loss": 0.7529,
+      "step": 13316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2091969023736981e-05,
+      "loss": 0.96,
+      "step": 13317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.208910743563346e-05,
+      "loss": 0.9746,
+      "step": 13318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2086246039504035e-05,
+      "loss": 0.7783,
+      "step": 13319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2083384835418147e-05,
+      "loss": 0.9111,
+      "step": 13320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2080523823445223e-05,
+      "loss": 0.873,
+      "step": 13321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2077663003654698e-05,
+      "loss": 0.7415,
+      "step": 13322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2074802376115992e-05,
+      "loss": 0.8164,
+      "step": 13323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.207194194089853e-05,
+      "loss": 1.0547,
+      "step": 13324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2069081698071715e-05,
+      "loss": 0.8066,
+      "step": 13325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2066221647704967e-05,
+      "loss": 0.9209,
+      "step": 13326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2063361789867685e-05,
+      "loss": 0.8828,
+      "step": 13327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2060502124629277e-05,
+      "loss": 0.8291,
+      "step": 13328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2057642652059127e-05,
+      "loss": 0.876,
+      "step": 13329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2054783372226635e-05,
+      "loss": 1.0088,
+      "step": 13330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2051924285201182e-05,
+      "loss": 0.9111,
+      "step": 13331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2049065391052154e-05,
+      "loss": 0.8936,
+      "step": 13332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2046206689848923e-05,
+      "loss": 0.9355,
+      "step": 13333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2043348181660862e-05,
+      "loss": 0.6699,
+      "step": 13334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2040489866557352e-05,
+      "loss": 0.8164,
+      "step": 13335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2037631744607736e-05,
+      "loss": 0.9775,
+      "step": 13336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2034773815881386e-05,
+      "loss": 1.0996,
+      "step": 13337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2031916080447648e-05,
+      "loss": 0.9561,
+      "step": 13338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2029058538375872e-05,
+      "loss": 0.7725,
+      "step": 13339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2026201189735409e-05,
+      "loss": 0.751,
+      "step": 13340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2023344034595588e-05,
+      "loss": 0.7861,
+      "step": 13341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2020487073025754e-05,
+      "loss": 1.0078,
+      "step": 13342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2017630305095222e-05,
+      "loss": 0.9004,
+      "step": 13343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2014773730873337e-05,
+      "loss": 0.6548,
+      "step": 13344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2011917350429405e-05,
+      "loss": 0.8799,
+      "step": 13345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.200906116383275e-05,
+      "loss": 0.9375,
+      "step": 13346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2006205171152679e-05,
+      "loss": 0.8643,
+      "step": 13347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.20033493724585e-05,
+      "loss": 1.0186,
+      "step": 13348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2000493767819509e-05,
+      "loss": 1.0146,
+      "step": 13349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1997638357305017e-05,
+      "loss": 0.8701,
+      "step": 13350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1994783140984302e-05,
+      "loss": 1.0127,
+      "step": 13351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1991928118926661e-05,
+      "loss": 1.043,
+      "step": 13352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1989073291201372e-05,
+      "loss": 0.8242,
+      "step": 13353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1986218657877715e-05,
+      "loss": 0.9717,
+      "step": 13354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1983364219024962e-05,
+      "loss": 0.9268,
+      "step": 13355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.198050997471239e-05,
+      "loss": 0.918,
+      "step": 13356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1977655925009248e-05,
+      "loss": 1.0146,
+      "step": 13357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1974802069984808e-05,
+      "loss": 0.959,
+      "step": 13358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1971948409708318e-05,
+      "loss": 0.7295,
+      "step": 13359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1969094944249035e-05,
+      "loss": 0.9258,
+      "step": 13360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1966241673676194e-05,
+      "loss": 0.8076,
+      "step": 13361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1963388598059048e-05,
+      "loss": 0.6909,
+      "step": 13362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.196053571746682e-05,
+      "loss": 0.7029,
+      "step": 13363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.195768303196875e-05,
+      "loss": 0.8848,
+      "step": 13364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1954830541634058e-05,
+      "loss": 0.8223,
+      "step": 13365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1951978246531972e-05,
+      "loss": 0.8926,
+      "step": 13366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1949126146731702e-05,
+      "loss": 0.8916,
+      "step": 13367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1946274242302469e-05,
+      "loss": 0.8467,
+      "step": 13368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1943422533313472e-05,
+      "loss": 0.9531,
+      "step": 13369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1940571019833918e-05,
+      "loss": 0.877,
+      "step": 13370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1937719701933e-05,
+      "loss": 0.7285,
+      "step": 13371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1934868579679919e-05,
+      "loss": 0.8428,
+      "step": 13372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1932017653143855e-05,
+      "loss": 1.0674,
+      "step": 13373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1929166922393998e-05,
+      "loss": 0.8984,
+      "step": 13374
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1926316387499521e-05,
+      "loss": 0.8047,
+      "step": 13375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1923466048529606e-05,
+      "loss": 0.9688,
+      "step": 13376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1920615905553413e-05,
+      "loss": 0.8877,
+      "step": 13377
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1917765958640115e-05,
+      "loss": 1.0352,
+      "step": 13378
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1914916207858867e-05,
+      "loss": 0.9658,
+      "step": 13379
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1912066653278827e-05,
+      "loss": 0.9561,
+      "step": 13380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.190921729496914e-05,
+      "loss": 0.7012,
+      "step": 13381
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.190636813299896e-05,
+      "loss": 0.667,
+      "step": 13382
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1903519167437417e-05,
+      "loss": 0.7979,
+      "step": 13383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1900670398353658e-05,
+      "loss": 0.9268,
+      "step": 13384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1897821825816804e-05,
+      "loss": 0.9121,
+      "step": 13385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1894973449895988e-05,
+      "loss": 1.0293,
+      "step": 13386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1892125270660333e-05,
+      "loss": 0.874,
+      "step": 13387
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.188927728817895e-05,
+      "loss": 0.7588,
+      "step": 13388
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1886429502520959e-05,
+      "loss": 0.8213,
+      "step": 13389
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1883581913755458e-05,
+      "loss": 0.9463,
+      "step": 13390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.188073452195156e-05,
+      "loss": 0.9307,
+      "step": 13391
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1877887327178353e-05,
+      "loss": 0.7012,
+      "step": 13392
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1875040329504937e-05,
+      "loss": 0.8838,
+      "step": 13393
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1872193529000392e-05,
+      "loss": 0.8418,
+      "step": 13394
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1869346925733813e-05,
+      "loss": 1.0117,
+      "step": 13395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.186650051977427e-05,
+      "loss": 0.9287,
+      "step": 13396
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1863654311190844e-05,
+      "loss": 0.999,
+      "step": 13397
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1860808300052592e-05,
+      "loss": 1.0537,
+      "step": 13398
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1857962486428592e-05,
+      "loss": 0.8779,
+      "step": 13399
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1855116870387895e-05,
+      "loss": 0.8311,
+      "step": 13400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.185227145199956e-05,
+      "loss": 1.0342,
+      "step": 13401
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1849426231332633e-05,
+      "loss": 0.8057,
+      "step": 13402
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1846581208456164e-05,
+      "loss": 1.0225,
+      "step": 13403
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1843736383439188e-05,
+      "loss": 0.8398,
+      "step": 13404
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1840891756350748e-05,
+      "loss": 0.8311,
+      "step": 13405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1838047327259866e-05,
+      "loss": 0.8682,
+      "step": 13406
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1835203096235572e-05,
+      "loss": 0.8809,
+      "step": 13407
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1832359063346885e-05,
+      "loss": 0.9199,
+      "step": 13408
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.182951522866283e-05,
+      "loss": 0.9473,
+      "step": 13409
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1826671592252405e-05,
+      "loss": 0.835,
+      "step": 13410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1823828154184628e-05,
+      "loss": 0.8389,
+      "step": 13411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1820984914528492e-05,
+      "loss": 0.8848,
+      "step": 13412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1818141873353004e-05,
+      "loss": 0.9414,
+      "step": 13413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1815299030727148e-05,
+      "loss": 0.9502,
+      "step": 13414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1812456386719914e-05,
+      "loss": 0.8965,
+      "step": 13415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1809613941400282e-05,
+      "loss": 0.8818,
+      "step": 13416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1806771694837233e-05,
+      "loss": 1.2002,
+      "step": 13417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1803929647099743e-05,
+      "loss": 0.9785,
+      "step": 13418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1801087798256774e-05,
+      "loss": 0.9355,
+      "step": 13419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1798246148377297e-05,
+      "loss": 0.9443,
+      "step": 13420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.179540469753026e-05,
+      "loss": 0.8711,
+      "step": 13421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1792563445784626e-05,
+      "loss": 0.793,
+      "step": 13422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1789722393209337e-05,
+      "loss": 0.9404,
+      "step": 13423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1786881539873339e-05,
+      "loss": 0.8066,
+      "step": 13424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1784040885845576e-05,
+      "loss": 0.8848,
+      "step": 13425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1781200431194976e-05,
+      "loss": 0.8545,
+      "step": 13426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1778360175990473e-05,
+      "loss": 1.0312,
+      "step": 13427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1775520120300987e-05,
+      "loss": 0.7993,
+      "step": 13428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1772680264195446e-05,
+      "loss": 0.9961,
+      "step": 13429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1769840607742751e-05,
+      "loss": 0.7568,
+      "step": 13430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1767001151011828e-05,
+      "loss": 0.7798,
+      "step": 13431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1764161894071568e-05,
+      "loss": 0.8125,
+      "step": 13432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1761322836990883e-05,
+      "loss": 0.9492,
+      "step": 13433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1758483979838661e-05,
+      "loss": 0.8828,
+      "step": 13434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1755645322683797e-05,
+      "loss": 0.8838,
+      "step": 13435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1752806865595175e-05,
+      "loss": 0.7812,
+      "step": 13436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1749968608641677e-05,
+      "loss": 0.8579,
+      "step": 13437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1747130551892176e-05,
+      "loss": 0.916,
+      "step": 13438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1744292695415549e-05,
+      "loss": 0.8525,
+      "step": 13439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1741455039280654e-05,
+      "loss": 0.8315,
+      "step": 13440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1738617583556364e-05,
+      "loss": 0.7725,
+      "step": 13441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1735780328311526e-05,
+      "loss": 0.9082,
+      "step": 13442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1732943273614996e-05,
+      "loss": 0.9629,
+      "step": 13443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1730106419535616e-05,
+      "loss": 0.9219,
+      "step": 13444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.172726976614224e-05,
+      "loss": 0.8672,
+      "step": 13445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1724433313503691e-05,
+      "loss": 0.7852,
+      "step": 13446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1721597061688815e-05,
+      "loss": 0.8389,
+      "step": 13447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1718761010766424e-05,
+      "loss": 0.8359,
+      "step": 13448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1715925160805355e-05,
+      "loss": 0.8975,
+      "step": 13449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1713089511874415e-05,
+      "loss": 0.7988,
+      "step": 13450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1710254064042426e-05,
+      "loss": 0.9355,
+      "step": 13451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1707418817378188e-05,
+      "loss": 0.8848,
+      "step": 13452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1704583771950514e-05,
+      "loss": 0.7441,
+      "step": 13453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1701748927828188e-05,
+      "loss": 0.8418,
+      "step": 13454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1698914285080017e-05,
+      "loss": 0.8555,
+      "step": 13455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.169607984377478e-05,
+      "loss": 1.043,
+      "step": 13456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1693245603981266e-05,
+      "loss": 0.8506,
+      "step": 13457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.169041156576825e-05,
+      "loss": 0.8135,
+      "step": 13458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1687577729204512e-05,
+      "loss": 0.8691,
+      "step": 13459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.168474409435881e-05,
+      "loss": 0.9199,
+      "step": 13460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.168191066129992e-05,
+      "loss": 0.9941,
+      "step": 13461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1679077430096593e-05,
+      "loss": 0.8887,
+      "step": 13462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1676244400817589e-05,
+      "loss": 0.8154,
+      "step": 13463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1673411573531654e-05,
+      "loss": 0.8809,
+      "step": 13464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1670578948307525e-05,
+      "loss": 0.9111,
+      "step": 13465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1667746525213954e-05,
+      "loss": 0.8516,
+      "step": 13466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1664914304319674e-05,
+      "loss": 0.8662,
+      "step": 13467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.166208228569341e-05,
+      "loss": 0.9385,
+      "step": 13468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1659250469403882e-05,
+      "loss": 0.8037,
+      "step": 13469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.165641885551982e-05,
+      "loss": 0.9971,
+      "step": 13470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1653587444109938e-05,
+      "loss": 0.8301,
+      "step": 13471
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1650756235242935e-05,
+      "loss": 0.7725,
+      "step": 13472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1647925228987528e-05,
+      "loss": 0.8398,
+      "step": 13473
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1645094425412418e-05,
+      "loss": 0.8213,
+      "step": 13474
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.164226382458629e-05,
+      "loss": 0.7402,
+      "step": 13475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1639433426577838e-05,
+      "loss": 0.8203,
+      "step": 13476
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1636603231455752e-05,
+      "loss": 0.9619,
+      "step": 13477
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1633773239288713e-05,
+      "loss": 0.8359,
+      "step": 13478
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.163094345014539e-05,
+      "loss": 0.7368,
+      "step": 13479
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1628113864094453e-05,
+      "loss": 0.9824,
+      "step": 13480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1625284481204577e-05,
+      "loss": 0.8369,
+      "step": 13481
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1622455301544418e-05,
+      "loss": 0.9453,
+      "step": 13482
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.161962632518263e-05,
+      "loss": 0.8203,
+      "step": 13483
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.161679755218786e-05,
+      "loss": 0.7412,
+      "step": 13484
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1613968982628765e-05,
+      "loss": 0.8608,
+      "step": 13485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1611140616573984e-05,
+      "loss": 0.8564,
+      "step": 13486
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1608312454092145e-05,
+      "loss": 1.0225,
+      "step": 13487
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1605484495251877e-05,
+      "loss": 0.8691,
+      "step": 13488
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1602656740121822e-05,
+      "loss": 0.9531,
+      "step": 13489
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1599829188770592e-05,
+      "loss": 0.8574,
+      "step": 13490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1597001841266805e-05,
+      "loss": 0.8682,
+      "step": 13491
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1594174697679059e-05,
+      "loss": 0.8486,
+      "step": 13492
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1591347758075984e-05,
+      "loss": 0.9004,
+      "step": 13493
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1588521022526169e-05,
+      "loss": 0.79,
+      "step": 13494
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1585694491098209e-05,
+      "loss": 0.8691,
+      "step": 13495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1582868163860695e-05,
+      "loss": 0.7061,
+      "step": 13496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.158004204088222e-05,
+      "loss": 0.957,
+      "step": 13497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1577216122231368e-05,
+      "loss": 0.8809,
+      "step": 13498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1574390407976704e-05,
+      "loss": 0.9229,
+      "step": 13499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.15715648981868e-05,
+      "loss": 0.7598,
+      "step": 13500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1568739592930229e-05,
+      "loss": 0.9033,
+      "step": 13501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1565914492275563e-05,
+      "loss": 0.9365,
+      "step": 13502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1563089596291347e-05,
+      "loss": 0.9111,
+      "step": 13503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1560264905046132e-05,
+      "loss": 0.8291,
+      "step": 13504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1557440418608463e-05,
+      "loss": 0.9834,
+      "step": 13505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1554616137046892e-05,
+      "loss": 0.8848,
+      "step": 13506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1551792060429951e-05,
+      "loss": 0.9512,
+      "step": 13507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1548968188826173e-05,
+      "loss": 0.8691,
+      "step": 13508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1546144522304074e-05,
+      "loss": 0.877,
+      "step": 13509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1543321060932193e-05,
+      "loss": 0.9209,
+      "step": 13510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.154049780477904e-05,
+      "loss": 1.002,
+      "step": 13511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.153767475391312e-05,
+      "loss": 0.7715,
+      "step": 13512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1534851908402956e-05,
+      "loss": 0.8086,
+      "step": 13513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1532029268317037e-05,
+      "loss": 0.8516,
+      "step": 13514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1529206833723867e-05,
+      "loss": 0.9189,
+      "step": 13515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1526384604691927e-05,
+      "loss": 0.7979,
+      "step": 13516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.152356258128972e-05,
+      "loss": 0.9707,
+      "step": 13517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.152074076358572e-05,
+      "loss": 0.8311,
+      "step": 13518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1517919151648407e-05,
+      "loss": 0.6709,
+      "step": 13519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1515097745546245e-05,
+      "loss": 0.8711,
+      "step": 13520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1512276545347709e-05,
+      "loss": 0.8242,
+      "step": 13521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1509455551121266e-05,
+      "loss": 0.8594,
+      "step": 13522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1506634762935365e-05,
+      "loss": 0.752,
+      "step": 13523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1503814180858451e-05,
+      "loss": 0.7249,
+      "step": 13524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.150099380495899e-05,
+      "loss": 0.7695,
+      "step": 13525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1498173635305414e-05,
+      "loss": 0.9199,
+      "step": 13526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.149535367196616e-05,
+      "loss": 0.9238,
+      "step": 13527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1492533915009656e-05,
+      "loss": 0.79,
+      "step": 13528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1489714364504342e-05,
+      "loss": 0.8359,
+      "step": 13529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.148689502051863e-05,
+      "loss": 0.9346,
+      "step": 13530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1484075883120939e-05,
+      "loss": 0.8945,
+      "step": 13531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1481256952379676e-05,
+      "loss": 0.9541,
+      "step": 13532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1478438228363263e-05,
+      "loss": 0.7891,
+      "step": 13533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1475619711140092e-05,
+      "loss": 0.8418,
+      "step": 13534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.147280140077856e-05,
+      "loss": 0.9453,
+      "step": 13535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1469983297347054e-05,
+      "loss": 0.8711,
+      "step": 13536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1467165400913976e-05,
+      "loss": 0.71,
+      "step": 13537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1464347711547696e-05,
+      "loss": 0.9453,
+      "step": 13538
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1461530229316597e-05,
+      "loss": 0.7759,
+      "step": 13539
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1458712954289042e-05,
+      "loss": 1.0488,
+      "step": 13540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1455895886533412e-05,
+      "loss": 1.0117,
+      "step": 13541
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.145307902611806e-05,
+      "loss": 0.7871,
+      "step": 13542
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1450262373111344e-05,
+      "loss": 0.8457,
+      "step": 13543
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1447445927581611e-05,
+      "loss": 0.8682,
+      "step": 13544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1444629689597222e-05,
+      "loss": 0.8281,
+      "step": 13545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1441813659226507e-05,
+      "loss": 0.79,
+      "step": 13546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1438997836537806e-05,
+      "loss": 0.9248,
+      "step": 13547
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1436182221599443e-05,
+      "loss": 0.9502,
+      "step": 13548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.143336681447976e-05,
+      "loss": 0.958,
+      "step": 13549
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1430551615247072e-05,
+      "loss": 0.8457,
+      "step": 13550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1427736623969695e-05,
+      "loss": 0.9805,
+      "step": 13551
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.142492184071593e-05,
+      "loss": 0.6855,
+      "step": 13552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1422107265554102e-05,
+      "loss": 0.8105,
+      "step": 13553
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1419292898552504e-05,
+      "loss": 1.0469,
+      "step": 13554
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1416478739779433e-05,
+      "loss": 0.8232,
+      "step": 13555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.141366478930317e-05,
+      "loss": 0.9463,
+      "step": 13556
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1410851047192023e-05,
+      "loss": 0.958,
+      "step": 13557
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1408037513514256e-05,
+      "loss": 0.7969,
+      "step": 13558
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1405224188338146e-05,
+      "loss": 0.7852,
+      "step": 13559
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1402411071731976e-05,
+      "loss": 0.8584,
+      "step": 13560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1399598163764005e-05,
+      "loss": 0.8711,
+      "step": 13561
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1396785464502492e-05,
+      "loss": 0.916,
+      "step": 13562
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1393972974015688e-05,
+      "loss": 0.8809,
+      "step": 13563
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1391160692371859e-05,
+      "loss": 0.9482,
+      "step": 13564
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.138834861963924e-05,
+      "loss": 0.6748,
+      "step": 13565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1385536755886078e-05,
+      "loss": 0.9014,
+      "step": 13566
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1382725101180593e-05,
+      "loss": 1.0156,
+      "step": 13567
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1379913655591038e-05,
+      "loss": 0.8877,
+      "step": 13568
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1377102419185627e-05,
+      "loss": 0.8633,
+      "step": 13569
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.137429139203258e-05,
+      "loss": 1.042,
+      "step": 13570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1371480574200104e-05,
+      "loss": 0.8818,
+      "step": 13571
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1368669965756431e-05,
+      "loss": 0.8428,
+      "step": 13572
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1365859566769752e-05,
+      "loss": 0.8682,
+      "step": 13573
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.136304937730827e-05,
+      "loss": 0.7256,
+      "step": 13574
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1360239397440171e-05,
+      "loss": 0.876,
+      "step": 13575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.135742962723366e-05,
+      "loss": 0.8301,
+      "step": 13576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1354620066756916e-05,
+      "loss": 0.8174,
+      "step": 13577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1351810716078119e-05,
+      "loss": 0.8975,
+      "step": 13578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1349001575265435e-05,
+      "loss": 0.7109,
+      "step": 13579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1346192644387047e-05,
+      "loss": 0.6538,
+      "step": 13580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1343383923511114e-05,
+      "loss": 0.8799,
+      "step": 13581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1340575412705795e-05,
+      "loss": 0.8818,
+      "step": 13582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1337767112039235e-05,
+      "loss": 0.832,
+      "step": 13583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1334959021579594e-05,
+      "loss": 0.8643,
+      "step": 13584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1332151141395023e-05,
+      "loss": 0.792,
+      "step": 13585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1329343471553653e-05,
+      "loss": 0.8096,
+      "step": 13586
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1326536012123617e-05,
+      "loss": 0.8799,
+      "step": 13587
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1323728763173034e-05,
+      "loss": 0.8926,
+      "step": 13588
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.132092172477005e-05,
+      "loss": 0.7856,
+      "step": 13589
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1318114896982767e-05,
+      "loss": 0.9141,
+      "step": 13590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1315308279879305e-05,
+      "loss": 1.0801,
+      "step": 13591
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1312501873527763e-05,
+      "loss": 0.8799,
+      "step": 13592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.130969567799626e-05,
+      "loss": 0.8389,
+      "step": 13593
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1306889693352883e-05,
+      "loss": 0.8252,
+      "step": 13594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1304083919665732e-05,
+      "loss": 0.9238,
+      "step": 13595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1301278357002876e-05,
+      "loss": 0.8018,
+      "step": 13596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1298473005432427e-05,
+      "loss": 0.999,
+      "step": 13597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1295667865022444e-05,
+      "loss": 0.9111,
+      "step": 13598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1292862935840995e-05,
+      "loss": 0.8369,
+      "step": 13599
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1290058217956166e-05,
+      "loss": 0.8672,
+      "step": 13600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1287253711436007e-05,
+      "loss": 0.9482,
+      "step": 13601
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1284449416348578e-05,
+      "loss": 0.8594,
+      "step": 13602
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1281645332761923e-05,
+      "loss": 0.8672,
+      "step": 13603
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1278841460744104e-05,
+      "loss": 0.9238,
+      "step": 13604
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1276037800363153e-05,
+      "loss": 0.834,
+      "step": 13605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.127323435168711e-05,
+      "loss": 0.9492,
+      "step": 13606
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1270431114783996e-05,
+      "loss": 0.7866,
+      "step": 13607
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1267628089721857e-05,
+      "loss": 0.8096,
+      "step": 13608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1264825276568698e-05,
+      "loss": 0.9033,
+      "step": 13609
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1262022675392544e-05,
+      "loss": 0.7617,
+      "step": 13610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1259220286261392e-05,
+      "loss": 0.8623,
+      "step": 13611
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1256418109243268e-05,
+      "loss": 0.9414,
+      "step": 13612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.125361614440616e-05,
+      "loss": 0.8066,
+      "step": 13613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1250814391818067e-05,
+      "loss": 0.8496,
+      "step": 13614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1248012851546971e-05,
+      "loss": 0.8184,
+      "step": 13615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1245211523660874e-05,
+      "loss": 0.8945,
+      "step": 13616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1242410408227744e-05,
+      "loss": 1.0381,
+      "step": 13617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1239609505315561e-05,
+      "loss": 0.8643,
+      "step": 13618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1236808814992282e-05,
+      "loss": 0.8623,
+      "step": 13619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1234008337325889e-05,
+      "loss": 0.9229,
+      "step": 13620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1231208072384333e-05,
+      "loss": 0.9463,
+      "step": 13621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.122840802023557e-05,
+      "loss": 0.7852,
+      "step": 13622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.122560818094754e-05,
+      "loss": 0.9336,
+      "step": 13623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1222808554588202e-05,
+      "loss": 0.8682,
+      "step": 13624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.122000914122549e-05,
+      "loss": 0.9844,
+      "step": 13625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1217209940927333e-05,
+      "loss": 0.8037,
+      "step": 13626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1214410953761655e-05,
+      "loss": 0.7104,
+      "step": 13627
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1211612179796391e-05,
+      "loss": 0.751,
+      "step": 13628
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1208813619099457e-05,
+      "loss": 0.6777,
+      "step": 13629
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.120601527173876e-05,
+      "loss": 0.7627,
+      "step": 13630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1203217137782202e-05,
+      "loss": 0.8076,
+      "step": 13631
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1200419217297704e-05,
+      "loss": 0.7559,
+      "step": 13632
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1197621510353151e-05,
+      "loss": 0.918,
+      "step": 13633
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1194824017016442e-05,
+      "loss": 0.8818,
+      "step": 13634
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1192026737355443e-05,
+      "loss": 0.9639,
+      "step": 13635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1189229671438064e-05,
+      "loss": 0.9873,
+      "step": 13636
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1186432819332169e-05,
+      "loss": 0.7998,
+      "step": 13637
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1183636181105633e-05,
+      "loss": 0.7158,
+      "step": 13638
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1180839756826306e-05,
+      "loss": 0.9688,
+      "step": 13639
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1178043546562071e-05,
+      "loss": 0.7705,
+      "step": 13640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1175247550380775e-05,
+      "loss": 0.9307,
+      "step": 13641
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.117245176835026e-05,
+      "loss": 0.7979,
+      "step": 13642
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.116965620053839e-05,
+      "loss": 0.8994,
+      "step": 13643
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1166860847012993e-05,
+      "loss": 0.8311,
+      "step": 13644
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1164065707841906e-05,
+      "loss": 0.833,
+      "step": 13645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.116127078309295e-05,
+      "loss": 0.8213,
+      "step": 13646
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1158476072833965e-05,
+      "loss": 0.8984,
+      "step": 13647
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1155681577132768e-05,
+      "loss": 0.7881,
+      "step": 13648
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1152887296057167e-05,
+      "loss": 0.793,
+      "step": 13649
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1150093229674965e-05,
+      "loss": 0.8789,
+      "step": 13650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1147299378053979e-05,
+      "loss": 0.8506,
+      "step": 13651
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1144505741262007e-05,
+      "loss": 0.9307,
+      "step": 13652
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1141712319366835e-05,
+      "loss": 0.9316,
+      "step": 13653
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1138919112436244e-05,
+      "loss": 0.9551,
+      "step": 13654
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1136126120538038e-05,
+      "loss": 0.9521,
+      "step": 13655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1133333343739983e-05,
+      "loss": 0.9307,
+      "step": 13656
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.113054078210985e-05,
+      "loss": 0.8887,
+      "step": 13657
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.11277484357154e-05,
+      "loss": 1.0459,
+      "step": 13658
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.112495630462441e-05,
+      "loss": 0.7227,
+      "step": 13659
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1122164388904632e-05,
+      "loss": 0.8281,
+      "step": 13660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1119372688623816e-05,
+      "loss": 1.0098,
+      "step": 13661
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.11165812038497e-05,
+      "loss": 0.9082,
+      "step": 13662
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1113789934650039e-05,
+      "loss": 0.8447,
+      "step": 13663
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1110998881092564e-05,
+      "loss": 0.9932,
+      "step": 13664
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1108208043245006e-05,
+      "loss": 0.8154,
+      "step": 13665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.110541742117508e-05,
+      "loss": 0.8555,
+      "step": 13666
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1102627014950513e-05,
+      "loss": 0.8091,
+      "step": 13667
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1099836824639033e-05,
+      "loss": 0.9414,
+      "step": 13668
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1097046850308337e-05,
+      "loss": 0.791,
+      "step": 13669
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.109425709202613e-05,
+      "loss": 1.0039,
+      "step": 13670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1091467549860106e-05,
+      "loss": 0.79,
+      "step": 13671
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1088678223877975e-05,
+      "loss": 0.8926,
+      "step": 13672
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1085889114147417e-05,
+      "loss": 0.7666,
+      "step": 13673
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.108310022073611e-05,
+      "loss": 0.7832,
+      "step": 13674
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1080311543711733e-05,
+      "loss": 0.9258,
+      "step": 13675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.107752308314197e-05,
+      "loss": 0.8789,
+      "step": 13676
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1074734839094484e-05,
+      "loss": 0.9463,
+      "step": 13677
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1071946811636932e-05,
+      "loss": 0.9072,
+      "step": 13678
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1069159000836967e-05,
+      "loss": 0.9727,
+      "step": 13679
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1066371406762255e-05,
+      "loss": 0.7529,
+      "step": 13680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.106358402948044e-05,
+      "loss": 0.8066,
+      "step": 13681
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1060796869059148e-05,
+      "loss": 0.9434,
+      "step": 13682
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1058009925566035e-05,
+      "loss": 0.9629,
+      "step": 13683
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1055223199068726e-05,
+      "loss": 0.8516,
+      "step": 13684
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1052436689634842e-05,
+      "loss": 1.0117,
+      "step": 13685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1049650397331999e-05,
+      "loss": 0.8486,
+      "step": 13686
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1046864322227826e-05,
+      "loss": 0.8691,
+      "step": 13687
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1044078464389926e-05,
+      "loss": 0.9258,
+      "step": 13688
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1041292823885904e-05,
+      "loss": 0.9551,
+      "step": 13689
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1038507400783351e-05,
+      "loss": 0.7793,
+      "step": 13690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1035722195149874e-05,
+      "loss": 0.9092,
+      "step": 13691
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1032937207053059e-05,
+      "loss": 0.9268,
+      "step": 13692
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1030152436560485e-05,
+      "loss": 0.9854,
+      "step": 13693
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1027367883739725e-05,
+      "loss": 0.8877,
+      "step": 13694
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1024583548658368e-05,
+      "loss": 0.8936,
+      "step": 13695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1021799431383969e-05,
+      "loss": 0.9297,
+      "step": 13696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1019015531984096e-05,
+      "loss": 0.6753,
+      "step": 13697
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1016231850526297e-05,
+      "loss": 1.0059,
+      "step": 13698
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1013448387078136e-05,
+      "loss": 0.96,
+      "step": 13699
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1010665141707157e-05,
+      "loss": 0.9248,
+      "step": 13700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1007882114480896e-05,
+      "loss": 0.915,
+      "step": 13701
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1005099305466884e-05,
+      "loss": 0.8311,
+      "step": 13702
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1002316714732667e-05,
+      "loss": 0.9307,
+      "step": 13703
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0999534342345764e-05,
+      "loss": 0.8008,
+      "step": 13704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0996752188373692e-05,
+      "loss": 0.8135,
+      "step": 13705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0993970252883956e-05,
+      "loss": 0.855,
+      "step": 13706
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0991188535944086e-05,
+      "loss": 1.0,
+      "step": 13707
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0988407037621575e-05,
+      "loss": 1.0391,
+      "step": 13708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0985625757983927e-05,
+      "loss": 0.7559,
+      "step": 13709
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.098284469709862e-05,
+      "loss": 0.8447,
+      "step": 13710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0980063855033162e-05,
+      "loss": 1.0137,
+      "step": 13711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.097728323185503e-05,
+      "loss": 0.8604,
+      "step": 13712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0974502827631697e-05,
+      "loss": 0.8262,
+      "step": 13713
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0971722642430632e-05,
+      "loss": 0.8477,
+      "step": 13714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0968942676319312e-05,
+      "loss": 0.9082,
+      "step": 13715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0966162929365195e-05,
+      "loss": 0.9912,
+      "step": 13716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0963383401635735e-05,
+      "loss": 0.9297,
+      "step": 13717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0960604093198377e-05,
+      "loss": 0.7646,
+      "step": 13718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0957825004120582e-05,
+      "loss": 0.7832,
+      "step": 13719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0955046134469782e-05,
+      "loss": 0.9785,
+      "step": 13720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0952267484313412e-05,
+      "loss": 0.8149,
+      "step": 13721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0949489053718893e-05,
+      "loss": 0.8037,
+      "step": 13722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0946710842753668e-05,
+      "loss": 0.874,
+      "step": 13723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0943932851485143e-05,
+      "loss": 0.8721,
+      "step": 13724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0941155079980739e-05,
+      "loss": 1.0029,
+      "step": 13725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0938377528307848e-05,
+      "loss": 0.9141,
+      "step": 13726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0935600196533896e-05,
+      "loss": 0.8179,
+      "step": 13727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0932823084726269e-05,
+      "loss": 0.8232,
+      "step": 13728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.093004619295235e-05,
+      "loss": 0.8623,
+      "step": 13729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0927269521279546e-05,
+      "loss": 0.9092,
+      "step": 13730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0924493069775228e-05,
+      "loss": 0.7695,
+      "step": 13731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0921716838506776e-05,
+      "loss": 0.7422,
+      "step": 13732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0918940827541546e-05,
+      "loss": 0.7803,
+      "step": 13733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0916165036946927e-05,
+      "loss": 0.7969,
+      "step": 13734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0913389466790267e-05,
+      "loss": 0.7388,
+      "step": 13735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.091061411713892e-05,
+      "loss": 0.8291,
+      "step": 13736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0907838988060231e-05,
+      "loss": 0.9473,
+      "step": 13737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.090506407962156e-05,
+      "loss": 0.8633,
+      "step": 13738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0902289391890232e-05,
+      "loss": 0.6855,
+      "step": 13739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.089951492493359e-05,
+      "loss": 0.8428,
+      "step": 13740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0896740678818946e-05,
+      "loss": 1.0039,
+      "step": 13741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0893966653613646e-05,
+      "loss": 0.8926,
+      "step": 13742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0891192849384993e-05,
+      "loss": 0.9375,
+      "step": 13743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0888419266200301e-05,
+      "loss": 0.8086,
+      "step": 13744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0885645904126867e-05,
+      "loss": 0.8467,
+      "step": 13745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0882872763232016e-05,
+      "loss": 0.8428,
+      "step": 13746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0880099843583024e-05,
+      "loss": 0.9248,
+      "step": 13747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0877327145247194e-05,
+      "loss": 0.8809,
+      "step": 13748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0874554668291794e-05,
+      "loss": 0.9307,
+      "step": 13749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0871782412784118e-05,
+      "loss": 0.9375,
+      "step": 13750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0869010378791443e-05,
+      "loss": 0.9336,
+      "step": 13751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0866238566381033e-05,
+      "loss": 1.0566,
+      "step": 13752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0863466975620152e-05,
+      "loss": 0.9727,
+      "step": 13753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.086069560657605e-05,
+      "loss": 1.1387,
+      "step": 13754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0857924459315998e-05,
+      "loss": 0.7061,
+      "step": 13755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.085515353390723e-05,
+      "loss": 0.9824,
+      "step": 13756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0852382830416994e-05,
+      "loss": 0.9404,
+      "step": 13757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0849612348912517e-05,
+      "loss": 0.8691,
+      "step": 13758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0846842089461042e-05,
+      "loss": 0.9092,
+      "step": 13759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0844072052129798e-05,
+      "loss": 0.9062,
+      "step": 13760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0841302236985993e-05,
+      "loss": 0.8896,
+      "step": 13761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.083853264409684e-05,
+      "loss": 0.9766,
+      "step": 13762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0835763273529567e-05,
+      "loss": 0.9199,
+      "step": 13763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0832994125351369e-05,
+      "loss": 0.8594,
+      "step": 13764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.083022519962944e-05,
+      "loss": 0.9893,
+      "step": 13765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0827456496430976e-05,
+      "loss": 0.9375,
+      "step": 13766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.082468801582317e-05,
+      "loss": 0.96,
+      "step": 13767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0821919757873206e-05,
+      "loss": 0.791,
+      "step": 13768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0819151722648248e-05,
+      "loss": 0.9365,
+      "step": 13769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0816383910215487e-05,
+      "loss": 0.8311,
+      "step": 13770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0813616320642079e-05,
+      "loss": 0.8398,
+      "step": 13771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0810848953995188e-05,
+      "loss": 0.8862,
+      "step": 13772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0808081810341959e-05,
+      "loss": 0.8564,
+      "step": 13773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0805314889749564e-05,
+      "loss": 0.8457,
+      "step": 13774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0802548192285132e-05,
+      "loss": 0.7925,
+      "step": 13775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0799781718015807e-05,
+      "loss": 1.1035,
+      "step": 13776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0797015467008713e-05,
+      "loss": 0.8809,
+      "step": 13777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0794249439331002e-05,
+      "loss": 1.0479,
+      "step": 13778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.079148363504978e-05,
+      "loss": 0.7969,
+      "step": 13779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.078871805423217e-05,
+      "loss": 0.8438,
+      "step": 13780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0785952696945277e-05,
+      "loss": 0.8984,
+      "step": 13781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0783187563256218e-05,
+      "loss": 0.9453,
+      "step": 13782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0780422653232096e-05,
+      "loss": 1.0215,
+      "step": 13783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0777657966940001e-05,
+      "loss": 0.9141,
+      "step": 13784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0774893504447014e-05,
+      "loss": 0.9746,
+      "step": 13785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0772129265820244e-05,
+      "loss": 0.9395,
+      "step": 13786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0769365251126756e-05,
+      "loss": 0.8594,
+      "step": 13787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0766601460433628e-05,
+      "loss": 1.0117,
+      "step": 13788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.076383789380792e-05,
+      "loss": 0.7656,
+      "step": 13789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.076107455131671e-05,
+      "loss": 0.8066,
+      "step": 13790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0758311433027053e-05,
+      "loss": 0.8027,
+      "step": 13791
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0755548539005998e-05,
+      "loss": 0.8906,
+      "step": 13792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0752785869320583e-05,
+      "loss": 0.8076,
+      "step": 13793
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.075002342403787e-05,
+      "loss": 0.8608,
+      "step": 13794
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0747261203224886e-05,
+      "loss": 0.9902,
+      "step": 13795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0744499206948658e-05,
+      "loss": 1.0518,
+      "step": 13796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.074173743527621e-05,
+      "loss": 0.9238,
+      "step": 13797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0738975888274571e-05,
+      "loss": 0.7959,
+      "step": 13798
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0736214566010753e-05,
+      "loss": 0.877,
+      "step": 13799
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0733453468551767e-05,
+      "loss": 0.9229,
+      "step": 13800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0730692595964603e-05,
+      "loss": 1.0186,
+      "step": 13801
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0727931948316278e-05,
+      "loss": 0.8506,
+      "step": 13802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0725171525673774e-05,
+      "loss": 0.835,
+      "step": 13803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0722411328104083e-05,
+      "loss": 1.0039,
+      "step": 13804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0719651355674177e-05,
+      "loss": 0.7817,
+      "step": 13805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0716891608451048e-05,
+      "loss": 0.8525,
+      "step": 13806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0714132086501658e-05,
+      "loss": 0.8457,
+      "step": 13807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0711372789892973e-05,
+      "loss": 0.8896,
+      "step": 13808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.070861371869195e-05,
+      "loss": 0.9092,
+      "step": 13809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0705854872965552e-05,
+      "loss": 0.9219,
+      "step": 13810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.070309625278072e-05,
+      "loss": 0.8877,
+      "step": 13811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0700337858204403e-05,
+      "loss": 0.9033,
+      "step": 13812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.069757968930353e-05,
+      "loss": 1.0137,
+      "step": 13813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0694821746145047e-05,
+      "loss": 1.1016,
+      "step": 13814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0692064028795876e-05,
+      "loss": 0.9209,
+      "step": 13815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.068930653732293e-05,
+      "loss": 0.8369,
+      "step": 13816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0686549271793138e-05,
+      "loss": 0.7627,
+      "step": 13817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.068379223227341e-05,
+      "loss": 0.8877,
+      "step": 13818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.068103541883064e-05,
+      "loss": 1.0029,
+      "step": 13819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0678278831531732e-05,
+      "loss": 0.8291,
+      "step": 13820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0675522470443586e-05,
+      "loss": 0.9014,
+      "step": 13821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0672766335633091e-05,
+      "loss": 0.8418,
+      "step": 13822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0670010427167123e-05,
+      "loss": 0.8867,
+      "step": 13823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0667254745112559e-05,
+      "loss": 0.9902,
+      "step": 13824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0664499289536283e-05,
+      "loss": 0.7725,
+      "step": 13825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0661744060505153e-05,
+      "loss": 0.9287,
+      "step": 13826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.065898905808603e-05,
+      "loss": 0.8604,
+      "step": 13827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0656234282345768e-05,
+      "loss": 0.8711,
+      "step": 13828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0653479733351229e-05,
+      "loss": 0.8799,
+      "step": 13829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0650725411169247e-05,
+      "loss": 0.8789,
+      "step": 13830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0647971315866666e-05,
+      "loss": 0.7881,
+      "step": 13831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0645217447510307e-05,
+      "loss": 0.9697,
+      "step": 13832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0642463806167012e-05,
+      "loss": 0.9004,
+      "step": 13833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0639710391903608e-05,
+      "loss": 0.9922,
+      "step": 13834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0636957204786906e-05,
+      "loss": 1.0088,
+      "step": 13835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0634204244883715e-05,
+      "loss": 0.7598,
+      "step": 13836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0631451512260838e-05,
+      "loss": 0.8008,
+      "step": 13837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0628699006985084e-05,
+      "loss": 1.0088,
+      "step": 13838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0625946729123249e-05,
+      "loss": 0.9326,
+      "step": 13839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.062319467874212e-05,
+      "loss": 0.9375,
+      "step": 13840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0620442855908468e-05,
+      "loss": 0.8418,
+      "step": 13841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0617691260689094e-05,
+      "loss": 0.8096,
+      "step": 13842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.061493989315076e-05,
+      "loss": 1.0117,
+      "step": 13843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0612188753360235e-05,
+      "loss": 0.8408,
+      "step": 13844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0609437841384271e-05,
+      "loss": 0.7026,
+      "step": 13845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0606687157289643e-05,
+      "loss": 0.8584,
+      "step": 13846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0603936701143095e-05,
+      "loss": 0.9453,
+      "step": 13847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0601186473011364e-05,
+      "loss": 0.7861,
+      "step": 13848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0598436472961194e-05,
+      "loss": 0.957,
+      "step": 13849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0595686701059329e-05,
+      "loss": 0.8574,
+      "step": 13850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0592937157372489e-05,
+      "loss": 0.8594,
+      "step": 13851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.05901878419674e-05,
+      "loss": 0.6973,
+      "step": 13852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0587438754910772e-05,
+      "loss": 0.7349,
+      "step": 13853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.058468989626933e-05,
+      "loss": 0.9004,
+      "step": 13854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0581941266109776e-05,
+      "loss": 0.8398,
+      "step": 13855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0579192864498802e-05,
+      "loss": 0.835,
+      "step": 13856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0576444691503119e-05,
+      "loss": 0.8643,
+      "step": 13857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.057369674718941e-05,
+      "loss": 0.8691,
+      "step": 13858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0570949031624358e-05,
+      "loss": 0.9912,
+      "step": 13859
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0568201544874639e-05,
+      "loss": 0.7666,
+      "step": 13860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0565454287006936e-05,
+      "loss": 0.9131,
+      "step": 13861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.056270725808791e-05,
+      "loss": 0.7344,
+      "step": 13862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0559960458184224e-05,
+      "loss": 0.834,
+      "step": 13863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0557213887362532e-05,
+      "loss": 1.0049,
+      "step": 13864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0554467545689496e-05,
+      "loss": 0.6992,
+      "step": 13865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.055172143323175e-05,
+      "loss": 0.8408,
+      "step": 13866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.054897555005594e-05,
+      "loss": 0.7959,
+      "step": 13867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0546229896228692e-05,
+      "loss": 0.958,
+      "step": 13868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0543484471816652e-05,
+      "loss": 0.7705,
+      "step": 13869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0540739276886431e-05,
+      "loss": 0.7549,
+      "step": 13870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0537994311504649e-05,
+      "loss": 0.8389,
+      "step": 13871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.053524957573791e-05,
+      "loss": 0.8213,
+      "step": 13872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0532505069652839e-05,
+      "loss": 0.7207,
+      "step": 13873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0529760793316025e-05,
+      "loss": 0.8379,
+      "step": 13874
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0527016746794065e-05,
+      "loss": 1.0449,
+      "step": 13875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0524272930153543e-05,
+      "loss": 0.8838,
+      "step": 13876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0521529343461059e-05,
+      "loss": 0.9004,
+      "step": 13877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0518785986783179e-05,
+      "loss": 0.9463,
+      "step": 13878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0516042860186482e-05,
+      "loss": 0.9727,
+      "step": 13879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0513299963737527e-05,
+      "loss": 0.9336,
+      "step": 13880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0510557297502889e-05,
+      "loss": 0.8857,
+      "step": 13881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0507814861549118e-05,
+      "loss": 0.832,
+      "step": 13882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0505072655942765e-05,
+      "loss": 0.9717,
+      "step": 13883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0502330680750366e-05,
+      "loss": 0.9189,
+      "step": 13884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0499588936038479e-05,
+      "loss": 0.8877,
+      "step": 13885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0496847421873629e-05,
+      "loss": 0.7383,
+      "step": 13886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0494106138322342e-05,
+      "loss": 0.9541,
+      "step": 13887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0491365085451137e-05,
+      "loss": 0.9766,
+      "step": 13888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0488624263326545e-05,
+      "loss": 0.8389,
+      "step": 13889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0485883672015072e-05,
+      "loss": 0.9541,
+      "step": 13890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0483143311583222e-05,
+      "loss": 0.9697,
+      "step": 13891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0480403182097485e-05,
+      "loss": 0.916,
+      "step": 13892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0477663283624375e-05,
+      "loss": 0.8906,
+      "step": 13893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0474923616230374e-05,
+      "loss": 0.9883,
+      "step": 13894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0472184179981966e-05,
+      "loss": 0.9609,
+      "step": 13895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0469444974945619e-05,
+      "loss": 0.8945,
+      "step": 13896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.046670600118782e-05,
+      "loss": 0.9014,
+      "step": 13897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0463967258775031e-05,
+      "loss": 0.9268,
+      "step": 13898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0461228747773712e-05,
+      "loss": 0.7324,
+      "step": 13899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0458490468250312e-05,
+      "loss": 0.9209,
+      "step": 13900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0455752420271295e-05,
+      "loss": 0.79,
+      "step": 13901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0453014603903098e-05,
+      "loss": 0.8066,
+      "step": 13902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0450277019212152e-05,
+      "loss": 0.9473,
+      "step": 13903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0447539666264908e-05,
+      "loss": 0.8857,
+      "step": 13904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0444802545127781e-05,
+      "loss": 0.9336,
+      "step": 13905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0442065655867197e-05,
+      "loss": 0.9004,
+      "step": 13906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.043932899854956e-05,
+      "loss": 0.7842,
+      "step": 13907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0436592573241301e-05,
+      "loss": 0.8799,
+      "step": 13908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0433856380008816e-05,
+      "loss": 0.958,
+      "step": 13909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0431120418918503e-05,
+      "loss": 0.9014,
+      "step": 13910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0428384690036749e-05,
+      "loss": 0.751,
+      "step": 13911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0425649193429956e-05,
+      "loss": 0.9893,
+      "step": 13912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0422913929164503e-05,
+      "loss": 0.8428,
+      "step": 13913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0420178897306759e-05,
+      "loss": 0.7393,
+      "step": 13914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0417444097923095e-05,
+      "loss": 0.8936,
+      "step": 13915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0414709531079882e-05,
+      "loss": 0.9365,
+      "step": 13916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0411975196843485e-05,
+      "loss": 0.8438,
+      "step": 13917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0409241095280253e-05,
+      "loss": 0.9209,
+      "step": 13918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0406507226456535e-05,
+      "loss": 0.7744,
+      "step": 13919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0403773590438666e-05,
+      "loss": 0.8535,
+      "step": 13920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0401040187292997e-05,
+      "loss": 0.8945,
+      "step": 13921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0398307017085853e-05,
+      "loss": 0.7876,
+      "step": 13922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0395574079883562e-05,
+      "loss": 0.8818,
+      "step": 13923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0392841375752433e-05,
+      "loss": 0.8047,
+      "step": 13924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.03901089047588e-05,
+      "loss": 0.7349,
+      "step": 13925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0387376666968958e-05,
+      "loss": 0.8555,
+      "step": 13926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0384644662449219e-05,
+      "loss": 0.8252,
+      "step": 13927
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0381912891265866e-05,
+      "loss": 0.9121,
+      "step": 13928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0379181353485211e-05,
+      "loss": 0.7383,
+      "step": 13929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0376450049173528e-05,
+      "loss": 0.8291,
+      "step": 13930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0373718978397105e-05,
+      "loss": 0.8467,
+      "step": 13931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.03709881412222e-05,
+      "loss": 0.7812,
+      "step": 13932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0368257537715105e-05,
+      "loss": 0.7285,
+      "step": 13933
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0365527167942073e-05,
+      "loss": 0.8945,
+      "step": 13934
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0362797031969363e-05,
+      "loss": 0.7266,
+      "step": 13935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0360067129863218e-05,
+      "loss": 0.9648,
+      "step": 13936
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0357337461689902e-05,
+      "loss": 0.8701,
+      "step": 13937
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.035460802751565e-05,
+      "loss": 0.8145,
+      "step": 13938
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0351878827406685e-05,
+      "loss": 0.9512,
+      "step": 13939
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0349149861429254e-05,
+      "loss": 0.916,
+      "step": 13940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0346421129649577e-05,
+      "loss": 0.873,
+      "step": 13941
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0343692632133865e-05,
+      "loss": 1.0,
+      "step": 13942
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0340964368948332e-05,
+      "loss": 0.7393,
+      "step": 13943
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0338236340159191e-05,
+      "loss": 0.9082,
+      "step": 13944
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0335508545832642e-05,
+      "loss": 0.7461,
+      "step": 13945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0332780986034876e-05,
+      "loss": 0.8975,
+      "step": 13946
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.033005366083208e-05,
+      "loss": 0.791,
+      "step": 13947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0327326570290447e-05,
+      "loss": 0.8477,
+      "step": 13948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0324599714476155e-05,
+      "loss": 0.6973,
+      "step": 13949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0321873093455374e-05,
+      "loss": 0.8525,
+      "step": 13950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0319146707294263e-05,
+      "loss": 0.9199,
+      "step": 13951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0316420556058996e-05,
+      "loss": 0.8066,
+      "step": 13952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0313694639815728e-05,
+      "loss": 0.7314,
+      "step": 13953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0310968958630601e-05,
+      "loss": 0.9287,
+      "step": 13954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0308243512569759e-05,
+      "loss": 0.79,
+      "step": 13955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0305518301699349e-05,
+      "loss": 0.77,
+      "step": 13956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0302793326085499e-05,
+      "loss": 0.7803,
+      "step": 13957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0300068585794338e-05,
+      "loss": 0.9189,
+      "step": 13958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.029734408089198e-05,
+      "loss": 0.9961,
+      "step": 13959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0294619811444551e-05,
+      "loss": 0.8662,
+      "step": 13960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.029189577751816e-05,
+      "loss": 0.7695,
+      "step": 13961
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0289171979178905e-05,
+      "loss": 0.9258,
+      "step": 13962
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.028644841649288e-05,
+      "loss": 0.8682,
+      "step": 13963
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0283725089526193e-05,
+      "loss": 0.7646,
+      "step": 13964
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0281001998344926e-05,
+      "loss": 0.916,
+      "step": 13965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0278279143015157e-05,
+      "loss": 0.8486,
+      "step": 13966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0275556523602954e-05,
+      "loss": 0.834,
+      "step": 13967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0272834140174404e-05,
+      "loss": 0.8828,
+      "step": 13968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0270111992795564e-05,
+      "loss": 0.8906,
+      "step": 13969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0267390081532488e-05,
+      "loss": 0.8535,
+      "step": 13970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0264668406451226e-05,
+      "loss": 0.9639,
+      "step": 13971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.026194696761784e-05,
+      "loss": 0.8662,
+      "step": 13972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0259225765098361e-05,
+      "loss": 1.0078,
+      "step": 13973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0256504798958828e-05,
+      "loss": 0.7744,
+      "step": 13974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.025378406926526e-05,
+      "loss": 0.9209,
+      "step": 13975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0251063576083697e-05,
+      "loss": 0.9609,
+      "step": 13976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0248343319480152e-05,
+      "loss": 0.9902,
+      "step": 13977
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0245623299520635e-05,
+      "loss": 0.8096,
+      "step": 13978
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0242903516271148e-05,
+      "loss": 0.7949,
+      "step": 13979
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0240183969797704e-05,
+      "loss": 0.9102,
+      "step": 13980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0237464660166295e-05,
+      "loss": 0.9121,
+      "step": 13981
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0234745587442902e-05,
+      "loss": 0.8867,
+      "step": 13982
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0232026751693511e-05,
+      "loss": 0.999,
+      "step": 13983
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0229308152984114e-05,
+      "loss": 0.8623,
+      "step": 13984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.022658979138067e-05,
+      "loss": 0.8203,
+      "step": 13985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0223871666949146e-05,
+      "loss": 0.7793,
+      "step": 13986
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0221153779755508e-05,
+      "loss": 0.9336,
+      "step": 13987
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0218436129865715e-05,
+      "loss": 0.876,
+      "step": 13988
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0215718717345707e-05,
+      "loss": 0.7988,
+      "step": 13989
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0213001542261422e-05,
+      "loss": 0.958,
+      "step": 13990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0210284604678819e-05,
+      "loss": 0.9043,
+      "step": 13991
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0207567904663817e-05,
+      "loss": 0.8467,
+      "step": 13992
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0204851442282342e-05,
+      "loss": 0.9023,
+      "step": 13993
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0202135217600309e-05,
+      "loss": 0.9414,
+      "step": 13994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0199419230683648e-05,
+      "loss": 0.8545,
+      "step": 13995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0196703481598258e-05,
+      "loss": 1.0,
+      "step": 13996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0193987970410046e-05,
+      "loss": 0.8633,
+      "step": 13997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.01912726971849e-05,
+      "loss": 0.8857,
+      "step": 13998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0188557661988724e-05,
+      "loss": 0.9082,
+      "step": 13999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0185842864887392e-05,
+      "loss": 0.8809,
+      "step": 14000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.01831283059468e-05,
+      "loss": 0.8228,
+      "step": 14001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0180413985232816e-05,
+      "loss": 0.875,
+      "step": 14002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0177699902811295e-05,
+      "loss": 0.9482,
+      "step": 14003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0174986058748122e-05,
+      "loss": 0.9229,
+      "step": 14004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.017227245310914e-05,
+      "loss": 0.8193,
+      "step": 14005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0169559085960205e-05,
+      "loss": 0.8838,
+      "step": 14006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0166845957367153e-05,
+      "loss": 0.8506,
+      "step": 14007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0164133067395838e-05,
+      "loss": 0.8193,
+      "step": 14008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0161420416112086e-05,
+      "loss": 0.915,
+      "step": 14009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.015870800358173e-05,
+      "loss": 0.7598,
+      "step": 14010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0155995829870578e-05,
+      "loss": 0.8906,
+      "step": 14011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0153283895044467e-05,
+      "loss": 0.9727,
+      "step": 14012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0150572199169195e-05,
+      "loss": 0.9707,
+      "step": 14013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0147860742310571e-05,
+      "loss": 0.9365,
+      "step": 14014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0145149524534386e-05,
+      "loss": 0.8916,
+      "step": 14015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0142438545906446e-05,
+      "loss": 0.8213,
+      "step": 14016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0139727806492532e-05,
+      "loss": 0.9297,
+      "step": 14017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0137017306358428e-05,
+      "loss": 1.0137,
+      "step": 14018
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.01343070455699e-05,
+      "loss": 0.8643,
+      "step": 14019
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0131597024192735e-05,
+      "loss": 0.9541,
+      "step": 14020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0128887242292684e-05,
+      "loss": 0.9014,
+      "step": 14021
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0126177699935513e-05,
+      "loss": 0.8945,
+      "step": 14022
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0123468397186962e-05,
+      "loss": 0.8066,
+      "step": 14023
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0120759334112797e-05,
+      "loss": 0.6909,
+      "step": 14024
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0118050510778747e-05,
+      "loss": 0.9297,
+      "step": 14025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0115341927250544e-05,
+      "loss": 0.876,
+      "step": 14026
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0112633583593928e-05,
+      "loss": 0.8745,
+      "step": 14027
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0109925479874617e-05,
+      "loss": 0.957,
+      "step": 14028
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0107217616158329e-05,
+      "loss": 0.8408,
+      "step": 14029
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0104509992510772e-05,
+      "loss": 0.8145,
+      "step": 14030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0101802608997659e-05,
+      "loss": 0.8936,
+      "step": 14031
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0099095465684687e-05,
+      "loss": 0.9229,
+      "step": 14032
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0096388562637553e-05,
+      "loss": 0.6885,
+      "step": 14033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0093681899921936e-05,
+      "loss": 0.9727,
+      "step": 14034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0090975477603532e-05,
+      "loss": 0.9297,
+      "step": 14035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0088269295748013e-05,
+      "loss": 0.8457,
+      "step": 14036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0085563354421048e-05,
+      "loss": 0.873,
+      "step": 14037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0082857653688296e-05,
+      "loss": 0.7998,
+      "step": 14038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.008015219361543e-05,
+      "loss": 0.9971,
+      "step": 14039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0077446974268098e-05,
+      "loss": 0.6094,
+      "step": 14040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0074741995711945e-05,
+      "loss": 0.9854,
+      "step": 14041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.007203725801261e-05,
+      "loss": 0.6187,
+      "step": 14042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.006933276123574e-05,
+      "loss": 0.8115,
+      "step": 14043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0066628505446956e-05,
+      "loss": 0.8301,
+      "step": 14044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0063924490711886e-05,
+      "loss": 0.9482,
+      "step": 14045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0061220717096142e-05,
+      "loss": 0.9629,
+      "step": 14046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0058517184665351e-05,
+      "loss": 0.9453,
+      "step": 14047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0055813893485112e-05,
+      "loss": 0.7715,
+      "step": 14048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0053110843621022e-05,
+      "loss": 0.835,
+      "step": 14049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0050408035138672e-05,
+      "loss": 0.8413,
+      "step": 14050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0047705468103665e-05,
+      "loss": 0.9326,
+      "step": 14051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0045003142581579e-05,
+      "loss": 0.9648,
+      "step": 14052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0042301058637988e-05,
+      "loss": 0.8574,
+      "step": 14053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0039599216338463e-05,
+      "loss": 0.9463,
+      "step": 14054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0036897615748575e-05,
+      "loss": 0.998,
+      "step": 14055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0034196256933884e-05,
+      "loss": 0.8857,
+      "step": 14056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0031495139959941e-05,
+      "loss": 1.0059,
+      "step": 14057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0028794264892288e-05,
+      "loss": 1.0781,
+      "step": 14058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0026093631796481e-05,
+      "loss": 0.6172,
+      "step": 14059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0023393240738048e-05,
+      "loss": 1.1143,
+      "step": 14060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0020693091782521e-05,
+      "loss": 0.9355,
+      "step": 14061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.001799318499542e-05,
+      "loss": 0.8516,
+      "step": 14062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0015293520442272e-05,
+      "loss": 0.9424,
+      "step": 14063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0012594098188588e-05,
+      "loss": 0.9502,
+      "step": 14064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0009894918299874e-05,
+      "loss": 0.9434,
+      "step": 14065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0007195980841625e-05,
+      "loss": 1.0977,
+      "step": 14066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0004497285879345e-05,
+      "loss": 0.9043,
+      "step": 14067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.000179883347852e-05,
+      "loss": 0.8799,
+      "step": 14068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.999100623704636e-06,
+      "loss": 0.8115,
+      "step": 14069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.996402656623161e-06,
+      "loss": 0.875,
+      "step": 14070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.99370493229958e-06,
+      "loss": 0.8223,
+      "step": 14071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.991007450799355e-06,
+      "loss": 0.8506,
+      "step": 14072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.988310212187933e-06,
+      "loss": 0.7363,
+      "step": 14073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.98561321653079e-06,
+      "loss": 0.8008,
+      "step": 14074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.98291646389336e-06,
+      "loss": 0.9307,
+      "step": 14075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.98021995434109e-06,
+      "loss": 0.9404,
+      "step": 14076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.977523687939406e-06,
+      "loss": 0.8955,
+      "step": 14077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.974827664753754e-06,
+      "loss": 0.791,
+      "step": 14078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.972131884849554e-06,
+      "loss": 0.9121,
+      "step": 14079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.969436348292221e-06,
+      "loss": 0.9707,
+      "step": 14080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.96674105514716e-06,
+      "loss": 0.793,
+      "step": 14081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.964046005479797e-06,
+      "loss": 0.5869,
+      "step": 14082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.961351199355513e-06,
+      "loss": 0.9043,
+      "step": 14083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.958656636839724e-06,
+      "loss": 0.8447,
+      "step": 14084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.955962317997802e-06,
+      "loss": 0.9424,
+      "step": 14085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.953268242895133e-06,
+      "loss": 0.9814,
+      "step": 14086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.950574411597102e-06,
+      "loss": 0.8613,
+      "step": 14087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.947880824169079e-06,
+      "loss": 0.8662,
+      "step": 14088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.945187480676424e-06,
+      "loss": 0.7827,
+      "step": 14089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.942494381184493e-06,
+      "loss": 0.6816,
+      "step": 14090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.93980152575865e-06,
+      "loss": 0.7705,
+      "step": 14091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.93710891446424e-06,
+      "loss": 0.7686,
+      "step": 14092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.934416547366604e-06,
+      "loss": 1.0293,
+      "step": 14093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.931724424531066e-06,
+      "loss": 1.0508,
+      "step": 14094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.929032546022976e-06,
+      "loss": 0.7227,
+      "step": 14095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.926340911907647e-06,
+      "loss": 0.9326,
+      "step": 14096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.9236495222504e-06,
+      "loss": 0.833,
+      "step": 14097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.920958377116536e-06,
+      "loss": 0.8525,
+      "step": 14098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.918267476571381e-06,
+      "loss": 0.8252,
+      "step": 14099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.915576820680224e-06,
+      "loss": 0.9854,
+      "step": 14100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.91288640950836e-06,
+      "loss": 0.9443,
+      "step": 14101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.910196243121071e-06,
+      "loss": 0.917,
+      "step": 14102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.907506321583654e-06,
+      "loss": 0.79,
+      "step": 14103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.904816644961375e-06,
+      "loss": 1.0146,
+      "step": 14104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.902127213319508e-06,
+      "loss": 1.0742,
+      "step": 14105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.89943802672331e-06,
+      "loss": 0.7227,
+      "step": 14106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.896749085238052e-06,
+      "loss": 0.9658,
+      "step": 14107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.894060388928986e-06,
+      "loss": 0.8047,
+      "step": 14108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.89137193786135e-06,
+      "loss": 0.915,
+      "step": 14109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.88868373210038e-06,
+      "loss": 0.8389,
+      "step": 14110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.885995771711331e-06,
+      "loss": 0.8672,
+      "step": 14111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.88330805675942e-06,
+      "loss": 0.8613,
+      "step": 14112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.88062058730986e-06,
+      "loss": 0.7056,
+      "step": 14113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.877933363427888e-06,
+      "loss": 0.8604,
+      "step": 14114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.875246385178703e-06,
+      "loss": 0.7993,
+      "step": 14115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.872559652627514e-06,
+      "loss": 0.9414,
+      "step": 14116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.869873165839511e-06,
+      "loss": 0.9326,
+      "step": 14117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.867186924879901e-06,
+      "loss": 0.7725,
+      "step": 14118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.864500929813867e-06,
+      "loss": 0.9229,
+      "step": 14119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.861815180706587e-06,
+      "loss": 0.8623,
+      "step": 14120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.85912967762323e-06,
+      "loss": 0.957,
+      "step": 14121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.856444420628979e-06,
+      "loss": 0.9463,
+      "step": 14122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.85375940978899e-06,
+      "loss": 0.7822,
+      "step": 14123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.851074645168423e-06,
+      "loss": 0.8291,
+      "step": 14124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.84839012683242e-06,
+      "loss": 0.6084,
+      "step": 14125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.84570585484614e-06,
+      "loss": 1.0244,
+      "step": 14126
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.843021829274716e-06,
+      "loss": 0.8896,
+      "step": 14127
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.840338050183284e-06,
+      "loss": 0.6826,
+      "step": 14128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.837654517636959e-06,
+      "loss": 0.8877,
+      "step": 14129
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.834971231700881e-06,
+      "loss": 0.8477,
+      "step": 14130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.832288192440158e-06,
+      "loss": 0.8232,
+      "step": 14131
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.829605399919898e-06,
+      "loss": 0.9248,
+      "step": 14132
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.826922854205196e-06,
+      "loss": 0.8975,
+      "step": 14133
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.824240555361166e-06,
+      "loss": 0.832,
+      "step": 14134
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.821558503452893e-06,
+      "loss": 0.9033,
+      "step": 14135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.81887669854546e-06,
+      "loss": 0.9062,
+      "step": 14136
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.816195140703942e-06,
+      "loss": 0.8701,
+      "step": 14137
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.813513829993423e-06,
+      "loss": 0.9238,
+      "step": 14138
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.81083276647897e-06,
+      "loss": 0.7227,
+      "step": 14139
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.80815195022564e-06,
+      "loss": 0.8721,
+      "step": 14140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.80547138129848e-06,
+      "loss": 0.957,
+      "step": 14141
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.802791059762556e-06,
+      "loss": 0.7979,
+      "step": 14142
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.800110985682905e-06,
+      "loss": 0.8193,
+      "step": 14143
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.797431159124562e-06,
+      "loss": 0.8604,
+      "step": 14144
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.794751580152553e-06,
+      "loss": 0.8379,
+      "step": 14145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.79207224883192e-06,
+      "loss": 0.8154,
+      "step": 14146
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.78939316522767e-06,
+      "loss": 0.8408,
+      "step": 14147
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.78671432940482e-06,
+      "loss": 0.8589,
+      "step": 14148
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.784035741428373e-06,
+      "loss": 0.8975,
+      "step": 14149
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.781357401363338e-06,
+      "loss": 0.7949,
+      "step": 14150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.77867930927471e-06,
+      "loss": 0.8965,
+      "step": 14151
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.776001465227473e-06,
+      "loss": 0.9141,
+      "step": 14152
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.773323869286606e-06,
+      "loss": 0.9717,
+      "step": 14153
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.770646521517102e-06,
+      "loss": 0.7949,
+      "step": 14154
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.767969421983922e-06,
+      "loss": 0.8345,
+      "step": 14155
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.765292570752034e-06,
+      "loss": 1.0586,
+      "step": 14156
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.762615967886386e-06,
+      "loss": 0.9473,
+      "step": 14157
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.759939613451951e-06,
+      "loss": 0.7949,
+      "step": 14158
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.757263507513667e-06,
+      "loss": 0.8984,
+      "step": 14159
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.754587650136467e-06,
+      "loss": 0.7656,
+      "step": 14160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.751912041385301e-06,
+      "loss": 0.8809,
+      "step": 14161
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.749236681325091e-06,
+      "loss": 0.9717,
+      "step": 14162
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.74656157002076e-06,
+      "loss": 0.9736,
+      "step": 14163
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.74388670753722e-06,
+      "loss": 0.8447,
+      "step": 14164
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.741212093939393e-06,
+      "loss": 0.7793,
+      "step": 14165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.738537729292172e-06,
+      "loss": 0.7607,
+      "step": 14166
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.73586361366047e-06,
+      "loss": 0.7988,
+      "step": 14167
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.73318974710917e-06,
+      "loss": 0.8555,
+      "step": 14168
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.730516129703158e-06,
+      "loss": 0.8887,
+      "step": 14169
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.727842761507323e-06,
+      "loss": 0.8057,
+      "step": 14170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.725169642586534e-06,
+      "loss": 0.9229,
+      "step": 14171
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.722496773005662e-06,
+      "loss": 0.8965,
+      "step": 14172
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.71982415282956e-06,
+      "loss": 0.9189,
+      "step": 14173
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.717151782123102e-06,
+      "loss": 0.8071,
+      "step": 14174
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.714479660951128e-06,
+      "loss": 0.8223,
+      "step": 14175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.711807789378484e-06,
+      "loss": 0.8623,
+      "step": 14176
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.709136167470001e-06,
+      "loss": 1.0146,
+      "step": 14177
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.706464795290526e-06,
+      "loss": 0.9189,
+      "step": 14178
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.70379367290488e-06,
+      "loss": 0.9902,
+      "step": 14179
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.70112280037788e-06,
+      "loss": 0.6914,
+      "step": 14180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.698452177774336e-06,
+      "loss": 0.8374,
+      "step": 14181
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.695781805159066e-06,
+      "loss": 0.8926,
+      "step": 14182
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.69311168259687e-06,
+      "loss": 0.917,
+      "step": 14183
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.690441810152542e-06,
+      "loss": 0.9756,
+      "step": 14184
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.687772187890865e-06,
+      "loss": 0.9492,
+      "step": 14185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.685102815876636e-06,
+      "loss": 0.9346,
+      "step": 14186
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.68243369417463e-06,
+      "loss": 0.7451,
+      "step": 14187
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.679764822849611e-06,
+      "loss": 0.7959,
+      "step": 14188
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.677096201966343e-06,
+      "loss": 0.7754,
+      "step": 14189
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.6744278315896e-06,
+      "loss": 0.9922,
+      "step": 14190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.671759711784126e-06,
+      "loss": 0.7861,
+      "step": 14191
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.66909184261467e-06,
+      "loss": 0.9668,
+      "step": 14192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.666424224145965e-06,
+      "loss": 0.8838,
+      "step": 14193
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.663756856442763e-06,
+      "loss": 0.9258,
+      "step": 14194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.661089739569782e-06,
+      "loss": 0.8428,
+      "step": 14195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.658422873591749e-06,
+      "loss": 0.8701,
+      "step": 14196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.655756258573373e-06,
+      "loss": 0.7529,
+      "step": 14197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.653089894579377e-06,
+      "loss": 0.8203,
+      "step": 14198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.65042378167446e-06,
+      "loss": 0.9219,
+      "step": 14199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.647757919923312e-06,
+      "loss": 0.8662,
+      "step": 14200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.645092309390644e-06,
+      "loss": 0.7515,
+      "step": 14201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.642426950141133e-06,
+      "loss": 0.8965,
+      "step": 14202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.63976184223946e-06,
+      "loss": 1.0049,
+      "step": 14203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.63709698575029e-06,
+      "loss": 0.8311,
+      "step": 14204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.634432380738308e-06,
+      "loss": 0.7959,
+      "step": 14205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.631768027268166e-06,
+      "loss": 0.71,
+      "step": 14206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.629103925404526e-06,
+      "loss": 0.7461,
+      "step": 14207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.626440075212024e-06,
+      "loss": 0.9395,
+      "step": 14208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.623776476755322e-06,
+      "loss": 1.043,
+      "step": 14209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.621113130099049e-06,
+      "loss": 0.7832,
+      "step": 14210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.618450035307838e-06,
+      "loss": 0.9072,
+      "step": 14211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.615787192446304e-06,
+      "loss": 0.7715,
+      "step": 14212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.613124601579084e-06,
+      "loss": 1.0322,
+      "step": 14213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.610462262770786e-06,
+      "loss": 0.8574,
+      "step": 14214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.607800176086013e-06,
+      "loss": 0.9014,
+      "step": 14215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.60513834158936e-06,
+      "loss": 0.8252,
+      "step": 14216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.602476759345434e-06,
+      "loss": 0.8564,
+      "step": 14217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.599815429418821e-06,
+      "loss": 0.9424,
+      "step": 14218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.597154351874103e-06,
+      "loss": 0.7891,
+      "step": 14219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.594493526775845e-06,
+      "loss": 0.8291,
+      "step": 14220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.591832954188639e-06,
+      "loss": 0.9658,
+      "step": 14221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.589172634177035e-06,
+      "loss": 0.8486,
+      "step": 14222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.586512566805595e-06,
+      "loss": 0.9873,
+      "step": 14223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.583852752138862e-06,
+      "loss": 0.79,
+      "step": 14224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.581193190241398e-06,
+      "loss": 0.8203,
+      "step": 14225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.578533881177733e-06,
+      "loss": 0.7178,
+      "step": 14226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.575874825012403e-06,
+      "loss": 1.0391,
+      "step": 14227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.573216021809928e-06,
+      "loss": 0.9512,
+      "step": 14228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.570557471634845e-06,
+      "loss": 0.8672,
+      "step": 14229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.567899174551658e-06,
+      "loss": 0.918,
+      "step": 14230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.565241130624878e-06,
+      "loss": 0.8955,
+      "step": 14231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.562583339919003e-06,
+      "loss": 0.9844,
+      "step": 14232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.55992580249854e-06,
+      "loss": 0.6436,
+      "step": 14233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.557268518427976e-06,
+      "loss": 0.959,
+      "step": 14234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.554611487771795e-06,
+      "loss": 0.917,
+      "step": 14235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.551954710594464e-06,
+      "loss": 0.9033,
+      "step": 14236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.549298186960476e-06,
+      "loss": 0.918,
+      "step": 14237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.546641916934285e-06,
+      "loss": 0.9121,
+      "step": 14238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.543985900580352e-06,
+      "loss": 0.8896,
+      "step": 14239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.541330137963126e-06,
+      "loss": 1.0723,
+      "step": 14240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.538674629147065e-06,
+      "loss": 0.7246,
+      "step": 14241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.536019374196605e-06,
+      "loss": 0.9678,
+      "step": 14242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.533364373176175e-06,
+      "loss": 0.7139,
+      "step": 14243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.530709626150219e-06,
+      "loss": 0.9258,
+      "step": 14244
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.528055133183149e-06,
+      "loss": 0.9639,
+      "step": 14245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.525400894339385e-06,
+      "loss": 0.8428,
+      "step": 14246
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.52274690968333e-06,
+      "loss": 0.7959,
+      "step": 14247
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.520093179279402e-06,
+      "loss": 1.0957,
+      "step": 14248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.51743970319199e-06,
+      "loss": 0.7725,
+      "step": 14249
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.514786481485483e-06,
+      "loss": 0.9561,
+      "step": 14250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.51213351422428e-06,
+      "loss": 0.6978,
+      "step": 14251
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.509480801472744e-06,
+      "loss": 0.8896,
+      "step": 14252
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.506828343295267e-06,
+      "loss": 0.8008,
+      "step": 14253
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.504176139756205e-06,
+      "loss": 0.9004,
+      "step": 14254
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.50152419091992e-06,
+      "loss": 0.9297,
+      "step": 14255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.498872496850762e-06,
+      "loss": 0.9863,
+      "step": 14256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.496221057613094e-06,
+      "loss": 0.8911,
+      "step": 14257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.493569873271248e-06,
+      "loss": 1.0332,
+      "step": 14258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.490918943889567e-06,
+      "loss": 0.8584,
+      "step": 14259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.488268269532366e-06,
+      "loss": 0.834,
+      "step": 14260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.485617850263988e-06,
+      "loss": 0.7676,
+      "step": 14261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.482967686148744e-06,
+      "loss": 0.7441,
+      "step": 14262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.480317777250945e-06,
+      "loss": 0.9541,
+      "step": 14263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.477668123634889e-06,
+      "loss": 0.877,
+      "step": 14264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.475018725364888e-06,
+      "loss": 0.9355,
+      "step": 14265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.472369582505232e-06,
+      "loss": 0.9512,
+      "step": 14266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.469720695120205e-06,
+      "loss": 0.9365,
+      "step": 14267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.46707206327408e-06,
+      "loss": 1.0205,
+      "step": 14268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.464423687031148e-06,
+      "loss": 0.8438,
+      "step": 14269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.461775566455668e-06,
+      "loss": 0.9316,
+      "step": 14270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.459127701611906e-06,
+      "loss": 0.8848,
+      "step": 14271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.456480092564106e-06,
+      "loss": 0.9277,
+      "step": 14272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.453832739376534e-06,
+      "loss": 0.9736,
+      "step": 14273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.451185642113425e-06,
+      "loss": 0.96,
+      "step": 14274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.44853880083902e-06,
+      "loss": 0.8438,
+      "step": 14275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.445892215617538e-06,
+      "loss": 0.7642,
+      "step": 14276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.443245886513223e-06,
+      "loss": 0.8467,
+      "step": 14277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.440599813590284e-06,
+      "loss": 0.9385,
+      "step": 14278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.437953996912932e-06,
+      "loss": 0.9219,
+      "step": 14279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.435308436545368e-06,
+      "loss": 0.9385,
+      "step": 14280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.432663132551806e-06,
+      "loss": 0.7725,
+      "step": 14281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.430018084996431e-06,
+      "loss": 1.0283,
+      "step": 14282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.427373293943425e-06,
+      "loss": 0.6797,
+      "step": 14283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.424728759456982e-06,
+      "loss": 0.6929,
+      "step": 14284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.422084481601274e-06,
+      "loss": 0.5874,
+      "step": 14285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.419440460440465e-06,
+      "loss": 0.9531,
+      "step": 14286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.416796696038715e-06,
+      "loss": 0.7998,
+      "step": 14287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.414153188460191e-06,
+      "loss": 0.998,
+      "step": 14288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.411509937769037e-06,
+      "loss": 0.9043,
+      "step": 14289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.408866944029398e-06,
+      "loss": 0.9395,
+      "step": 14290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.406224207305404e-06,
+      "loss": 0.874,
+      "step": 14291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.4035817276612e-06,
+      "loss": 0.8984,
+      "step": 14292
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.400939505160902e-06,
+      "loss": 0.8652,
+      "step": 14293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.398297539868636e-06,
+      "loss": 0.9619,
+      "step": 14294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.395655831848498e-06,
+      "loss": 0.9111,
+      "step": 14295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.393014381164618e-06,
+      "loss": 0.9814,
+      "step": 14296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.390373187881083e-06,
+      "loss": 0.7695,
+      "step": 14297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.38773225206199e-06,
+      "loss": 0.7051,
+      "step": 14298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.385091573771419e-06,
+      "loss": 0.8477,
+      "step": 14299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.382451153073465e-06,
+      "loss": 0.8232,
+      "step": 14300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.379810990032194e-06,
+      "loss": 0.9248,
+      "step": 14301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.37717108471168e-06,
+      "loss": 0.9609,
+      "step": 14302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.374531437175974e-06,
+      "loss": 1.0107,
+      "step": 14303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.37189204748915e-06,
+      "loss": 0.8491,
+      "step": 14304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.36925291571525e-06,
+      "loss": 0.9189,
+      "step": 14305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.366614041918318e-06,
+      "loss": 0.7129,
+      "step": 14306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.363975426162383e-06,
+      "loss": 0.8369,
+      "step": 14307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.361337068511493e-06,
+      "loss": 0.9092,
+      "step": 14308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.358698969029666e-06,
+      "loss": 0.9062,
+      "step": 14309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.35606112778092e-06,
+      "loss": 0.9824,
+      "step": 14310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.353423544829261e-06,
+      "loss": 0.6992,
+      "step": 14311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.350786220238707e-06,
+      "loss": 0.7393,
+      "step": 14312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.348149154073254e-06,
+      "loss": 0.8613,
+      "step": 14313
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.345512346396897e-06,
+      "loss": 0.8887,
+      "step": 14314
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.342875797273612e-06,
+      "loss": 0.8594,
+      "step": 14315
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.340239506767399e-06,
+      "loss": 0.7461,
+      "step": 14316
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.337603474942222e-06,
+      "loss": 0.9434,
+      "step": 14317
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.334967701862055e-06,
+      "loss": 1.0557,
+      "step": 14318
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.332332187590847e-06,
+      "loss": 0.9082,
+      "step": 14319
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.329696932192571e-06,
+      "loss": 0.9414,
+      "step": 14320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.327061935731172e-06,
+      "loss": 0.624,
+      "step": 14321
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.324427198270591e-06,
+      "loss": 0.9072,
+      "step": 14322
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.321792719874761e-06,
+      "loss": 0.9072,
+      "step": 14323
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.319158500607621e-06,
+      "loss": 0.8672,
+      "step": 14324
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.316524540533096e-06,
+      "loss": 0.8955,
+      "step": 14325
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.313890839715103e-06,
+      "loss": 0.9961,
+      "step": 14326
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.311257398217542e-06,
+      "loss": 0.8848,
+      "step": 14327
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.308624216104339e-06,
+      "loss": 0.793,
+      "step": 14328
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.305991293439383e-06,
+      "loss": 0.75,
+      "step": 14329
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.30335863028656e-06,
+      "loss": 0.9795,
+      "step": 14330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.300726226709776e-06,
+      "loss": 0.9102,
+      "step": 14331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.298094082772897e-06,
+      "loss": 0.9463,
+      "step": 14332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.295462198539797e-06,
+      "loss": 1.0557,
+      "step": 14333
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.292830574074355e-06,
+      "loss": 0.8569,
+      "step": 14334
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.290199209440418e-06,
+      "loss": 0.8535,
+      "step": 14335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.28756810470186e-06,
+      "loss": 0.9414,
+      "step": 14336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.284937259922517e-06,
+      "loss": 0.8066,
+      "step": 14337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.282306675166238e-06,
+      "loss": 0.9697,
+      "step": 14338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.279676350496844e-06,
+      "loss": 0.8184,
+      "step": 14339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.277046285978188e-06,
+      "loss": 0.7715,
+      "step": 14340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.274416481674084e-06,
+      "loss": 0.7197,
+      "step": 14341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.271786937648349e-06,
+      "loss": 0.8604,
+      "step": 14342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.269157653964787e-06,
+      "loss": 0.9277,
+      "step": 14343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.266528630687217e-06,
+      "loss": 0.9019,
+      "step": 14344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.263899867879431e-06,
+      "loss": 0.8286,
+      "step": 14345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.261271365605224e-06,
+      "loss": 0.8877,
+      "step": 14346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.25864312392837e-06,
+      "loss": 0.876,
+      "step": 14347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.256015142912666e-06,
+      "loss": 0.7852,
+      "step": 14348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.253387422621878e-06,
+      "loss": 0.9355,
+      "step": 14349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.25075996311977e-06,
+      "loss": 0.8789,
+      "step": 14350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.248132764470101e-06,
+      "loss": 0.8291,
+      "step": 14351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.245505826736635e-06,
+      "loss": 0.8271,
+      "step": 14352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.242879149983118e-06,
+      "loss": 0.8564,
+      "step": 14353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.240252734273284e-06,
+      "loss": 0.8945,
+      "step": 14354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.237626579670869e-06,
+      "loss": 0.751,
+      "step": 14355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.235000686239613e-06,
+      "loss": 0.8672,
+      "step": 14356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.23237505404323e-06,
+      "loss": 0.751,
+      "step": 14357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.22974968314544e-06,
+      "loss": 0.6997,
+      "step": 14358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.22712457360994e-06,
+      "loss": 0.6865,
+      "step": 14359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.224499725500455e-06,
+      "loss": 1.0039,
+      "step": 14360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.221875138880671e-06,
+      "loss": 0.8525,
+      "step": 14361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.219250813814282e-06,
+      "loss": 0.8057,
+      "step": 14362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.21662675036496e-06,
+      "loss": 0.835,
+      "step": 14363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.214002948596406e-06,
+      "loss": 0.8281,
+      "step": 14364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.211379408572277e-06,
+      "loss": 0.8652,
+      "step": 14365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.20875613035624e-06,
+      "loss": 0.9834,
+      "step": 14366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.206133114011951e-06,
+      "loss": 0.8721,
+      "step": 14367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.203510359603074e-06,
+      "loss": 0.8838,
+      "step": 14368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.20088786719325e-06,
+      "loss": 0.8643,
+      "step": 14369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.198265636846111e-06,
+      "loss": 0.7861,
+      "step": 14370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.195643668625304e-06,
+      "loss": 0.5459,
+      "step": 14371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.193021962594453e-06,
+      "loss": 0.7617,
+      "step": 14372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.190400518817175e-06,
+      "loss": 0.8711,
+      "step": 14373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.18777933735708e-06,
+      "loss": 1.0146,
+      "step": 14374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.185158418277792e-06,
+      "loss": 0.8936,
+      "step": 14375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.1825377616429e-06,
+      "loss": 0.7305,
+      "step": 14376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.179917367516005e-06,
+      "loss": 0.7959,
+      "step": 14377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.177297235960687e-06,
+      "loss": 0.8892,
+      "step": 14378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.174677367040546e-06,
+      "loss": 0.9404,
+      "step": 14379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.17205776081915e-06,
+      "loss": 0.7581,
+      "step": 14380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.169438417360063e-06,
+      "loss": 1.0156,
+      "step": 14381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.16681933672685e-06,
+      "loss": 0.7271,
+      "step": 14382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.164200518983082e-06,
+      "loss": 0.8496,
+      "step": 14383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.161581964192298e-06,
+      "loss": 0.8799,
+      "step": 14384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.158963672418046e-06,
+      "loss": 0.8809,
+      "step": 14385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.156345643723854e-06,
+      "loss": 0.9902,
+      "step": 14386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.153727878173272e-06,
+      "loss": 0.7632,
+      "step": 14387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.151110375829817e-06,
+      "loss": 0.8789,
+      "step": 14388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.148493136757008e-06,
+      "loss": 0.7266,
+      "step": 14389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.14587616101835e-06,
+      "loss": 0.9766,
+      "step": 14390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.143259448677363e-06,
+      "loss": 0.9053,
+      "step": 14391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.14064299979754e-06,
+      "loss": 0.875,
+      "step": 14392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.138026814442378e-06,
+      "loss": 0.8076,
+      "step": 14393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.135410892675351e-06,
+      "loss": 0.8936,
+      "step": 14394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.132795234559959e-06,
+      "loss": 0.8525,
+      "step": 14395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.130179840159668e-06,
+      "loss": 0.7397,
+      "step": 14396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.127564709537943e-06,
+      "loss": 0.8701,
+      "step": 14397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.124949842758242e-06,
+      "loss": 0.9287,
+      "step": 14398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.122335239884033e-06,
+      "loss": 0.9268,
+      "step": 14399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.11972090097876e-06,
+      "loss": 0.9346,
+      "step": 14400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.117106826105859e-06,
+      "loss": 0.7686,
+      "step": 14401
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.114493015328765e-06,
+      "loss": 0.7871,
+      "step": 14402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.111879468710919e-06,
+      "loss": 0.9805,
+      "step": 14403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.109266186315737e-06,
+      "loss": 0.9707,
+      "step": 14404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.106653168206635e-06,
+      "loss": 0.9482,
+      "step": 14405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.10404041444702e-06,
+      "loss": 0.832,
+      "step": 14406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.101427925100306e-06,
+      "loss": 0.9561,
+      "step": 14407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.098815700229885e-06,
+      "loss": 0.9297,
+      "step": 14408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.096203739899147e-06,
+      "loss": 0.915,
+      "step": 14409
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.093592044171473e-06,
+      "loss": 0.9092,
+      "step": 14410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.09098061311025e-06,
+      "loss": 0.7041,
+      "step": 14411
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.088369446778846e-06,
+      "loss": 0.8018,
+      "step": 14412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.085758545240626e-06,
+      "loss": 0.9004,
+      "step": 14413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.08314790855894e-06,
+      "loss": 0.8662,
+      "step": 14414
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.080537536797156e-06,
+      "loss": 0.8633,
+      "step": 14415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.077927430018607e-06,
+      "loss": 0.7666,
+      "step": 14416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.075317588286644e-06,
+      "loss": 0.7305,
+      "step": 14417
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.072708011664591e-06,
+      "loss": 1.002,
+      "step": 14418
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.070098700215784e-06,
+      "loss": 0.6865,
+      "step": 14419
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.067489654003536e-06,
+      "loss": 0.8066,
+      "step": 14420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.064880873091166e-06,
+      "loss": 0.8135,
+      "step": 14421
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.062272357541968e-06,
+      "loss": 0.9238,
+      "step": 14422
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.059664107419264e-06,
+      "loss": 1.0186,
+      "step": 14423
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.057056122786335e-06,
+      "loss": 0.9766,
+      "step": 14424
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.05444840370647e-06,
+      "loss": 0.8574,
+      "step": 14425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.05184095024295e-06,
+      "loss": 0.9102,
+      "step": 14426
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.049233762459057e-06,
+      "loss": 1.0908,
+      "step": 14427
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.046626840418056e-06,
+      "loss": 0.9541,
+      "step": 14428
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.04402018418321e-06,
+      "loss": 0.8184,
+      "step": 14429
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.041413793817764e-06,
+      "loss": 0.9551,
+      "step": 14430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.038807669384986e-06,
+      "loss": 0.7959,
+      "step": 14431
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.036201810948111e-06,
+      "loss": 0.9395,
+      "step": 14432
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.033596218570372e-06,
+      "loss": 0.9121,
+      "step": 14433
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.030990892314996e-06,
+      "loss": 0.9844,
+      "step": 14434
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.028385832245221e-06,
+      "loss": 0.8135,
+      "step": 14435
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.025781038424254e-06,
+      "loss": 0.9463,
+      "step": 14436
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.02317651091531e-06,
+      "loss": 0.8574,
+      "step": 14437
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.020572249781581e-06,
+      "loss": 0.7715,
+      "step": 14438
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.017968255086281e-06,
+      "loss": 0.8613,
+      "step": 14439
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.015364526892594e-06,
+      "loss": 0.8428,
+      "step": 14440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.012761065263707e-06,
+      "loss": 0.8096,
+      "step": 14441
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.010157870262786e-06,
+      "loss": 0.6133,
+      "step": 14442
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.007554941953022e-06,
+      "loss": 0.8369,
+      "step": 14443
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.004952280397573e-06,
+      "loss": 0.791,
+      "step": 14444
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.002349885659599e-06,
+      "loss": 0.8525,
+      "step": 14445
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.99974775780224e-06,
+      "loss": 0.8682,
+      "step": 14446
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.997145896888661e-06,
+      "loss": 0.9844,
+      "step": 14447
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.994544302981993e-06,
+      "loss": 1.0254,
+      "step": 14448
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.991942976145369e-06,
+      "loss": 0.7656,
+      "step": 14449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.989341916441908e-06,
+      "loss": 0.9688,
+      "step": 14450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.986741123934744e-06,
+      "loss": 0.9639,
+      "step": 14451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.984140598686984e-06,
+      "loss": 0.8125,
+      "step": 14452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.981540340761734e-06,
+      "loss": 0.9453,
+      "step": 14453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.978940350222092e-06,
+      "loss": 0.7939,
+      "step": 14454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.97634062713116e-06,
+      "loss": 0.8916,
+      "step": 14455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.973741171552025e-06,
+      "loss": 0.9229,
+      "step": 14456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.971141983547753e-06,
+      "loss": 0.8418,
+      "step": 14457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.968543063181441e-06,
+      "loss": 0.8691,
+      "step": 14458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.965944410516145e-06,
+      "loss": 0.8145,
+      "step": 14459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.963346025614927e-06,
+      "loss": 0.8701,
+      "step": 14460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.960747908540838e-06,
+      "loss": 0.7656,
+      "step": 14461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.95815005935694e-06,
+      "loss": 0.8623,
+      "step": 14462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.955552478126266e-06,
+      "loss": 0.9873,
+      "step": 14463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.952955164911852e-06,
+      "loss": 0.8516,
+      "step": 14464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.950358119776722e-06,
+      "loss": 0.8535,
+      "step": 14465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.947761342783911e-06,
+      "loss": 0.8965,
+      "step": 14466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.94516483399643e-06,
+      "loss": 0.9043,
+      "step": 14467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.942568593477286e-06,
+      "loss": 0.8643,
+      "step": 14468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.939972621289477e-06,
+      "loss": 0.7832,
+      "step": 14469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.937376917496012e-06,
+      "loss": 0.9531,
+      "step": 14470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.934781482159877e-06,
+      "loss": 0.7646,
+      "step": 14471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.932186315344053e-06,
+      "loss": 0.9658,
+      "step": 14472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.929591417111512e-06,
+      "loss": 0.9434,
+      "step": 14473
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.926996787525236e-06,
+      "loss": 0.8252,
+      "step": 14474
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.924402426648184e-06,
+      "loss": 0.7637,
+      "step": 14475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.921808334543311e-06,
+      "loss": 0.9336,
+      "step": 14476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.919214511273567e-06,
+      "loss": 0.728,
+      "step": 14477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.916620956901903e-06,
+      "loss": 0.9121,
+      "step": 14478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.914027671491254e-06,
+      "loss": 0.8408,
+      "step": 14479
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.911434655104552e-06,
+      "loss": 0.918,
+      "step": 14480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.908841907804712e-06,
+      "loss": 0.8955,
+      "step": 14481
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.906249429654668e-06,
+      "loss": 0.9473,
+      "step": 14482
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.903657220717326e-06,
+      "loss": 1.0088,
+      "step": 14483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.901065281055589e-06,
+      "loss": 0.5425,
+      "step": 14484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.89847361073235e-06,
+      "loss": 0.9229,
+      "step": 14485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.895882209810516e-06,
+      "loss": 0.8916,
+      "step": 14486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.893291078352965e-06,
+      "loss": 0.8047,
+      "step": 14487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.890700216422574e-06,
+      "loss": 0.6587,
+      "step": 14488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.888109624082214e-06,
+      "loss": 0.8594,
+      "step": 14489
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.885519301394758e-06,
+      "loss": 0.9883,
+      "step": 14490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.882929248423063e-06,
+      "loss": 0.8867,
+      "step": 14491
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.880339465229982e-06,
+      "loss": 0.7021,
+      "step": 14492
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.877749951878352e-06,
+      "loss": 1.0264,
+      "step": 14493
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.875160708431028e-06,
+      "loss": 0.8438,
+      "step": 14494
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.872571734950837e-06,
+      "loss": 0.7432,
+      "step": 14495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.869983031500606e-06,
+      "loss": 0.9248,
+      "step": 14496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.867394598143149e-06,
+      "loss": 0.8013,
+      "step": 14497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.86480643494129e-06,
+      "loss": 0.9727,
+      "step": 14498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.862218541957831e-06,
+      "loss": 0.8691,
+      "step": 14499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.859630919255568e-06,
+      "loss": 0.9365,
+      "step": 14500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.8570435668973e-06,
+      "loss": 0.9834,
+      "step": 14501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.85445648494582e-06,
+      "loss": 0.8525,
+      "step": 14502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.851869673463902e-06,
+      "loss": 1.0908,
+      "step": 14503
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.849283132514324e-06,
+      "loss": 0.8926,
+      "step": 14504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.846696862159843e-06,
+      "loss": 0.793,
+      "step": 14505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.844110862463238e-06,
+      "loss": 0.7793,
+      "step": 14506
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.84152513348725e-06,
+      "loss": 0.7993,
+      "step": 14507
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.838939675294633e-06,
+      "loss": 0.7319,
+      "step": 14508
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.83635448794812e-06,
+      "loss": 0.7842,
+      "step": 14509
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.83376957151046e-06,
+      "loss": 0.8364,
+      "step": 14510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.831184926044374e-06,
+      "loss": 0.9053,
+      "step": 14511
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.828600551612582e-06,
+      "loss": 0.7969,
+      "step": 14512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.826016448277795e-06,
+      "loss": 0.8613,
+      "step": 14513
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.823432616102731e-06,
+      "loss": 0.9062,
+      "step": 14514
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.820849055150094e-06,
+      "loss": 0.8311,
+      "step": 14515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.81826576548257e-06,
+      "loss": 0.9248,
+      "step": 14516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.815682747162843e-06,
+      "loss": 0.7979,
+      "step": 14517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.813100000253614e-06,
+      "loss": 0.8584,
+      "step": 14518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.810517524817548e-06,
+      "loss": 0.917,
+      "step": 14519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.807935320917314e-06,
+      "loss": 1.0332,
+      "step": 14520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.805353388615569e-06,
+      "loss": 0.8594,
+      "step": 14521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.802771727974981e-06,
+      "loss": 0.7451,
+      "step": 14522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.800190339058195e-06,
+      "loss": 0.8223,
+      "step": 14523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.79760922192785e-06,
+      "loss": 0.8203,
+      "step": 14524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.795028376646578e-06,
+      "loss": 0.8223,
+      "step": 14525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.79244780327702e-06,
+      "loss": 0.8809,
+      "step": 14526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.789867501881795e-06,
+      "loss": 0.9023,
+      "step": 14527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.787287472523517e-06,
+      "loss": 0.873,
+      "step": 14528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.78470771526479e-06,
+      "loss": 0.917,
+      "step": 14529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.782128230168232e-06,
+      "loss": 0.7014,
+      "step": 14530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.779549017296432e-06,
+      "loss": 0.873,
+      "step": 14531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.776970076711976e-06,
+      "loss": 0.8062,
+      "step": 14532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.774391408477445e-06,
+      "loss": 0.873,
+      "step": 14533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.771813012655428e-06,
+      "loss": 0.9316,
+      "step": 14534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.769234889308489e-06,
+      "loss": 0.6328,
+      "step": 14535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.766657038499192e-06,
+      "loss": 0.8857,
+      "step": 14536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.764079460290083e-06,
+      "loss": 0.917,
+      "step": 14537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.76150215474373e-06,
+      "loss": 0.916,
+      "step": 14538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.758925121922668e-06,
+      "loss": 0.7969,
+      "step": 14539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.756348361889429e-06,
+      "loss": 0.6494,
+      "step": 14540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.753771874706554e-06,
+      "loss": 0.8438,
+      "step": 14541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.751195660436563e-06,
+      "loss": 0.8896,
+      "step": 14542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.748619719141972e-06,
+      "loss": 1.0469,
+      "step": 14543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.74604405088528e-06,
+      "loss": 0.9482,
+      "step": 14544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.743468655729016e-06,
+      "loss": 0.9067,
+      "step": 14545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.740893533735657e-06,
+      "loss": 0.8525,
+      "step": 14546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.738318684967704e-06,
+      "loss": 0.8496,
+      "step": 14547
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.73574410948763e-06,
+      "loss": 0.8926,
+      "step": 14548
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.733169807357924e-06,
+      "loss": 0.9473,
+      "step": 14549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.730595778641056e-06,
+      "loss": 0.9492,
+      "step": 14550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.728022023399482e-06,
+      "loss": 0.8809,
+      "step": 14551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.72544854169566e-06,
+      "loss": 0.9395,
+      "step": 14552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.722875333592052e-06,
+      "loss": 0.7588,
+      "step": 14553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.720302399151092e-06,
+      "loss": 1.0332,
+      "step": 14554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.717729738435223e-06,
+      "loss": 0.8379,
+      "step": 14555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.715157351506864e-06,
+      "loss": 0.7388,
+      "step": 14556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.712585238428457e-06,
+      "loss": 0.793,
+      "step": 14557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.710013399262412e-06,
+      "loss": 0.7778,
+      "step": 14558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.707441834071137e-06,
+      "loss": 0.8247,
+      "step": 14559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.704870542917033e-06,
+      "loss": 0.8027,
+      "step": 14560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.70229952586251e-06,
+      "loss": 0.8848,
+      "step": 14561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.699728782969952e-06,
+      "loss": 1.0098,
+      "step": 14562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.697158314301742e-06,
+      "loss": 0.8262,
+      "step": 14563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.694588119920253e-06,
+      "loss": 0.8486,
+      "step": 14564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.692018199887871e-06,
+      "loss": 0.835,
+      "step": 14565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.68944855426695e-06,
+      "loss": 0.7397,
+      "step": 14566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.68687918311985e-06,
+      "loss": 0.7974,
+      "step": 14567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.684310086508913e-06,
+      "loss": 0.8115,
+      "step": 14568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.681741264496501e-06,
+      "loss": 0.7451,
+      "step": 14569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.679172717144944e-06,
+      "loss": 0.7109,
+      "step": 14570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.67660444451657e-06,
+      "loss": 0.875,
+      "step": 14571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.674036446673698e-06,
+      "loss": 0.7939,
+      "step": 14572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.67146872367866e-06,
+      "loss": 0.5972,
+      "step": 14573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.668901275593759e-06,
+      "loss": 0.7461,
+      "step": 14574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.666334102481303e-06,
+      "loss": 0.7812,
+      "step": 14575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.66376720440358e-06,
+      "loss": 0.7476,
+      "step": 14576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.661200581422893e-06,
+      "loss": 0.9023,
+      "step": 14577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.658634233601524e-06,
+      "loss": 0.8555,
+      "step": 14578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.656068161001747e-06,
+      "loss": 0.875,
+      "step": 14579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.653502363685827e-06,
+      "loss": 0.8066,
+      "step": 14580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.650936841716044e-06,
+      "loss": 0.8145,
+      "step": 14581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.648371595154648e-06,
+      "loss": 0.8594,
+      "step": 14582
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.64580662406388e-06,
+      "loss": 0.8877,
+      "step": 14583
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.643241928505995e-06,
+      "loss": 0.7773,
+      "step": 14584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.640677508543236e-06,
+      "loss": 0.8145,
+      "step": 14585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.638113364237828e-06,
+      "loss": 0.9883,
+      "step": 14586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.635549495651996e-06,
+      "loss": 0.8545,
+      "step": 14587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.632985902847945e-06,
+      "loss": 0.8672,
+      "step": 14588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.630422585887907e-06,
+      "loss": 1.0127,
+      "step": 14589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.627859544834076e-06,
+      "loss": 0.9512,
+      "step": 14590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.62529677974865e-06,
+      "loss": 0.9043,
+      "step": 14591
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.62273429069381e-06,
+      "loss": 0.7793,
+      "step": 14592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.620172077731759e-06,
+      "loss": 0.8418,
+      "step": 14593
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.617610140924664e-06,
+      "loss": 0.7041,
+      "step": 14594
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.615048480334696e-06,
+      "loss": 0.8398,
+      "step": 14595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.612487096024014e-06,
+      "loss": 0.7285,
+      "step": 14596
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.609925988054785e-06,
+      "loss": 0.8896,
+      "step": 14597
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.607365156489158e-06,
+      "loss": 0.874,
+      "step": 14598
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.604804601389271e-06,
+      "loss": 0.8174,
+      "step": 14599
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.602244322817259e-06,
+      "loss": 0.7327,
+      "step": 14600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.599684320835267e-06,
+      "loss": 0.8135,
+      "step": 14601
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.597124595505406e-06,
+      "loss": 0.9404,
+      "step": 14602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.594565146889797e-06,
+      "loss": 0.7651,
+      "step": 14603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.592005975050544e-06,
+      "loss": 0.7686,
+      "step": 14604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.589447080049761e-06,
+      "loss": 0.8604,
+      "step": 14605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.586888461949543e-06,
+      "loss": 0.9824,
+      "step": 14606
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.584330120811976e-06,
+      "loss": 1.1094,
+      "step": 14607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.581772056699138e-06,
+      "loss": 0.8652,
+      "step": 14608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.57921426967312e-06,
+      "loss": 0.8604,
+      "step": 14609
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.576656759795982e-06,
+      "loss": 0.8408,
+      "step": 14610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.574099527129791e-06,
+      "loss": 0.8115,
+      "step": 14611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.571542571736595e-06,
+      "loss": 0.7803,
+      "step": 14612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.568985893678458e-06,
+      "loss": 0.9238,
+      "step": 14613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.566429493017417e-06,
+      "loss": 0.9189,
+      "step": 14614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.563873369815505e-06,
+      "loss": 0.9482,
+      "step": 14615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.561317524134749e-06,
+      "loss": 0.7939,
+      "step": 14616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.558761956037182e-06,
+      "loss": 0.8975,
+      "step": 14617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.556206665584816e-06,
+      "loss": 0.7695,
+      "step": 14618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.553651652839658e-06,
+      "loss": 0.8896,
+      "step": 14619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.551096917863707e-06,
+      "loss": 0.8896,
+      "step": 14620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.54854246071897e-06,
+      "loss": 0.752,
+      "step": 14621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.545988281467429e-06,
+      "loss": 0.9385,
+      "step": 14622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.543434380171067e-06,
+      "loss": 0.8594,
+      "step": 14623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.540880756891854e-06,
+      "loss": 0.959,
+      "step": 14624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.538327411691771e-06,
+      "loss": 0.8594,
+      "step": 14625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.535774344632777e-06,
+      "loss": 0.9424,
+      "step": 14626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.533221555776813e-06,
+      "loss": 0.8643,
+      "step": 14627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.530669045185851e-06,
+      "loss": 0.9482,
+      "step": 14628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.528116812921818e-06,
+      "loss": 0.7949,
+      "step": 14629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.525564859046656e-06,
+      "loss": 0.8604,
+      "step": 14630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.523013183622279e-06,
+      "loss": 1.0801,
+      "step": 14631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.520461786710628e-06,
+      "loss": 0.9053,
+      "step": 14632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.517910668373608e-06,
+      "loss": 0.6787,
+      "step": 14633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.515359828673131e-06,
+      "loss": 0.7842,
+      "step": 14634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.512809267671087e-06,
+      "loss": 0.8555,
+      "step": 14635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.510258985429385e-06,
+      "loss": 0.9639,
+      "step": 14636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.50770898200991e-06,
+      "loss": 0.8994,
+      "step": 14637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.50515925747454e-06,
+      "loss": 0.8857,
+      "step": 14638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.50260981188514e-06,
+      "loss": 0.8662,
+      "step": 14639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.500060645303598e-06,
+      "loss": 0.7939,
+      "step": 14640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.497511757791763e-06,
+      "loss": 0.7871,
+      "step": 14641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.494963149411489e-06,
+      "loss": 0.8799,
+      "step": 14642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.492414820224617e-06,
+      "loss": 0.7422,
+      "step": 14643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.489866770293002e-06,
+      "loss": 0.9404,
+      "step": 14644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.487318999678473e-06,
+      "loss": 0.8809,
+      "step": 14645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.484771508442851e-06,
+      "loss": 0.9248,
+      "step": 14646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.482224296647954e-06,
+      "loss": 0.8623,
+      "step": 14647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.479677364355608e-06,
+      "loss": 0.8105,
+      "step": 14648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.47713071162761e-06,
+      "loss": 0.9629,
+      "step": 14649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.474584338525765e-06,
+      "loss": 0.8906,
+      "step": 14650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.472038245111855e-06,
+      "loss": 0.8604,
+      "step": 14651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.46949243144768e-06,
+      "loss": 0.8359,
+      "step": 14652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.466946897595016e-06,
+      "loss": 1.0039,
+      "step": 14653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.464401643615631e-06,
+      "loss": 0.8301,
+      "step": 14654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.461856669571289e-06,
+      "loss": 0.8857,
+      "step": 14655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.459311975523758e-06,
+      "loss": 0.6431,
+      "step": 14656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.456767561534785e-06,
+      "loss": 0.8193,
+      "step": 14657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.454223427666118e-06,
+      "loss": 0.9258,
+      "step": 14658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.451679573979483e-06,
+      "loss": 0.71,
+      "step": 14659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.449136000536633e-06,
+      "loss": 0.7725,
+      "step": 14660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.44659270739928e-06,
+      "loss": 0.8994,
+      "step": 14661
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.444049694629146e-06,
+      "loss": 0.7876,
+      "step": 14662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.441506962287933e-06,
+      "loss": 0.8486,
+      "step": 14663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.438964510437361e-06,
+      "loss": 0.8916,
+      "step": 14664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.436422339139124e-06,
+      "loss": 0.7861,
+      "step": 14665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.433880448454898e-06,
+      "loss": 1.0273,
+      "step": 14666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.431338838446378e-06,
+      "loss": 0.9775,
+      "step": 14667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.428797509175253e-06,
+      "loss": 0.9111,
+      "step": 14668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.426256460703182e-06,
+      "loss": 0.9072,
+      "step": 14669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.423715693091827e-06,
+      "loss": 0.6875,
+      "step": 14670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.421175206402842e-06,
+      "loss": 0.8672,
+      "step": 14671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.418635000697893e-06,
+      "loss": 0.9766,
+      "step": 14672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.416095076038609e-06,
+      "loss": 0.8711,
+      "step": 14673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.413555432486631e-06,
+      "loss": 0.8271,
+      "step": 14674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.41101607010358e-06,
+      "loss": 0.9131,
+      "step": 14675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.408476988951097e-06,
+      "loss": 1.0469,
+      "step": 14676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.405938189090786e-06,
+      "loss": 0.9395,
+      "step": 14677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.40339967058426e-06,
+      "loss": 0.917,
+      "step": 14678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.40086143349311e-06,
+      "loss": 0.9189,
+      "step": 14679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.398323477878948e-06,
+      "loss": 1.0312,
+      "step": 14680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.39578580380336e-06,
+      "loss": 0.8926,
+      "step": 14681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.393248411327921e-06,
+      "loss": 0.9062,
+      "step": 14682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.390711300514202e-06,
+      "loss": 0.7422,
+      "step": 14683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.388174471423784e-06,
+      "loss": 1.0352,
+      "step": 14684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.385637924118224e-06,
+      "loss": 0.8965,
+      "step": 14685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.383101658659075e-06,
+      "loss": 0.6924,
+      "step": 14686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.380565675107876e-06,
+      "loss": 0.8838,
+      "step": 14687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.378029973526186e-06,
+      "loss": 0.8076,
+      "step": 14688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.375494553975528e-06,
+      "loss": 0.7686,
+      "step": 14689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.372959416517432e-06,
+      "loss": 0.6934,
+      "step": 14690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.370424561213408e-06,
+      "loss": 0.6685,
+      "step": 14691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.367889988124987e-06,
+      "loss": 0.874,
+      "step": 14692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.365355697313667e-06,
+      "loss": 0.7793,
+      "step": 14693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.362821688840947e-06,
+      "loss": 0.7397,
+      "step": 14694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.360287962768313e-06,
+      "loss": 0.9414,
+      "step": 14695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.357754519157268e-06,
+      "loss": 0.9434,
+      "step": 14696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.355221358069284e-06,
+      "loss": 0.7686,
+      "step": 14697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.352688479565829e-06,
+      "loss": 0.8008,
+      "step": 14698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.350155883708365e-06,
+      "loss": 0.6963,
+      "step": 14699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.347623570558363e-06,
+      "loss": 0.875,
+      "step": 14700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.345091540177269e-06,
+      "loss": 0.9121,
+      "step": 14701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.342559792626528e-06,
+      "loss": 0.7471,
+      "step": 14702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.340028327967569e-06,
+      "loss": 0.8965,
+      "step": 14703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.337497146261839e-06,
+      "loss": 0.9043,
+      "step": 14704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.334966247570755e-06,
+      "loss": 0.9229,
+      "step": 14705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.332435631955735e-06,
+      "loss": 0.9336,
+      "step": 14706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.329905299478182e-06,
+      "loss": 0.8574,
+      "step": 14707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.327375250199513e-06,
+      "loss": 0.7939,
+      "step": 14708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.32484548418112e-06,
+      "loss": 0.8193,
+      "step": 14709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.322316001484393e-06,
+      "loss": 0.8809,
+      "step": 14710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.319786802170706e-06,
+      "loss": 0.8809,
+      "step": 14711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.317257886301449e-06,
+      "loss": 1.0312,
+      "step": 14712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.314729253937985e-06,
+      "loss": 0.6494,
+      "step": 14713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.312200905141672e-06,
+      "loss": 0.6875,
+      "step": 14714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.309672839973877e-06,
+      "loss": 0.6924,
+      "step": 14715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.30714505849594e-06,
+      "loss": 1.0186,
+      "step": 14716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.304617560769206e-06,
+      "loss": 0.7695,
+      "step": 14717
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.302090346855004e-06,
+      "loss": 0.8008,
+      "step": 14718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.29956341681467e-06,
+      "loss": 0.7705,
+      "step": 14719
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.297036770709526e-06,
+      "loss": 0.998,
+      "step": 14720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.29451040860088e-06,
+      "loss": 0.8203,
+      "step": 14721
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.291984330550036e-06,
+      "loss": 0.9248,
+      "step": 14722
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.289458536618307e-06,
+      "loss": 0.9326,
+      "step": 14723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.28693302686698e-06,
+      "loss": 0.5752,
+      "step": 14724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.28440780135734e-06,
+      "loss": 0.8647,
+      "step": 14725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.281882860150662e-06,
+      "loss": 0.8662,
+      "step": 14726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.279358203308232e-06,
+      "loss": 0.6221,
+      "step": 14727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.276833830891312e-06,
+      "loss": 0.9189,
+      "step": 14728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.274309742961156e-06,
+      "loss": 0.8643,
+      "step": 14729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.271785939579011e-06,
+      "loss": 0.8486,
+      "step": 14730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.269262420806136e-06,
+      "loss": 0.8164,
+      "step": 14731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.266739186703768e-06,
+      "loss": 0.8125,
+      "step": 14732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.26421623733313e-06,
+      "loss": 0.8496,
+      "step": 14733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.261693572755442e-06,
+      "loss": 0.8047,
+      "step": 14734
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.259171193031939e-06,
+      "loss": 0.584,
+      "step": 14735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.256649098223822e-06,
+      "loss": 0.7764,
+      "step": 14736
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.254127288392296e-06,
+      "loss": 0.8027,
+      "step": 14737
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.25160576359855e-06,
+      "loss": 0.6758,
+      "step": 14738
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.249084523903788e-06,
+      "loss": 0.9795,
+      "step": 14739
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.246563569369186e-06,
+      "loss": 0.9023,
+      "step": 14740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.244042900055921e-06,
+      "loss": 0.9043,
+      "step": 14741
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.241522516025156e-06,
+      "loss": 0.8379,
+      "step": 14742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.239002417338065e-06,
+      "loss": 0.8545,
+      "step": 14743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.236482604055797e-06,
+      "loss": 0.8525,
+      "step": 14744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.233963076239503e-06,
+      "loss": 0.9561,
+      "step": 14745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.231443833950311e-06,
+      "loss": 1.0195,
+      "step": 14746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.228924877249378e-06,
+      "loss": 0.9297,
+      "step": 14747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.22640620619782e-06,
+      "loss": 0.9297,
+      "step": 14748
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.22388782085676e-06,
+      "loss": 0.8818,
+      "step": 14749
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.2213697212873e-06,
+      "loss": 0.8467,
+      "step": 14750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.218851907550557e-06,
+      "loss": 0.7803,
+      "step": 14751
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.216334379707643e-06,
+      "loss": 0.7988,
+      "step": 14752
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.213817137819635e-06,
+      "loss": 0.751,
+      "step": 14753
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.211300181947617e-06,
+      "loss": 0.8779,
+      "step": 14754
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.20878351215268e-06,
+      "loss": 0.8887,
+      "step": 14755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.206267128495894e-06,
+      "loss": 0.8076,
+      "step": 14756
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.203751031038316e-06,
+      "loss": 0.9629,
+      "step": 14757
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.201235219841006e-06,
+      "loss": 0.9189,
+      "step": 14758
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.198719694965023e-06,
+      "loss": 0.8496,
+      "step": 14759
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.196204456471409e-06,
+      "loss": 0.6367,
+      "step": 14760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.193689504421196e-06,
+      "loss": 0.8799,
+      "step": 14761
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.191174838875413e-06,
+      "loss": 0.8311,
+      "step": 14762
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.188660459895093e-06,
+      "loss": 0.8428,
+      "step": 14763
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.18614636754125e-06,
+      "loss": 0.7861,
+      "step": 14764
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.18363256187489e-06,
+      "loss": 0.8203,
+      "step": 14765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.18111904295701e-06,
+      "loss": 0.8916,
+      "step": 14766
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.178605810848622e-06,
+      "loss": 0.7617,
+      "step": 14767
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.176092865610703e-06,
+      "loss": 0.7568,
+      "step": 14768
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.173580207304237e-06,
+      "loss": 0.9717,
+      "step": 14769
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.171067835990194e-06,
+      "loss": 0.7612,
+      "step": 14770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.168555751729552e-06,
+      "loss": 0.7085,
+      "step": 14771
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.166043954583267e-06,
+      "loss": 0.8721,
+      "step": 14772
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.163532444612295e-06,
+      "loss": 0.8154,
+      "step": 14773
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.161021221877571e-06,
+      "loss": 0.7539,
+      "step": 14774
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.158510286440053e-06,
+      "loss": 0.9297,
+      "step": 14775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.155999638360668e-06,
+      "loss": 0.874,
+      "step": 14776
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.153489277700337e-06,
+      "loss": 0.917,
+      "step": 14777
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.150979204519976e-06,
+      "loss": 0.7822,
+      "step": 14778
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.148469418880508e-06,
+      "loss": 0.8916,
+      "step": 14779
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.145959920842837e-06,
+      "loss": 0.9648,
+      "step": 14780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.143450710467855e-06,
+      "loss": 0.7129,
+      "step": 14781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.140941787816448e-06,
+      "loss": 0.8682,
+      "step": 14782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.138433152949517e-06,
+      "loss": 0.833,
+      "step": 14783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.135924805927926e-06,
+      "loss": 0.9131,
+      "step": 14784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.13341674681255e-06,
+      "loss": 0.8828,
+      "step": 14785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.130908975664247e-06,
+      "loss": 0.7832,
+      "step": 14786
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.12840149254388e-06,
+      "loss": 0.7817,
+      "step": 14787
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.1258942975123e-06,
+      "loss": 0.71,
+      "step": 14788
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.123387390630342e-06,
+      "loss": 0.7705,
+      "step": 14789
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.120880771958839e-06,
+      "loss": 0.6367,
+      "step": 14790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.11837444155863e-06,
+      "loss": 0.9443,
+      "step": 14791
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.115868399490532e-06,
+      "loss": 0.9736,
+      "step": 14792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.113362645815356e-06,
+      "loss": 0.9316,
+      "step": 14793
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.110857180593905e-06,
+      "loss": 0.8486,
+      "step": 14794
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.108352003886994e-06,
+      "loss": 0.8828,
+      "step": 14795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.105847115755407e-06,
+      "loss": 0.9307,
+      "step": 14796
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.103342516259931e-06,
+      "loss": 0.9668,
+      "step": 14797
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.100838205461339e-06,
+      "loss": 0.9277,
+      "step": 14798
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.098334183420416e-06,
+      "loss": 0.8857,
+      "step": 14799
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.095830450197918e-06,
+      "loss": 0.7969,
+      "step": 14800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.093327005854603e-06,
+      "loss": 0.832,
+      "step": 14801
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.09082385045123e-06,
+      "loss": 0.8125,
+      "step": 14802
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.088320984048541e-06,
+      "loss": 0.8848,
+      "step": 14803
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.085818406707268e-06,
+      "loss": 0.8789,
+      "step": 14804
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.083316118488139e-06,
+      "loss": 0.9678,
+      "step": 14805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.080814119451887e-06,
+      "loss": 0.8301,
+      "step": 14806
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.078312409659226e-06,
+      "loss": 0.7832,
+      "step": 14807
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.07581098917086e-06,
+      "loss": 0.7969,
+      "step": 14808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.073309858047487e-06,
+      "loss": 0.9355,
+      "step": 14809
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.070809016349817e-06,
+      "loss": 0.9717,
+      "step": 14810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.068308464138528e-06,
+      "loss": 0.8643,
+      "step": 14811
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.065808201474303e-06,
+      "loss": 0.7793,
+      "step": 14812
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.06330822841781e-06,
+      "loss": 0.6221,
+      "step": 14813
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.060808545029727e-06,
+      "loss": 0.8262,
+      "step": 14814
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.058309151370708e-06,
+      "loss": 0.8242,
+      "step": 14815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.055810047501412e-06,
+      "loss": 1.0283,
+      "step": 14816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.053311233482467e-06,
+      "loss": 0.7705,
+      "step": 14817
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.050812709374534e-06,
+      "loss": 0.9287,
+      "step": 14818
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.048314475238235e-06,
+      "loss": 0.7578,
+      "step": 14819
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.045816531134194e-06,
+      "loss": 0.79,
+      "step": 14820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.043318877123023e-06,
+      "loss": 0.9385,
+      "step": 14821
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.040821513265347e-06,
+      "loss": 0.7373,
+      "step": 14822
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.038324439621761e-06,
+      "loss": 0.8721,
+      "step": 14823
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.035827656252864e-06,
+      "loss": 0.8516,
+      "step": 14824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.033331163219239e-06,
+      "loss": 0.876,
+      "step": 14825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.030834960581477e-06,
+      "loss": 0.8027,
+      "step": 14826
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.028339048400153e-06,
+      "loss": 0.8311,
+      "step": 14827
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.025843426735834e-06,
+      "loss": 0.8975,
+      "step": 14828
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.02334809564907e-06,
+      "loss": 1.0068,
+      "step": 14829
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.020853055200435e-06,
+      "loss": 0.8945,
+      "step": 14830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.018358305450467e-06,
+      "loss": 0.9492,
+      "step": 14831
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.015863846459705e-06,
+      "loss": 0.9062,
+      "step": 14832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.013369678288679e-06,
+      "loss": 0.9346,
+      "step": 14833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.010875800997918e-06,
+      "loss": 0.9971,
+      "step": 14834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.008382214647948e-06,
+      "loss": 0.771,
+      "step": 14835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.005888919299275e-06,
+      "loss": 0.9229,
+      "step": 14836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.0033959150124e-06,
+      "loss": 0.9971,
+      "step": 14837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.00090320184783e-06,
+      "loss": 0.8428,
+      "step": 14838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.998410779866055e-06,
+      "loss": 0.9404,
+      "step": 14839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.995918649127551e-06,
+      "loss": 0.834,
+      "step": 14840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.993426809692794e-06,
+      "loss": 0.8359,
+      "step": 14841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.990935261622266e-06,
+      "loss": 0.8896,
+      "step": 14842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.988444004976418e-06,
+      "loss": 1.0635,
+      "step": 14843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.985953039815713e-06,
+      "loss": 0.7988,
+      "step": 14844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.983462366200587e-06,
+      "loss": 0.8525,
+      "step": 14845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.9809719841915e-06,
+      "loss": 0.8252,
+      "step": 14846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.978481893848873e-06,
+      "loss": 0.7686,
+      "step": 14847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.975992095233137e-06,
+      "loss": 1.0,
+      "step": 14848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.973502588404704e-06,
+      "loss": 0.9141,
+      "step": 14849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.971013373424002e-06,
+      "loss": 0.8926,
+      "step": 14850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.968524450351428e-06,
+      "loss": 0.9551,
+      "step": 14851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.966035819247387e-06,
+      "loss": 0.9307,
+      "step": 14852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.963547480172256e-06,
+      "loss": 0.8838,
+      "step": 14853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.961059433186436e-06,
+      "loss": 0.8262,
+      "step": 14854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.958571678350298e-06,
+      "loss": 0.8125,
+      "step": 14855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.956084215724214e-06,
+      "loss": 0.9893,
+      "step": 14856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.95359704536854e-06,
+      "loss": 0.8994,
+      "step": 14857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.951110167343642e-06,
+      "loss": 0.8545,
+      "step": 14858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.948623581709868e-06,
+      "loss": 0.8359,
+      "step": 14859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.94613728852756e-06,
+      "loss": 0.8535,
+      "step": 14860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.943651287857041e-06,
+      "loss": 0.9668,
+      "step": 14861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.941165579758659e-06,
+      "loss": 0.7998,
+      "step": 14862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.93868016429272e-06,
+      "loss": 0.8994,
+      "step": 14863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.936195041519546e-06,
+      "loss": 0.6919,
+      "step": 14864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.933710211499432e-06,
+      "loss": 0.7861,
+      "step": 14865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.931225674292692e-06,
+      "loss": 1.0869,
+      "step": 14866
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.928741429959612e-06,
+      "loss": 0.959,
+      "step": 14867
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.926257478560477e-06,
+      "loss": 0.79,
+      "step": 14868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.923773820155558e-06,
+      "loss": 0.6479,
+      "step": 14869
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.921290454805141e-06,
+      "loss": 0.7646,
+      "step": 14870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.918807382569482e-06,
+      "loss": 0.7998,
+      "step": 14871
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.91632460350884e-06,
+      "loss": 0.7588,
+      "step": 14872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.913842117683455e-06,
+      "loss": 0.8701,
+      "step": 14873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.911359925153586e-06,
+      "loss": 0.9229,
+      "step": 14874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.908878025979459e-06,
+      "loss": 0.8711,
+      "step": 14875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.906396420221302e-06,
+      "loss": 0.7246,
+      "step": 14876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.903915107939333e-06,
+      "loss": 0.9502,
+      "step": 14877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.901434089193777e-06,
+      "loss": 1.084,
+      "step": 14878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.898953364044835e-06,
+      "loss": 0.8662,
+      "step": 14879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.896472932552709e-06,
+      "loss": 0.7178,
+      "step": 14880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.893992794777581e-06,
+      "loss": 0.8984,
+      "step": 14881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.891512950779652e-06,
+      "loss": 0.8467,
+      "step": 14882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.889033400619093e-06,
+      "loss": 0.873,
+      "step": 14883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.88655414435607e-06,
+      "loss": 0.8516,
+      "step": 14884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.884075182050763e-06,
+      "loss": 0.7754,
+      "step": 14885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.881596513763316e-06,
+      "loss": 0.9102,
+      "step": 14886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.879118139553885e-06,
+      "loss": 0.6699,
+      "step": 14887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.876640059482604e-06,
+      "loss": 0.8428,
+      "step": 14888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.874162273609622e-06,
+      "loss": 1.0088,
+      "step": 14889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.87168478199506e-06,
+      "loss": 0.729,
+      "step": 14890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.86920758469904e-06,
+      "loss": 0.7275,
+      "step": 14891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.866730681781668e-06,
+      "loss": 0.9238,
+      "step": 14892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.864254073303071e-06,
+      "loss": 0.959,
+      "step": 14893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.861777759323335e-06,
+      "loss": 0.8779,
+      "step": 14894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.859301739902557e-06,
+      "loss": 0.8965,
+      "step": 14895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.856826015100812e-06,
+      "loss": 0.8003,
+      "step": 14896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.854350584978196e-06,
+      "loss": 0.8555,
+      "step": 14897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.851875449594772e-06,
+      "loss": 1.0215,
+      "step": 14898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.849400609010604e-06,
+      "loss": 0.9824,
+      "step": 14899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.846926063285745e-06,
+      "loss": 0.874,
+      "step": 14900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.844451812480254e-06,
+      "loss": 0.9043,
+      "step": 14901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.84197785665417e-06,
+      "loss": 0.7891,
+      "step": 14902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.839504195867529e-06,
+      "loss": 0.8506,
+      "step": 14903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.83703083018035e-06,
+      "loss": 0.8105,
+      "step": 14904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.83455775965267e-06,
+      "loss": 0.7588,
+      "step": 14905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.832084984344496e-06,
+      "loss": 0.8584,
+      "step": 14906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.829612504315836e-06,
+      "loss": 0.8965,
+      "step": 14907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.827140319626681e-06,
+      "loss": 0.9668,
+      "step": 14908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.824668430337038e-06,
+      "loss": 0.7881,
+      "step": 14909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.822196836506886e-06,
+      "loss": 0.8555,
+      "step": 14910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.819725538196204e-06,
+      "loss": 0.7212,
+      "step": 14911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.817254535464954e-06,
+      "loss": 0.9863,
+      "step": 14912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.814783828373113e-06,
+      "loss": 0.8877,
+      "step": 14913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.812313416980637e-06,
+      "loss": 0.915,
+      "step": 14914
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.80984330134747e-06,
+      "loss": 0.8691,
+      "step": 14915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.807373481533548e-06,
+      "loss": 0.8271,
+      "step": 14916
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.804903957598818e-06,
+      "loss": 0.748,
+      "step": 14917
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.80243472960321e-06,
+      "loss": 0.876,
+      "step": 14918
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.799965797606638e-06,
+      "loss": 0.6724,
+      "step": 14919
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.797497161669016e-06,
+      "loss": 0.8838,
+      "step": 14920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.795028821850248e-06,
+      "loss": 0.8169,
+      "step": 14921
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.792560778210242e-06,
+      "loss": 0.9316,
+      "step": 14922
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.790093030808885e-06,
+      "loss": 0.8467,
+      "step": 14923
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.787625579706056e-06,
+      "loss": 0.8643,
+      "step": 14924
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.785158424961645e-06,
+      "loss": 0.8594,
+      "step": 14925
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.782691566635519e-06,
+      "loss": 0.8613,
+      "step": 14926
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.780225004787536e-06,
+      "loss": 0.8203,
+      "step": 14927
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.777758739477548e-06,
+      "loss": 0.876,
+      "step": 14928
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.77529277076542e-06,
+      "loss": 0.916,
+      "step": 14929
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.772827098710982e-06,
+      "loss": 0.9365,
+      "step": 14930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.770361723374073e-06,
+      "loss": 0.9443,
+      "step": 14931
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.767896644814512e-06,
+      "loss": 0.9043,
+      "step": 14932
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.765431863092135e-06,
+      "loss": 1.0176,
+      "step": 14933
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.762967378266743e-06,
+      "loss": 0.9678,
+      "step": 14934
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.760503190398147e-06,
+      "loss": 0.874,
+      "step": 14935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.758039299546137e-06,
+      "loss": 0.7969,
+      "step": 14936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.755575705770516e-06,
+      "loss": 0.7744,
+      "step": 14937
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.753112409131062e-06,
+      "loss": 0.8848,
+      "step": 14938
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.750649409687557e-06,
+      "loss": 0.7578,
+      "step": 14939
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.748186707499756e-06,
+      "loss": 0.7012,
+      "step": 14940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.745724302627442e-06,
+      "loss": 0.8506,
+      "step": 14941
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.743262195130359e-06,
+      "loss": 0.8682,
+      "step": 14942
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.740800385068256e-06,
+      "loss": 0.8662,
+      "step": 14943
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.73833887250087e-06,
+      "loss": 0.8848,
+      "step": 14944
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.735877657487947e-06,
+      "loss": 0.7568,
+      "step": 14945
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.733416740089203e-06,
+      "loss": 0.9805,
+      "step": 14946
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.730956120364362e-06,
+      "loss": 0.9697,
+      "step": 14947
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.728495798373126e-06,
+      "loss": 0.7832,
+      "step": 14948
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.726035774175213e-06,
+      "loss": 0.7627,
+      "step": 14949
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.723576047830319e-06,
+      "loss": 0.9502,
+      "step": 14950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.721116619398128e-06,
+      "loss": 0.9463,
+      "step": 14951
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.718657488938319e-06,
+      "loss": 0.769,
+      "step": 14952
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.716198656510582e-06,
+      "loss": 0.7588,
+      "step": 14953
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.713740122174579e-06,
+      "loss": 0.8779,
+      "step": 14954
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.71128188598997e-06,
+      "loss": 0.8384,
+      "step": 14955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.708823948016401e-06,
+      "loss": 0.9219,
+      "step": 14956
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.706366308313538e-06,
+      "loss": 0.9121,
+      "step": 14957
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.70390896694101e-06,
+      "loss": 0.9902,
+      "step": 14958
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.701451923958447e-06,
+      "loss": 0.7158,
+      "step": 14959
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.698995179425473e-06,
+      "loss": 0.8965,
+      "step": 14960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.696538733401714e-06,
+      "loss": 0.875,
+      "step": 14961
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.69408258594678e-06,
+      "loss": 0.8379,
+      "step": 14962
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.69162673712027e-06,
+      "loss": 1.0293,
+      "step": 14963
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.689171186981774e-06,
+      "loss": 0.9053,
+      "step": 14964
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.686715935590896e-06,
+      "loss": 0.8418,
+      "step": 14965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.68426098300721e-06,
+      "loss": 0.834,
+      "step": 14966
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.68180632929029e-06,
+      "loss": 0.8223,
+      "step": 14967
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.6793519744997e-06,
+      "loss": 0.876,
+      "step": 14968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.676897918695007e-06,
+      "loss": 0.7822,
+      "step": 14969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.674444161935763e-06,
+      "loss": 0.9131,
+      "step": 14970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.671990704281502e-06,
+      "loss": 0.9316,
+      "step": 14971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.669537545791781e-06,
+      "loss": 0.9395,
+      "step": 14972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.667084686526119e-06,
+      "loss": 0.8369,
+      "step": 14973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.664632126544043e-06,
+      "loss": 0.8584,
+      "step": 14974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.66217986590506e-06,
+      "loss": 0.8311,
+      "step": 14975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.659727904668695e-06,
+      "loss": 0.9414,
+      "step": 14976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.657276242894443e-06,
+      "loss": 0.7271,
+      "step": 14977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.654824880641796e-06,
+      "loss": 0.7402,
+      "step": 14978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.652373817970238e-06,
+      "loss": 0.8457,
+      "step": 14979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.64992305493926e-06,
+      "loss": 0.9414,
+      "step": 14980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.647472591608332e-06,
+      "loss": 0.9385,
+      "step": 14981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.645022428036915e-06,
+      "loss": 0.666,
+      "step": 14982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.642572564284461e-06,
+      "loss": 0.9502,
+      "step": 14983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.640123000410438e-06,
+      "loss": 0.7856,
+      "step": 14984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.637673736474278e-06,
+      "loss": 0.8271,
+      "step": 14985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.63522477253542e-06,
+      "loss": 0.8525,
+      "step": 14986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.63277610865329e-06,
+      "loss": 0.7549,
+      "step": 14987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.630327744887317e-06,
+      "loss": 0.7871,
+      "step": 14988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.627879681296914e-06,
+      "loss": 0.5752,
+      "step": 14989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.625431917941483e-06,
+      "loss": 0.8301,
+      "step": 14990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.622984454880424e-06,
+      "loss": 0.7656,
+      "step": 14991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.620537292173138e-06,
+      "loss": 0.8975,
+      "step": 14992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.618090429879004e-06,
+      "loss": 0.8965,
+      "step": 14993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.615643868057403e-06,
+      "loss": 0.8145,
+      "step": 14994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.613197606767695e-06,
+      "loss": 0.9951,
+      "step": 14995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.610751646069263e-06,
+      "loss": 0.9092,
+      "step": 14996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.608305986021451e-06,
+      "loss": 0.7212,
+      "step": 14997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.60586062668361e-06,
+      "loss": 0.9004,
+      "step": 14998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.603415568115076e-06,
+      "loss": 0.834,
+      "step": 14999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.600970810375188e-06,
+      "loss": 0.9053,
+      "step": 15000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.598526353523281e-06,
+      "loss": 0.958,
+      "step": 15001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.596082197618668e-06,
+      "loss": 0.9307,
+      "step": 15002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5936383427206615e-06,
+      "loss": 0.79,
+      "step": 15003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.591194788888561e-06,
+      "loss": 0.7676,
+      "step": 15004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5887515361816755e-06,
+      "loss": 1.0,
+      "step": 15005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.586308584659292e-06,
+      "loss": 0.7593,
+      "step": 15006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5838659343806895e-06,
+      "loss": 0.918,
+      "step": 15007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.581423585405141e-06,
+      "loss": 0.9521,
+      "step": 15008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.578981537791925e-06,
+      "loss": 0.7354,
+      "step": 15009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.576539791600302e-06,
+      "loss": 0.8516,
+      "step": 15010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.574098346889511e-06,
+      "loss": 0.9297,
+      "step": 15011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.571657203718818e-06,
+      "loss": 0.873,
+      "step": 15012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.569216362147456e-06,
+      "loss": 0.8789,
+      "step": 15013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.566775822234655e-06,
+      "loss": 0.6938,
+      "step": 15014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.564335584039633e-06,
+      "loss": 0.8691,
+      "step": 15015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.56189564762162e-06,
+      "loss": 0.688,
+      "step": 15016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5594560130398205e-06,
+      "loss": 0.5952,
+      "step": 15017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.557016680353437e-06,
+      "loss": 0.6978,
+      "step": 15018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.55457764962166e-06,
+      "loss": 0.9727,
+      "step": 15019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.552138920903689e-06,
+      "loss": 0.7158,
+      "step": 15020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5497004942586985e-06,
+      "loss": 0.8496,
+      "step": 15021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.54726236974586e-06,
+      "loss": 0.7891,
+      "step": 15022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5448245474243385e-06,
+      "loss": 0.8965,
+      "step": 15023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5423870273533e-06,
+      "loss": 0.9434,
+      "step": 15024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.539949809591893e-06,
+      "loss": 0.9414,
+      "step": 15025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.537512894199259e-06,
+      "loss": 0.7803,
+      "step": 15026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.53507628123453e-06,
+      "loss": 0.7559,
+      "step": 15027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.53263997075685e-06,
+      "loss": 0.8613,
+      "step": 15028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.530203962825331e-06,
+      "loss": 0.7822,
+      "step": 15029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.527768257499091e-06,
+      "loss": 0.9287,
+      "step": 15030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.525332854837228e-06,
+      "loss": 0.8506,
+      "step": 15031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.522897754898857e-06,
+      "loss": 1.0098,
+      "step": 15032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.520462957743062e-06,
+      "loss": 0.9082,
+      "step": 15033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.518028463428933e-06,
+      "loss": 0.8438,
+      "step": 15034
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.515594272015536e-06,
+      "loss": 0.9395,
+      "step": 15035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.513160383561959e-06,
+      "loss": 0.8916,
+      "step": 15036
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.510726798127255e-06,
+      "loss": 0.9551,
+      "step": 15037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.508293515770484e-06,
+      "loss": 0.8955,
+      "step": 15038
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5058605365506845e-06,
+      "loss": 0.8203,
+      "step": 15039
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.503427860526913e-06,
+      "loss": 0.9355,
+      "step": 15040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.500995487758196e-06,
+      "loss": 0.8574,
+      "step": 15041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.498563418303562e-06,
+      "loss": 0.8545,
+      "step": 15042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.496131652222021e-06,
+      "loss": 0.8193,
+      "step": 15043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.493700189572597e-06,
+      "loss": 0.7588,
+      "step": 15044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.491269030414292e-06,
+      "loss": 0.9248,
+      "step": 15045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.488838174806099e-06,
+      "loss": 0.9912,
+      "step": 15046
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.486407622807004e-06,
+      "loss": 0.9346,
+      "step": 15047
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.483977374476001e-06,
+      "loss": 0.832,
+      "step": 15048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.481547429872058e-06,
+      "loss": 0.9111,
+      "step": 15049
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.479117789054144e-06,
+      "loss": 0.9551,
+      "step": 15050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.4766884520812135e-06,
+      "loss": 0.8672,
+      "step": 15051
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.474259419012229e-06,
+      "loss": 0.793,
+      "step": 15052
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.471830689906135e-06,
+      "loss": 0.8311,
+      "step": 15053
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.469402264821861e-06,
+      "loss": 0.8867,
+      "step": 15054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.4669741438183394e-06,
+      "loss": 0.7031,
+      "step": 15055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.464546326954503e-06,
+      "loss": 1.0068,
+      "step": 15056
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.462118814289263e-06,
+      "loss": 0.8945,
+      "step": 15057
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.459691605881518e-06,
+      "loss": 0.8916,
+      "step": 15058
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.4572647017901855e-06,
+      "loss": 0.9736,
+      "step": 15059
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.454838102074153e-06,
+      "loss": 0.8027,
+      "step": 15060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.452411806792306e-06,
+      "loss": 0.9941,
+      "step": 15061
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.449985816003515e-06,
+      "loss": 0.8232,
+      "step": 15062
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.447560129766667e-06,
+      "loss": 0.8506,
+      "step": 15063
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.445134748140619e-06,
+      "loss": 1.0195,
+      "step": 15064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.44270967118423e-06,
+      "loss": 0.8545,
+      "step": 15065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.440284898956338e-06,
+      "loss": 0.8281,
+      "step": 15066
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.437860431515805e-06,
+      "loss": 0.918,
+      "step": 15067
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.435436268921452e-06,
+      "loss": 0.9609,
+      "step": 15068
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.43301241123211e-06,
+      "loss": 0.915,
+      "step": 15069
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.430588858506593e-06,
+      "loss": 0.6523,
+      "step": 15070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.428165610803726e-06,
+      "loss": 0.8594,
+      "step": 15071
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.425742668182308e-06,
+      "loss": 0.8721,
+      "step": 15072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.423320030701133e-06,
+      "loss": 0.7969,
+      "step": 15073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.420897698418989e-06,
+      "loss": 1.0117,
+      "step": 15074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.418475671394671e-06,
+      "loss": 0.7422,
+      "step": 15075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.416053949686948e-06,
+      "loss": 1.0088,
+      "step": 15076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.413632533354589e-06,
+      "loss": 0.8584,
+      "step": 15077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.411211422456344e-06,
+      "loss": 0.7812,
+      "step": 15078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.408790617050982e-06,
+      "loss": 0.8418,
+      "step": 15079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.406370117197243e-06,
+      "loss": 1.0566,
+      "step": 15080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.403949922953866e-06,
+      "loss": 0.9902,
+      "step": 15081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.401530034379571e-06,
+      "loss": 0.8271,
+      "step": 15082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.399110451533094e-06,
+      "loss": 0.8203,
+      "step": 15083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.396691174473154e-06,
+      "loss": 0.8848,
+      "step": 15084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.394272203258455e-06,
+      "loss": 0.8115,
+      "step": 15085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3918535379476975e-06,
+      "loss": 0.8428,
+      "step": 15086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.389435178599569e-06,
+      "loss": 0.8818,
+      "step": 15087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3870171252727666e-06,
+      "loss": 0.7393,
+      "step": 15088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.384599378025967e-06,
+      "loss": 0.7783,
+      "step": 15089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.382181936917841e-06,
+      "loss": 0.8457,
+      "step": 15090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.379764802007043e-06,
+      "loss": 0.8203,
+      "step": 15091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.377347973352247e-06,
+      "loss": 0.9053,
+      "step": 15092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.374931451012093e-06,
+      "loss": 0.814,
+      "step": 15093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3725152350452235e-06,
+      "loss": 0.7334,
+      "step": 15094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3700993255102645e-06,
+      "loss": 0.7803,
+      "step": 15095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.367683722465861e-06,
+      "loss": 0.9121,
+      "step": 15096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3652684259706216e-06,
+      "loss": 0.9189,
+      "step": 15097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.362853436083155e-06,
+      "loss": 0.8369,
+      "step": 15098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.360438752862078e-06,
+      "loss": 0.7305,
+      "step": 15099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.35802437636598e-06,
+      "loss": 0.8394,
+      "step": 15100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3556103066534514e-06,
+      "loss": 0.8096,
+      "step": 15101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.353196543783068e-06,
+      "loss": 0.8633,
+      "step": 15102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.350783087813418e-06,
+      "loss": 0.9736,
+      "step": 15103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.348369938803064e-06,
+      "loss": 0.8516,
+      "step": 15104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.345957096810563e-06,
+      "loss": 0.8291,
+      "step": 15105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.343544561894462e-06,
+      "loss": 0.8555,
+      "step": 15106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.341132334113321e-06,
+      "loss": 0.9287,
+      "step": 15107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.33872041352567e-06,
+      "loss": 0.8809,
+      "step": 15108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.33630880019004e-06,
+      "loss": 0.751,
+      "step": 15109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.333897494164943e-06,
+      "loss": 0.7319,
+      "step": 15110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3314864955089125e-06,
+      "loss": 0.8242,
+      "step": 15111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.32907580428045e-06,
+      "loss": 0.8496,
+      "step": 15112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.326665420538051e-06,
+      "loss": 0.8428,
+      "step": 15113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.324255344340205e-06,
+      "loss": 0.7783,
+      "step": 15114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3218455757454125e-06,
+      "loss": 0.958,
+      "step": 15115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3194361148121395e-06,
+      "loss": 0.8682,
+      "step": 15116
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.317026961598864e-06,
+      "loss": 0.7041,
+      "step": 15117
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.314618116164034e-06,
+      "loss": 0.7539,
+      "step": 15118
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.312209578566125e-06,
+      "loss": 0.8037,
+      "step": 15119
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3098013488635765e-06,
+      "loss": 0.8848,
+      "step": 15120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.307393427114826e-06,
+      "loss": 0.7695,
+      "step": 15121
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.304985813378305e-06,
+      "loss": 0.8242,
+      "step": 15122
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.30257850771245e-06,
+      "loss": 0.8535,
+      "step": 15123
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.300171510175673e-06,
+      "loss": 0.8398,
+      "step": 15124
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.297764820826385e-06,
+      "loss": 0.7134,
+      "step": 15125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.295358439722982e-06,
+      "loss": 0.9229,
+      "step": 15126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.292952366923871e-06,
+      "loss": 0.834,
+      "step": 15127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.290546602487438e-06,
+      "loss": 0.833,
+      "step": 15128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.288141146472063e-06,
+      "loss": 0.9609,
+      "step": 15129
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.28573599893611e-06,
+      "loss": 1.0244,
+      "step": 15130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.283331159937959e-06,
+      "loss": 0.7012,
+      "step": 15131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.280926629535963e-06,
+      "loss": 0.9404,
+      "step": 15132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.278522407788473e-06,
+      "loss": 0.958,
+      "step": 15133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.276118494753823e-06,
+      "loss": 0.835,
+      "step": 15134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.273714890490367e-06,
+      "loss": 0.9033,
+      "step": 15135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.271311595056421e-06,
+      "loss": 0.8477,
+      "step": 15136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.26890860851031e-06,
+      "loss": 0.8457,
+      "step": 15137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.266505930910342e-06,
+      "loss": 0.8008,
+      "step": 15138
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.264103562314831e-06,
+      "loss": 0.75,
+      "step": 15139
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.261701502782072e-06,
+      "loss": 0.8701,
+      "step": 15140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.25929975237035e-06,
+      "loss": 0.8408,
+      "step": 15141
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.256898311137959e-06,
+      "loss": 0.7627,
+      "step": 15142
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.254497179143172e-06,
+      "loss": 0.7861,
+      "step": 15143
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.252096356444254e-06,
+      "loss": 0.7666,
+      "step": 15144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.24969584309946e-06,
+      "loss": 0.9189,
+      "step": 15145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.247295639167058e-06,
+      "loss": 0.7891,
+      "step": 15146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.244895744705285e-06,
+      "loss": 0.7637,
+      "step": 15147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.242496159772383e-06,
+      "loss": 0.8174,
+      "step": 15148
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.240096884426573e-06,
+      "loss": 0.8447,
+      "step": 15149
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.237697918726092e-06,
+      "loss": 0.8945,
+      "step": 15150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.2352992627291514e-06,
+      "loss": 0.9131,
+      "step": 15151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.232900916493959e-06,
+      "loss": 0.7773,
+      "step": 15152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.230502880078707e-06,
+      "loss": 0.957,
+      "step": 15153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.228105153541603e-06,
+      "loss": 0.8594,
+      "step": 15154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.2257077369408234e-06,
+      "loss": 0.7871,
+      "step": 15155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.223310630334552e-06,
+      "loss": 0.8115,
+      "step": 15156
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.220913833780949e-06,
+      "loss": 0.9434,
+      "step": 15157
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.218517347338194e-06,
+      "loss": 0.9336,
+      "step": 15158
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.216121171064432e-06,
+      "loss": 0.7617,
+      "step": 15159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.213725305017814e-06,
+      "loss": 0.8047,
+      "step": 15160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.211329749256472e-06,
+      "loss": 0.9033,
+      "step": 15161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.208934503838556e-06,
+      "loss": 0.7002,
+      "step": 15162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.206539568822179e-06,
+      "loss": 0.8877,
+      "step": 15163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.204144944265466e-06,
+      "loss": 0.9072,
+      "step": 15164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.201750630226514e-06,
+      "loss": 0.7852,
+      "step": 15165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1993566267634385e-06,
+      "loss": 0.7646,
+      "step": 15166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1969629339343375e-06,
+      "loss": 0.873,
+      "step": 15167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.194569551797295e-06,
+      "loss": 0.8184,
+      "step": 15168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.192176480410389e-06,
+      "loss": 0.79,
+      "step": 15169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1897837198316865e-06,
+      "loss": 0.9092,
+      "step": 15170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.187391270119266e-06,
+      "loss": 1.0723,
+      "step": 15171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.184999131331178e-06,
+      "loss": 0.7783,
+      "step": 15172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.182607303525478e-06,
+      "loss": 0.833,
+      "step": 15173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1802157867601915e-06,
+      "loss": 0.9131,
+      "step": 15174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.177824581093378e-06,
+      "loss": 0.9473,
+      "step": 15175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.175433686583051e-06,
+      "loss": 0.8262,
+      "step": 15176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.173043103287231e-06,
+      "loss": 0.8457,
+      "step": 15177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.170652831263929e-06,
+      "loss": 0.9941,
+      "step": 15178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.168262870571158e-06,
+      "loss": 0.7275,
+      "step": 15179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.16587322126691e-06,
+      "loss": 0.708,
+      "step": 15180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.163483883409168e-06,
+      "loss": 0.8887,
+      "step": 15181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.16109485705593e-06,
+      "loss": 0.7046,
+      "step": 15182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.158706142265161e-06,
+      "loss": 0.7559,
+      "step": 15183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.156317739094829e-06,
+      "loss": 0.6699,
+      "step": 15184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.153929647602887e-06,
+      "loss": 0.8379,
+      "step": 15185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1515418678472985e-06,
+      "loss": 0.8857,
+      "step": 15186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1491543998860045e-06,
+      "loss": 0.7959,
+      "step": 15187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.14676724377694e-06,
+      "loss": 0.6797,
+      "step": 15188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.144380399578028e-06,
+      "loss": 0.8301,
+      "step": 15189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.141993867347203e-06,
+      "loss": 0.9824,
+      "step": 15190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.139607647142375e-06,
+      "loss": 0.8203,
+      "step": 15191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1372217390214446e-06,
+      "loss": 0.958,
+      "step": 15192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1348361430423095e-06,
+      "loss": 1.0078,
+      "step": 15193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.132450859262874e-06,
+      "loss": 0.918,
+      "step": 15194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.130065887741013e-06,
+      "loss": 0.7236,
+      "step": 15195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1276812285346045e-06,
+      "loss": 1.0547,
+      "step": 15196
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.125296881701509e-06,
+      "loss": 0.9082,
+      "step": 15197
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.122912847299601e-06,
+      "loss": 0.8613,
+      "step": 15198
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1205291253867285e-06,
+      "loss": 1.0449,
+      "step": 15199
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.118145716020737e-06,
+      "loss": 0.8848,
+      "step": 15200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.115762619259459e-06,
+      "loss": 0.834,
+      "step": 15201
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1133798351607365e-06,
+      "loss": 0.9033,
+      "step": 15202
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.110997363782388e-06,
+      "loss": 1.0869,
+      "step": 15203
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.108615205182227e-06,
+      "loss": 0.8506,
+      "step": 15204
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.106233359418058e-06,
+      "loss": 0.8193,
+      "step": 15205
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.103851826547692e-06,
+      "loss": 0.7705,
+      "step": 15206
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.101470606628918e-06,
+      "loss": 0.9678,
+      "step": 15207
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.099089699719517e-06,
+      "loss": 1.0723,
+      "step": 15208
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0967091058772615e-06,
+      "loss": 0.7471,
+      "step": 15209
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.094328825159935e-06,
+      "loss": 0.8174,
+      "step": 15210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0919488576252944e-06,
+      "loss": 1.0215,
+      "step": 15211
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.089569203331095e-06,
+      "loss": 0.835,
+      "step": 15212
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.087189862335076e-06,
+      "loss": 0.7891,
+      "step": 15213
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.084810834694991e-06,
+      "loss": 0.9512,
+      "step": 15214
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.082432120468563e-06,
+      "loss": 0.7295,
+      "step": 15215
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.08005371971352e-06,
+      "loss": 1.0176,
+      "step": 15216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.077675632487568e-06,
+      "loss": 0.6934,
+      "step": 15217
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.075297858848433e-06,
+      "loss": 0.7207,
+      "step": 15218
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.072920398853811e-06,
+      "loss": 0.9658,
+      "step": 15219
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.070543252561391e-06,
+      "loss": 1.0801,
+      "step": 15220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.068166420028857e-06,
+      "loss": 0.9004,
+      "step": 15221
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0657899013138976e-06,
+      "loss": 0.8232,
+      "step": 15222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0634136964741815e-06,
+      "loss": 0.8623,
+      "step": 15223
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.061037805567368e-06,
+      "loss": 0.9326,
+      "step": 15224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.05866222865111e-06,
+      "loss": 0.9277,
+      "step": 15225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.056286965783066e-06,
+      "loss": 0.8838,
+      "step": 15226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.053912017020869e-06,
+      "loss": 0.6758,
+      "step": 15227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0515373824221515e-06,
+      "loss": 0.7344,
+      "step": 15228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0491630620445466e-06,
+      "loss": 0.9395,
+      "step": 15229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.046789055945666e-06,
+      "loss": 0.9521,
+      "step": 15230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0444153641831235e-06,
+      "loss": 0.9414,
+      "step": 15231
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.042041986814512e-06,
+      "loss": 0.6934,
+      "step": 15232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.039668923897438e-06,
+      "loss": 0.6206,
+      "step": 15233
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.037296175489487e-06,
+      "loss": 0.9473,
+      "step": 15234
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.034923741648234e-06,
+      "loss": 0.8701,
+      "step": 15235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.032551622431248e-06,
+      "loss": 0.8936,
+      "step": 15236
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.030179817896103e-06,
+      "loss": 0.8818,
+      "step": 15237
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.027808328100352e-06,
+      "loss": 0.7119,
+      "step": 15238
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0254371531015416e-06,
+      "loss": 0.7036,
+      "step": 15239
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.023066292957211e-06,
+      "loss": 0.8496,
+      "step": 15240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.020695747724901e-06,
+      "loss": 0.8247,
+      "step": 15241
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0183255174621364e-06,
+      "loss": 0.7227,
+      "step": 15242
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.015955602226432e-06,
+      "loss": 0.9062,
+      "step": 15243
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.013586002075297e-06,
+      "loss": 0.8867,
+      "step": 15244
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.011216717066241e-06,
+      "loss": 0.8506,
+      "step": 15245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.00884774725676e-06,
+      "loss": 0.8721,
+      "step": 15246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.006479092704335e-06,
+      "loss": 0.9326,
+      "step": 15247
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0041107534664466e-06,
+      "loss": 0.9678,
+      "step": 15248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.001742729600576e-06,
+      "loss": 0.9502,
+      "step": 15249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.999375021164174e-06,
+      "loss": 0.8535,
+      "step": 15250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.997007628214716e-06,
+      "loss": 0.8428,
+      "step": 15251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.994640550809641e-06,
+      "loss": 0.7397,
+      "step": 15252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.992273789006387e-06,
+      "loss": 0.8232,
+      "step": 15253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.989907342862401e-06,
+      "loss": 0.7373,
+      "step": 15254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.987541212435103e-06,
+      "loss": 0.9375,
+      "step": 15255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.985175397781909e-06,
+      "loss": 0.7705,
+      "step": 15256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.9828098989602275e-06,
+      "loss": 0.835,
+      "step": 15257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.9804447160274745e-06,
+      "loss": 0.9688,
+      "step": 15258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.978079849041037e-06,
+      "loss": 0.876,
+      "step": 15259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.975715298058307e-06,
+      "loss": 0.8184,
+      "step": 15260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.973351063136658e-06,
+      "loss": 0.6826,
+      "step": 15261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.970987144333474e-06,
+      "loss": 0.9766,
+      "step": 15262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.9686235417061166e-06,
+      "loss": 0.9502,
+      "step": 15263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.966260255311941e-06,
+      "loss": 0.7197,
+      "step": 15264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.963897285208292e-06,
+      "loss": 0.7988,
+      "step": 15265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.961534631452522e-06,
+      "loss": 0.9199,
+      "step": 15266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.959172294101966e-06,
+      "loss": 0.834,
+      "step": 15267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.956810273213936e-06,
+      "loss": 0.9014,
+      "step": 15268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.954448568845773e-06,
+      "loss": 0.8203,
+      "step": 15269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.952087181054774e-06,
+      "loss": 0.8994,
+      "step": 15270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.949726109898247e-06,
+      "loss": 0.7861,
+      "step": 15271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.947365355433484e-06,
+      "loss": 0.9561,
+      "step": 15272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.945004917717781e-06,
+      "loss": 0.8076,
+      "step": 15273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.942644796808417e-06,
+      "loss": 0.8496,
+      "step": 15274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.940284992762663e-06,
+      "loss": 0.8516,
+      "step": 15275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.937925505637777e-06,
+      "loss": 0.8467,
+      "step": 15276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.935566335491033e-06,
+      "loss": 0.7881,
+      "step": 15277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.933207482379674e-06,
+      "loss": 0.7627,
+      "step": 15278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.9308489463609395e-06,
+      "loss": 0.8545,
+      "step": 15279
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.928490727492059e-06,
+      "loss": 0.8828,
+      "step": 15280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.926132825830274e-06,
+      "loss": 0.6055,
+      "step": 15281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.923775241432797e-06,
+      "loss": 0.9248,
+      "step": 15282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.9214179743568386e-06,
+      "loss": 0.9287,
+      "step": 15283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.919061024659597e-06,
+      "loss": 0.8643,
+      "step": 15284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.916704392398282e-06,
+      "loss": 0.9346,
+      "step": 15285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.914348077630073e-06,
+      "loss": 0.9268,
+      "step": 15286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.911992080412153e-06,
+      "loss": 0.6943,
+      "step": 15287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.909636400801689e-06,
+      "loss": 0.9531,
+      "step": 15288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.907281038855857e-06,
+      "loss": 0.9658,
+      "step": 15289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.904925994631813e-06,
+      "loss": 0.7852,
+      "step": 15290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.9025712681867015e-06,
+      "loss": 0.8906,
+      "step": 15291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.900216859577662e-06,
+      "loss": 0.7588,
+      "step": 15292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.897862768861841e-06,
+      "loss": 0.7651,
+      "step": 15293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.895508996096356e-06,
+      "loss": 0.7461,
+      "step": 15294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.89315554133833e-06,
+      "loss": 0.8594,
+      "step": 15295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.890802404644867e-06,
+      "loss": 0.832,
+      "step": 15296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.888449586073085e-06,
+      "loss": 0.7549,
+      "step": 15297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.88609708568007e-06,
+      "loss": 0.7393,
+      "step": 15298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.883744903522913e-06,
+      "loss": 0.9102,
+      "step": 15299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.8813930396586856e-06,
+      "loss": 1.0449,
+      "step": 15300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.879041494144476e-06,
+      "loss": 0.8271,
+      "step": 15301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.876690267037342e-06,
+      "loss": 0.9746,
+      "step": 15302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.8743393583943415e-06,
+      "loss": 0.8994,
+      "step": 15303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.871988768272517e-06,
+      "loss": 0.8262,
+      "step": 15304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.869638496728923e-06,
+      "loss": 0.8066,
+      "step": 15305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.867288543820587e-06,
+      "loss": 0.8975,
+      "step": 15306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.864938909604537e-06,
+      "loss": 0.9326,
+      "step": 15307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.8625895941377855e-06,
+      "loss": 0.9902,
+      "step": 15308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.8602405974773545e-06,
+      "loss": 0.874,
+      "step": 15309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.8578919196802395e-06,
+      "loss": 0.8789,
+      "step": 15310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.8555435608034415e-06,
+      "loss": 0.9219,
+      "step": 15311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.853195520903937e-06,
+      "loss": 0.7393,
+      "step": 15312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.850847800038718e-06,
+      "loss": 0.9033,
+      "step": 15313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.848500398264754e-06,
+      "loss": 0.7021,
+      "step": 15314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.846153315639003e-06,
+      "loss": 0.8926,
+      "step": 15315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.8438065522184325e-06,
+      "loss": 1.0,
+      "step": 15316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.841460108059987e-06,
+      "loss": 0.7822,
+      "step": 15317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.839113983220606e-06,
+      "loss": 0.8818,
+      "step": 15318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.836768177757218e-06,
+      "loss": 0.8994,
+      "step": 15319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.834422691726761e-06,
+      "loss": 0.7168,
+      "step": 15320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.832077525186145e-06,
+      "loss": 0.7676,
+      "step": 15321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.829732678192284e-06,
+      "loss": 0.7822,
+      "step": 15322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.82738815080207e-06,
+      "loss": 0.7363,
+      "step": 15323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.825043943072416e-06,
+      "loss": 0.9395,
+      "step": 15324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.822700055060196e-06,
+      "loss": 0.6689,
+      "step": 15325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.820356486822295e-06,
+      "loss": 0.8701,
+      "step": 15326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.818013238415573e-06,
+      "loss": 0.8179,
+      "step": 15327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.815670309896911e-06,
+      "loss": 0.9805,
+      "step": 15328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.813327701323158e-06,
+      "loss": 0.8564,
+      "step": 15329
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.810985412751159e-06,
+      "loss": 0.8501,
+      "step": 15330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.808643444237753e-06,
+      "loss": 0.7559,
+      "step": 15331
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.80630179583978e-06,
+      "loss": 0.9131,
+      "step": 15332
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.803960467614055e-06,
+      "loss": 0.9102,
+      "step": 15333
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.801619459617408e-06,
+      "loss": 0.9434,
+      "step": 15334
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.799278771906641e-06,
+      "loss": 0.7715,
+      "step": 15335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.796938404538551e-06,
+      "loss": 0.7817,
+      "step": 15336
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.794598357569944e-06,
+      "loss": 0.8491,
+      "step": 15337
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.792258631057598e-06,
+      "loss": 0.6758,
+      "step": 15338
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.7899192250582926e-06,
+      "loss": 1.0684,
+      "step": 15339
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.787580139628793e-06,
+      "loss": 0.9688,
+      "step": 15340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.785241374825873e-06,
+      "loss": 0.7578,
+      "step": 15341
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.782902930706281e-06,
+      "loss": 0.8516,
+      "step": 15342
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.780564807326764e-06,
+      "loss": 0.9365,
+      "step": 15343
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.778227004744056e-06,
+      "loss": 0.8037,
+      "step": 15344
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.775889523014903e-06,
+      "loss": 0.9609,
+      "step": 15345
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.773552362196018e-06,
+      "loss": 0.8135,
+      "step": 15346
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.771215522344121e-06,
+      "loss": 0.7852,
+      "step": 15347
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.768879003515913e-06,
+      "loss": 0.8906,
+      "step": 15348
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.766542805768106e-06,
+      "loss": 0.8076,
+      "step": 15349
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.764206929157387e-06,
+      "loss": 0.8984,
+      "step": 15350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.7618713737404404e-06,
+      "loss": 1.0078,
+      "step": 15351
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.759536139573937e-06,
+      "loss": 1.0029,
+      "step": 15352
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.757201226714561e-06,
+      "loss": 0.8857,
+      "step": 15353
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.754866635218964e-06,
+      "loss": 0.9521,
+      "step": 15354
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.752532365143794e-06,
+      "loss": 0.9746,
+      "step": 15355
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.750198416545712e-06,
+      "loss": 0.8945,
+      "step": 15356
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.747864789481349e-06,
+      "loss": 0.9443,
+      "step": 15357
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.745531484007333e-06,
+      "loss": 0.9297,
+      "step": 15358
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.743198500180281e-06,
+      "loss": 0.8369,
+      "step": 15359
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.740865838056821e-06,
+      "loss": 0.8193,
+      "step": 15360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.738533497693554e-06,
+      "loss": 0.8008,
+      "step": 15361
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.736201479147078e-06,
+      "loss": 0.9561,
+      "step": 15362
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.733869782473978e-06,
+      "loss": 0.8486,
+      "step": 15363
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.731538407730851e-06,
+      "loss": 0.7207,
+      "step": 15364
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.7292073549742675e-06,
+      "loss": 0.9404,
+      "step": 15365
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.726876624260792e-06,
+      "loss": 0.8125,
+      "step": 15366
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.724546215646978e-06,
+      "loss": 0.8223,
+      "step": 15367
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.722216129189394e-06,
+      "loss": 0.7949,
+      "step": 15368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.719886364944575e-06,
+      "loss": 0.9756,
+      "step": 15369
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.71755692296906e-06,
+      "loss": 0.8906,
+      "step": 15370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.7152278033193685e-06,
+      "loss": 0.9551,
+      "step": 15371
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.712899006052036e-06,
+      "loss": 0.9111,
+      "step": 15372
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.710570531223568e-06,
+      "loss": 0.9434,
+      "step": 15373
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.70824237889047e-06,
+      "loss": 0.7891,
+      "step": 15374
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.705914549109236e-06,
+      "loss": 0.8389,
+      "step": 15375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.703587041936364e-06,
+      "loss": 0.8994,
+      "step": 15376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.70125985742833e-06,
+      "loss": 0.8936,
+      "step": 15377
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.698932995641611e-06,
+      "loss": 0.8213,
+      "step": 15378
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.696606456632664e-06,
+      "loss": 0.8281,
+      "step": 15379
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.69428024045796e-06,
+      "loss": 0.7646,
+      "step": 15380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.691954347173946e-06,
+      "loss": 0.7856,
+      "step": 15381
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6896287768370605e-06,
+      "loss": 0.959,
+      "step": 15382
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.687303529503735e-06,
+      "loss": 0.8486,
+      "step": 15383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.684978605230405e-06,
+      "loss": 0.9053,
+      "step": 15384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.68265400407349e-06,
+      "loss": 0.9541,
+      "step": 15385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6803297260893966e-06,
+      "loss": 0.9229,
+      "step": 15386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.678005771334521e-06,
+      "loss": 0.9541,
+      "step": 15387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.675682139865274e-06,
+      "loss": 0.8457,
+      "step": 15388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.673358831738037e-06,
+      "loss": 1.0283,
+      "step": 15389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.671035847009189e-06,
+      "loss": 0.8975,
+      "step": 15390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.668713185735094e-06,
+      "loss": 0.9502,
+      "step": 15391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.666390847972131e-06,
+      "loss": 0.8779,
+      "step": 15392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.664068833776649e-06,
+      "loss": 0.8994,
+      "step": 15393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.661747143204998e-06,
+      "loss": 0.8623,
+      "step": 15394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.65942577631351e-06,
+      "loss": 0.915,
+      "step": 15395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.657104733158533e-06,
+      "loss": 0.9688,
+      "step": 15396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.654784013796383e-06,
+      "loss": 0.8223,
+      "step": 15397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.652463618283372e-06,
+      "loss": 0.6973,
+      "step": 15398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.65014354667582e-06,
+      "loss": 0.7803,
+      "step": 15399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.647823799030026e-06,
+      "loss": 1.0127,
+      "step": 15400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.645504375402281e-06,
+      "loss": 0.9058,
+      "step": 15401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.643185275848863e-06,
+      "loss": 0.916,
+      "step": 15402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.640866500426064e-06,
+      "loss": 0.9404,
+      "step": 15403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.638548049190148e-06,
+      "loss": 0.8594,
+      "step": 15404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.636229922197375e-06,
+      "loss": 0.5635,
+      "step": 15405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.633912119503994e-06,
+      "loss": 0.8994,
+      "step": 15406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6315946411662655e-06,
+      "loss": 0.8154,
+      "step": 15407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.629277487240417e-06,
+      "loss": 0.7451,
+      "step": 15408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6269606577826865e-06,
+      "loss": 0.915,
+      "step": 15409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.624644152849282e-06,
+      "loss": 0.7256,
+      "step": 15410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.622327972496436e-06,
+      "loss": 0.9131,
+      "step": 15411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.620012116780347e-06,
+      "loss": 0.9277,
+      "step": 15412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.617696585757219e-06,
+      "loss": 0.874,
+      "step": 15413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.615381379483228e-06,
+      "loss": 0.9932,
+      "step": 15414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.613066498014575e-06,
+      "loss": 0.918,
+      "step": 15415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.610751941407423e-06,
+      "loss": 1.0078,
+      "step": 15416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.608437709717952e-06,
+      "loss": 0.8574,
+      "step": 15417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6061238030023135e-06,
+      "loss": 0.8301,
+      "step": 15418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6038102213166555e-06,
+      "loss": 0.7852,
+      "step": 15419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.601496964717131e-06,
+      "loss": 0.6133,
+      "step": 15420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.599184033259873e-06,
+      "loss": 0.7456,
+      "step": 15421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.596871427001008e-06,
+      "loss": 0.8965,
+      "step": 15422
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.5945591459966485e-06,
+      "loss": 0.8545,
+      "step": 15423
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.592247190302923e-06,
+      "loss": 0.9277,
+      "step": 15424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.589935559975926e-06,
+      "loss": 0.7451,
+      "step": 15425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.587624255071756e-06,
+      "loss": 0.9365,
+      "step": 15426
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.585313275646495e-06,
+      "loss": 0.8574,
+      "step": 15427
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.583002621756234e-06,
+      "loss": 0.6963,
+      "step": 15428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.580692293457043e-06,
+      "loss": 0.9443,
+      "step": 15429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.578382290804986e-06,
+      "loss": 0.7676,
+      "step": 15430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.576072613856111e-06,
+      "loss": 0.9727,
+      "step": 15431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.573763262666484e-06,
+      "loss": 0.8301,
+      "step": 15432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.5714542372921345e-06,
+      "loss": 0.7588,
+      "step": 15433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.569145537789101e-06,
+      "loss": 0.9258,
+      "step": 15434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.5668371642134e-06,
+      "loss": 0.6455,
+      "step": 15435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.564529116621061e-06,
+      "loss": 0.9912,
+      "step": 15436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.562221395068089e-06,
+      "loss": 0.9717,
+      "step": 15437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.559913999610476e-06,
+      "loss": 0.916,
+      "step": 15438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.557606930304232e-06,
+      "loss": 0.7715,
+      "step": 15439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.555300187205336e-06,
+      "loss": 0.7729,
+      "step": 15440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.5529937703697644e-06,
+      "loss": 1.0107,
+      "step": 15441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.550687679853483e-06,
+      "loss": 0.7822,
+      "step": 15442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.548381915712463e-06,
+      "loss": 0.7783,
+      "step": 15443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.546076478002654e-06,
+      "loss": 0.959,
+      "step": 15444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.543771366780003e-06,
+      "loss": 0.9639,
+      "step": 15445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.541466582100442e-06,
+      "loss": 0.8516,
+      "step": 15446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.539162124019913e-06,
+      "loss": 0.9941,
+      "step": 15447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.536857992594332e-06,
+      "loss": 1.0859,
+      "step": 15448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.534554187879614e-06,
+      "loss": 0.9258,
+      "step": 15449
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.532250709931658e-06,
+      "loss": 0.7158,
+      "step": 15450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.529947558806378e-06,
+      "loss": 0.8545,
+      "step": 15451
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.527644734559655e-06,
+      "loss": 0.7832,
+      "step": 15452
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.525342237247374e-06,
+      "loss": 0.8945,
+      "step": 15453
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.523040066925403e-06,
+      "loss": 0.6929,
+      "step": 15454
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.520738223649623e-06,
+      "loss": 0.8447,
+      "step": 15455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.518436707475884e-06,
+      "loss": 0.8594,
+      "step": 15456
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.5161355184600385e-06,
+      "loss": 0.8604,
+      "step": 15457
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.513834656657922e-06,
+      "loss": 0.9131,
+      "step": 15458
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.511534122125385e-06,
+      "loss": 0.8955,
+      "step": 15459
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.509233914918247e-06,
+      "loss": 0.9336,
+      "step": 15460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.506934035092325e-06,
+      "loss": 0.7607,
+      "step": 15461
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.504634482703427e-06,
+      "loss": 0.8799,
+      "step": 15462
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.502335257807369e-06,
+      "loss": 0.9033,
+      "step": 15463
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.500036360459938e-06,
+      "loss": 0.9131,
+      "step": 15464
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.4977377907169225e-06,
+      "loss": 0.7876,
+      "step": 15465
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.495439548634095e-06,
+      "loss": 0.7637,
+      "step": 15466
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.493141634267241e-06,
+      "loss": 0.8652,
+      "step": 15467
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.49084404767212e-06,
+      "loss": 0.8545,
+      "step": 15468
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.488546788904482e-06,
+      "loss": 0.9238,
+      "step": 15469
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.486249858020073e-06,
+      "loss": 0.7285,
+      "step": 15470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.483953255074642e-06,
+      "loss": 0.8232,
+      "step": 15471
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.481656980123916e-06,
+      "loss": 0.8535,
+      "step": 15472
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.4793610332236215e-06,
+      "loss": 0.7266,
+      "step": 15473
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.477065414429465e-06,
+      "loss": 0.8916,
+      "step": 15474
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.474770123797165e-06,
+      "loss": 0.7383,
+      "step": 15475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.472475161382421e-06,
+      "loss": 0.875,
+      "step": 15476
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.470180527240919e-06,
+      "loss": 0.877,
+      "step": 15477
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.4678862214283414e-06,
+      "loss": 0.7979,
+      "step": 15478
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.465592244000376e-06,
+      "loss": 0.8203,
+      "step": 15479
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.4632985950126815e-06,
+      "loss": 0.9688,
+      "step": 15480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.461005274520919e-06,
+      "loss": 0.8477,
+      "step": 15481
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.458712282580739e-06,
+      "loss": 0.8008,
+      "step": 15482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.456419619247793e-06,
+      "loss": 0.9258,
+      "step": 15483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.454127284577711e-06,
+      "loss": 0.8301,
+      "step": 15484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.451835278626119e-06,
+      "loss": 0.8555,
+      "step": 15485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.449543601448647e-06,
+      "loss": 0.8848,
+      "step": 15486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.4472522531009015e-06,
+      "loss": 1.0156,
+      "step": 15487
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.4449612336384845e-06,
+      "loss": 0.9102,
+      "step": 15488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.442670543116991e-06,
+      "loss": 0.8877,
+      "step": 15489
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.440380181592016e-06,
+      "loss": 0.9199,
+      "step": 15490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.4380901491191386e-06,
+      "loss": 0.9082,
+      "step": 15491
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.4358004457539306e-06,
+      "loss": 0.9004,
+      "step": 15492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.433511071551946e-06,
+      "loss": 0.8721,
+      "step": 15493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.431222026568755e-06,
+      "loss": 0.6665,
+      "step": 15494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.428933310859904e-06,
+      "loss": 0.8662,
+      "step": 15495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.4266449244809296e-06,
+      "loss": 0.9062,
+      "step": 15496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.42435686748736e-06,
+      "loss": 0.7725,
+      "step": 15497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.42206913993473e-06,
+      "loss": 0.9189,
+      "step": 15498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.41978174187855e-06,
+      "loss": 0.8809,
+      "step": 15499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.417494673374323e-06,
+      "loss": 0.7539,
+      "step": 15500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.415207934477563e-06,
+      "loss": 0.8604,
+      "step": 15501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.412921525243747e-06,
+      "loss": 0.876,
+      "step": 15502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.4106354457283744e-06,
+      "loss": 0.8174,
+      "step": 15503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.408349695986915e-06,
+      "loss": 0.9658,
+      "step": 15504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.406064276074837e-06,
+      "loss": 0.9404,
+      "step": 15505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.403779186047594e-06,
+      "loss": 0.7646,
+      "step": 15506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.401494425960652e-06,
+      "loss": 0.708,
+      "step": 15507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.399209995869448e-06,
+      "loss": 0.7773,
+      "step": 15508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.39692589582942e-06,
+      "loss": 0.8662,
+      "step": 15509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.394642125895987e-06,
+      "loss": 0.8662,
+      "step": 15510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.392358686124584e-06,
+      "loss": 0.8301,
+      "step": 15511
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.39007557657062e-06,
+      "loss": 0.9395,
+      "step": 15512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.387792797289494e-06,
+      "loss": 0.9307,
+      "step": 15513
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.385510348336599e-06,
+      "loss": 0.7598,
+      "step": 15514
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.3832282297673356e-06,
+      "loss": 0.958,
+      "step": 15515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.380946441637077e-06,
+      "loss": 0.9443,
+      "step": 15516
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.378664984001195e-06,
+      "loss": 1.0479,
+      "step": 15517
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.376383856915052e-06,
+      "loss": 1.0586,
+      "step": 15518
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.37410306043401e-06,
+      "loss": 0.8789,
+      "step": 15519
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.371822594613417e-06,
+      "loss": 0.8105,
+      "step": 15520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.369542459508608e-06,
+      "loss": 1.0381,
+      "step": 15521
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.367262655174913e-06,
+      "loss": 0.8428,
+      "step": 15522
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.364983181667668e-06,
+      "loss": 0.9697,
+      "step": 15523
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.3627040390421806e-06,
+      "loss": 0.834,
+      "step": 15524
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.360425227353755e-06,
+      "loss": 0.8506,
+      "step": 15525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.3581467466577026e-06,
+      "loss": 0.6533,
+      "step": 15526
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.35586859700931e-06,
+      "loss": 0.833,
+      "step": 15527
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.353590778463859e-06,
+      "loss": 0.9111,
+      "step": 15528
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.3513132910766214e-06,
+      "loss": 0.9277,
+      "step": 15529
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.3490361349028775e-06,
+      "loss": 0.876,
+      "step": 15530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.346759309997881e-06,
+      "loss": 0.8037,
+      "step": 15531
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.344482816416884e-06,
+      "loss": 0.9385,
+      "step": 15532
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.342206654215122e-06,
+      "loss": 0.8926,
+      "step": 15533
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.339930823447844e-06,
+      "loss": 0.9561,
+      "step": 15534
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.3376553241702736e-06,
+      "loss": 0.9453,
+      "step": 15535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.335380156437627e-06,
+      "loss": 0.8623,
+      "step": 15536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.333105320305113e-06,
+      "loss": 0.8359,
+      "step": 15537
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.330830815827946e-06,
+      "loss": 0.9053,
+      "step": 15538
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.328556643061317e-06,
+      "loss": 0.6719,
+      "step": 15539
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.32628280206041e-06,
+      "loss": 0.9648,
+      "step": 15540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.3240092928803994e-06,
+      "loss": 0.7998,
+      "step": 15541
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.3217361155764735e-06,
+      "loss": 0.8896,
+      "step": 15542
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.319463270203783e-06,
+      "loss": 0.9893,
+      "step": 15543
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.317190756817486e-06,
+      "loss": 0.9512,
+      "step": 15544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.314918575472724e-06,
+      "loss": 0.9541,
+      "step": 15545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.312646726224647e-06,
+      "loss": 0.7529,
+      "step": 15546
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.310375209128383e-06,
+      "loss": 0.8135,
+      "step": 15547
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.308104024239052e-06,
+      "loss": 0.8369,
+      "step": 15548
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.305833171611764e-06,
+      "loss": 0.9951,
+      "step": 15549
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.3035626513016354e-06,
+      "loss": 0.6787,
+      "step": 15550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.301292463363764e-06,
+      "loss": 0.8818,
+      "step": 15551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.299022607853238e-06,
+      "loss": 0.9121,
+      "step": 15552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.296753084825134e-06,
+      "loss": 0.7246,
+      "step": 15553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.294483894334538e-06,
+      "loss": 0.8359,
+      "step": 15554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.292215036436513e-06,
+      "loss": 0.7705,
+      "step": 15555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.2899465111861155e-06,
+      "loss": 0.8916,
+      "step": 15556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.287678318638389e-06,
+      "loss": 0.9834,
+      "step": 15557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.285410458848389e-06,
+      "loss": 0.7617,
+      "step": 15558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.283142931871147e-06,
+      "loss": 0.7139,
+      "step": 15559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.280875737761682e-06,
+      "loss": 0.6104,
+      "step": 15560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.278608876575014e-06,
+      "loss": 0.9189,
+      "step": 15561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.27634234836616e-06,
+      "loss": 0.8887,
+      "step": 15562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.2740761531901165e-06,
+      "loss": 0.9375,
+      "step": 15563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.2718102911018786e-06,
+      "loss": 0.8584,
+      "step": 15564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.269544762156428e-06,
+      "loss": 0.7754,
+      "step": 15565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.26727956640875e-06,
+      "loss": 0.7832,
+      "step": 15566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.2650147039138094e-06,
+      "loss": 0.8994,
+      "step": 15567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.262750174726571e-06,
+      "loss": 0.9141,
+      "step": 15568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.26048597890198e-06,
+      "loss": 0.9697,
+      "step": 15569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.258222116494994e-06,
+      "loss": 0.9678,
+      "step": 15570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.255958587560542e-06,
+      "loss": 0.998,
+      "step": 15571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.253695392153551e-06,
+      "loss": 0.7734,
+      "step": 15572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.251432530328951e-06,
+      "loss": 0.8203,
+      "step": 15573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.249170002141653e-06,
+      "loss": 0.8057,
+      "step": 15574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.246907807646558e-06,
+      "loss": 0.835,
+      "step": 15575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.244645946898558e-06,
+      "loss": 0.9551,
+      "step": 15576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.242384419952554e-06,
+      "loss": 0.7837,
+      "step": 15577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.24012322686342e-06,
+      "loss": 0.9639,
+      "step": 15578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.237862367686029e-06,
+      "loss": 0.7881,
+      "step": 15579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.235601842475239e-06,
+      "loss": 0.876,
+      "step": 15580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.233341651285918e-06,
+      "loss": 0.8906,
+      "step": 15581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.231081794172911e-06,
+      "loss": 0.8643,
+      "step": 15582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.228822271191049e-06,
+      "loss": 0.6914,
+      "step": 15583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.226563082395176e-06,
+      "loss": 0.8789,
+      "step": 15584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.224304227840106e-06,
+      "loss": 1.042,
+      "step": 15585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.222045707580664e-06,
+      "loss": 0.8633,
+      "step": 15586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.219787521671654e-06,
+      "loss": 0.7773,
+      "step": 15587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.2175296701678765e-06,
+      "loss": 0.9375,
+      "step": 15588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.215272153124112e-06,
+      "loss": 0.7334,
+      "step": 15589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.213014970595161e-06,
+      "loss": 0.9297,
+      "step": 15590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.210758122635789e-06,
+      "loss": 0.6577,
+      "step": 15591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.208501609300768e-06,
+      "loss": 0.9814,
+      "step": 15592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.2062454306448435e-06,
+      "loss": 0.8555,
+      "step": 15593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.2039895867227854e-06,
+      "loss": 0.793,
+      "step": 15594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.201734077589325e-06,
+      "loss": 0.8916,
+      "step": 15595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.199478903299201e-06,
+      "loss": 0.9834,
+      "step": 15596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.197224063907133e-06,
+      "loss": 1.0,
+      "step": 15597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.194969559467847e-06,
+      "loss": 0.8271,
+      "step": 15598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.192715390036055e-06,
+      "loss": 0.877,
+      "step": 15599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.190461555666451e-06,
+      "loss": 0.9473,
+      "step": 15600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.188208056413729e-06,
+      "loss": 0.7959,
+      "step": 15601
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.185954892332584e-06,
+      "loss": 0.8047,
+      "step": 15602
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.183702063477688e-06,
+      "loss": 0.8516,
+      "step": 15603
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.18144956990371e-06,
+      "loss": 0.8916,
+      "step": 15604
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.1791974116653055e-06,
+      "loss": 0.9229,
+      "step": 15605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.176945588817143e-06,
+      "loss": 0.8779,
+      "step": 15606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.174694101413856e-06,
+      "loss": 0.8301,
+      "step": 15607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.172442949510087e-06,
+      "loss": 1.0273,
+      "step": 15608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.170192133160453e-06,
+      "loss": 0.8799,
+      "step": 15609
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.167941652419591e-06,
+      "loss": 0.8096,
+      "step": 15610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.165691507342109e-06,
+      "loss": 1.0273,
+      "step": 15611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.1634416979826e-06,
+      "loss": 0.6753,
+      "step": 15612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.1611922243956755e-06,
+      "loss": 0.9043,
+      "step": 15613
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.15894308663592e-06,
+      "loss": 0.8828,
+      "step": 15614
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.156694284757907e-06,
+      "loss": 0.877,
+      "step": 15615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.1544458188162084e-06,
+      "loss": 0.8057,
+      "step": 15616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.152197688865396e-06,
+      "loss": 0.7891,
+      "step": 15617
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.149949894960021e-06,
+      "loss": 0.7432,
+      "step": 15618
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.1477024371546325e-06,
+      "loss": 0.7432,
+      "step": 15619
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.145455315503759e-06,
+      "loss": 0.9736,
+      "step": 15620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.143208530061948e-06,
+      "loss": 0.708,
+      "step": 15621
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.140962080883714e-06,
+      "loss": 0.9004,
+      "step": 15622
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.138715968023572e-06,
+      "loss": 0.9072,
+      "step": 15623
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.136470191536024e-06,
+      "loss": 0.7744,
+      "step": 15624
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.1342247514755795e-06,
+      "loss": 0.9629,
+      "step": 15625
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.131979647896722e-06,
+      "loss": 0.7979,
+      "step": 15626
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.129734880853932e-06,
+      "loss": 0.9707,
+      "step": 15627
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.127490450401681e-06,
+      "loss": 0.8643,
+      "step": 15628
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.1252463565944454e-06,
+      "loss": 0.8164,
+      "step": 15629
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.123002599486676e-06,
+      "loss": 0.8008,
+      "step": 15630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.120759179132825e-06,
+      "loss": 0.9756,
+      "step": 15631
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.118516095587321e-06,
+      "loss": 0.6592,
+      "step": 15632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.116273348904618e-06,
+      "loss": 0.73,
+      "step": 15633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.1140309391391285e-06,
+      "loss": 0.8701,
+      "step": 15634
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.11178886634527e-06,
+      "loss": 0.7744,
+      "step": 15635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.109547130577447e-06,
+      "loss": 0.835,
+      "step": 15636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.1073057318900695e-06,
+      "loss": 0.8975,
+      "step": 15637
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.105064670337526e-06,
+      "loss": 0.9004,
+      "step": 15638
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.102823945974199e-06,
+      "loss": 0.7432,
+      "step": 15639
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.100583558854458e-06,
+      "loss": 0.7881,
+      "step": 15640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.098343509032685e-06,
+      "loss": 0.8218,
+      "step": 15641
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.096103796563231e-06,
+      "loss": 0.8564,
+      "step": 15642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.093864421500448e-06,
+      "loss": 0.8477,
+      "step": 15643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.091625383898672e-06,
+      "loss": 0.8867,
+      "step": 15644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.089386683812253e-06,
+      "loss": 0.8813,
+      "step": 15645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.087148321295509e-06,
+      "loss": 0.834,
+      "step": 15646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.0849102964027574e-06,
+      "loss": 0.8037,
+      "step": 15647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.082672609188305e-06,
+      "loss": 1.1133,
+      "step": 15648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.080435259706466e-06,
+      "loss": 0.8926,
+      "step": 15649
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.078198248011524e-06,
+      "loss": 0.7563,
+      "step": 15650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.075961574157771e-06,
+      "loss": 0.8193,
+      "step": 15651
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.073725238199475e-06,
+      "loss": 0.8936,
+      "step": 15652
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.071489240190917e-06,
+      "loss": 0.7949,
+      "step": 15653
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.069253580186354e-06,
+      "loss": 0.8135,
+      "step": 15654
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.067018258240039e-06,
+      "loss": 0.6909,
+      "step": 15655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.064783274406205e-06,
+      "loss": 0.5029,
+      "step": 15656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.062548628739109e-06,
+      "loss": 0.7295,
+      "step": 15657
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.060314321292966e-06,
+      "loss": 0.876,
+      "step": 15658
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.058080352121995e-06,
+      "loss": 0.957,
+      "step": 15659
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.055846721280419e-06,
+      "loss": 0.8594,
+      "step": 15660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.053613428822433e-06,
+      "loss": 0.8164,
+      "step": 15661
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.051380474802235e-06,
+      "loss": 0.8984,
+      "step": 15662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.049147859274007e-06,
+      "loss": 0.6855,
+      "step": 15663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.046915582291937e-06,
+      "loss": 0.6863,
+      "step": 15664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.044683643910192e-06,
+      "loss": 0.8877,
+      "step": 15665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.042452044182927e-06,
+      "loss": 0.8262,
+      "step": 15666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.040220783164308e-06,
+      "loss": 0.9141,
+      "step": 15667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.037989860908473e-06,
+      "loss": 0.8223,
+      "step": 15668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.035759277469568e-06,
+      "loss": 0.7217,
+      "step": 15669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.033529032901717e-06,
+      "loss": 0.9111,
+      "step": 15670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.031299127259045e-06,
+      "loss": 0.6724,
+      "step": 15671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.029069560595655e-06,
+      "loss": 0.8457,
+      "step": 15672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.026840332965665e-06,
+      "loss": 0.79,
+      "step": 15673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.024611444423167e-06,
+      "loss": 0.9111,
+      "step": 15674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.0223828950222495e-06,
+      "loss": 0.8232,
+      "step": 15675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.020154684816988e-06,
+      "loss": 0.8018,
+      "step": 15676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.017926813861463e-06,
+      "loss": 0.7578,
+      "step": 15677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.015699282209735e-06,
+      "loss": 0.9277,
+      "step": 15678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.0134720899158594e-06,
+      "loss": 1.001,
+      "step": 15679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.011245237033876e-06,
+      "loss": 0.8447,
+      "step": 15680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.009018723617838e-06,
+      "loss": 0.7529,
+      "step": 15681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.006792549721772e-06,
+      "loss": 0.4897,
+      "step": 15682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.004566715399696e-06,
+      "loss": 0.8945,
+      "step": 15683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.002341220705618e-06,
+      "loss": 0.9697,
+      "step": 15684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.000116065693564e-06,
+      "loss": 0.8916,
+      "step": 15685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.997891250417518e-06,
+      "loss": 0.9473,
+      "step": 15686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9956667749314745e-06,
+      "loss": 0.8154,
+      "step": 15687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.993442639289404e-06,
+      "loss": 0.7861,
+      "step": 15688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.991218843545296e-06,
+      "loss": 0.8701,
+      "step": 15689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9889953877531115e-06,
+      "loss": 0.8809,
+      "step": 15690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.986772271966801e-06,
+      "loss": 0.79,
+      "step": 15691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.984549496240311e-06,
+      "loss": 0.7988,
+      "step": 15692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.982327060627593e-06,
+      "loss": 0.8926,
+      "step": 15693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.980104965182572e-06,
+      "loss": 0.9238,
+      "step": 15694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.977883209959165e-06,
+      "loss": 0.8086,
+      "step": 15695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.975661795011303e-06,
+      "loss": 0.8242,
+      "step": 15696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.973440720392885e-06,
+      "loss": 0.7939,
+      "step": 15697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.97121998615781e-06,
+      "loss": 0.9551,
+      "step": 15698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.968999592359963e-06,
+      "loss": 0.7715,
+      "step": 15699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.966779539053238e-06,
+      "loss": 1.0605,
+      "step": 15700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.964559826291502e-06,
+      "loss": 0.9707,
+      "step": 15701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.962340454128624e-06,
+      "loss": 0.9473,
+      "step": 15702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9601214226184525e-06,
+      "loss": 0.7881,
+      "step": 15703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.957902731814851e-06,
+      "loss": 0.8174,
+      "step": 15704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.955684381771653e-06,
+      "loss": 1.0322,
+      "step": 15705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.953466372542696e-06,
+      "loss": 0.9961,
+      "step": 15706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.951248704181791e-06,
+      "loss": 0.8066,
+      "step": 15707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.949031376742771e-06,
+      "loss": 0.8613,
+      "step": 15708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9468143902794365e-06,
+      "loss": 1.0049,
+      "step": 15709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.944597744845588e-06,
+      "loss": 0.9482,
+      "step": 15710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.942381440495011e-06,
+      "loss": 0.9639,
+      "step": 15711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.940165477281501e-06,
+      "loss": 0.9414,
+      "step": 15712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.937949855258827e-06,
+      "loss": 0.7988,
+      "step": 15713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.935734574480754e-06,
+      "loss": 0.9688,
+      "step": 15714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.933519635001035e-06,
+      "loss": 0.7939,
+      "step": 15715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.931305036873432e-06,
+      "loss": 0.8359,
+      "step": 15716
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.929090780151683e-06,
+      "loss": 0.834,
+      "step": 15717
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.926876864889519e-06,
+      "loss": 0.9014,
+      "step": 15718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.92466329114066e-06,
+      "loss": 0.8516,
+      "step": 15719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.922450058958833e-06,
+      "loss": 0.8154,
+      "step": 15720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9202371683977445e-06,
+      "loss": 0.7451,
+      "step": 15721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.918024619511093e-06,
+      "loss": 0.8643,
+      "step": 15722
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.915812412352564e-06,
+      "loss": 0.793,
+      "step": 15723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.913600546975853e-06,
+      "loss": 1.0068,
+      "step": 15724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.911389023434631e-06,
+      "loss": 0.9238,
+      "step": 15725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.909177841782563e-06,
+      "loss": 0.8042,
+      "step": 15726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.906967002073303e-06,
+      "loss": 0.79,
+      "step": 15727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.904756504360516e-06,
+      "loss": 0.9482,
+      "step": 15728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.902546348697833e-06,
+      "loss": 0.79,
+      "step": 15729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.900336535138893e-06,
+      "loss": 0.9443,
+      "step": 15730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.898127063737313e-06,
+      "loss": 0.8496,
+      "step": 15731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.895917934546724e-06,
+      "loss": 0.9121,
+      "step": 15732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.893709147620727e-06,
+      "loss": 0.8301,
+      "step": 15733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.891500703012925e-06,
+      "loss": 1.0039,
+      "step": 15734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.889292600776904e-06,
+      "loss": 0.9902,
+      "step": 15735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.887084840966257e-06,
+      "loss": 0.6821,
+      "step": 15736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.884877423634558e-06,
+      "loss": 0.998,
+      "step": 15737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.882670348835371e-06,
+      "loss": 0.9824,
+      "step": 15738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.880463616622254e-06,
+      "loss": 0.9004,
+      "step": 15739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.878257227048765e-06,
+      "loss": 0.8271,
+      "step": 15740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8760511801684426e-06,
+      "loss": 0.8525,
+      "step": 15741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.873845476034814e-06,
+      "loss": 0.7031,
+      "step": 15742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.87164011470142e-06,
+      "loss": 0.8193,
+      "step": 15743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.869435096221771e-06,
+      "loss": 0.7998,
+      "step": 15744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8672304206493746e-06,
+      "loss": 0.9541,
+      "step": 15745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.865026088037726e-06,
+      "loss": 0.9336,
+      "step": 15746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.862822098440333e-06,
+      "loss": 0.8174,
+      "step": 15747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.860618451910669e-06,
+      "loss": 0.6377,
+      "step": 15748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.858415148502214e-06,
+      "loss": 0.9102,
+      "step": 15749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.856212188268429e-06,
+      "loss": 0.876,
+      "step": 15750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.854009571262778e-06,
+      "loss": 0.7646,
+      "step": 15751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.851807297538721e-06,
+      "loss": 0.8418,
+      "step": 15752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.849605367149689e-06,
+      "loss": 0.7949,
+      "step": 15753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.847403780149123e-06,
+      "loss": 0.8145,
+      "step": 15754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.845202536590437e-06,
+      "loss": 0.9336,
+      "step": 15755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.843001636527066e-06,
+      "loss": 0.9971,
+      "step": 15756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8408010800124105e-06,
+      "loss": 0.9414,
+      "step": 15757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.838600867099873e-06,
+      "loss": 0.9688,
+      "step": 15758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8364009978428374e-06,
+      "loss": 1.0049,
+      "step": 15759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8342014722947025e-06,
+      "loss": 0.9463,
+      "step": 15760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.832002290508838e-06,
+      "loss": 1.043,
+      "step": 15761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.829803452538611e-06,
+      "loss": 0.8945,
+      "step": 15762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.827604958437376e-06,
+      "loss": 0.8359,
+      "step": 15763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.825406808258496e-06,
+      "loss": 0.9189,
+      "step": 15764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.823209002055306e-06,
+      "loss": 0.7988,
+      "step": 15765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.821011539881141e-06,
+      "loss": 1.0527,
+      "step": 15766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8188144217893184e-06,
+      "loss": 0.6719,
+      "step": 15767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8166176478331735e-06,
+      "loss": 0.752,
+      "step": 15768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.814421218066006e-06,
+      "loss": 0.8691,
+      "step": 15769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.812225132541116e-06,
+      "loss": 0.9824,
+      "step": 15770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8100293913117935e-06,
+      "loss": 0.8838,
+      "step": 15771
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.807833994431331e-06,
+      "loss": 0.832,
+      "step": 15772
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.805638941953e-06,
+      "loss": 0.9072,
+      "step": 15773
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.803444233930067e-06,
+      "loss": 0.8369,
+      "step": 15774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.801249870415786e-06,
+      "loss": 0.8916,
+      "step": 15775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7990558514634195e-06,
+      "loss": 0.8037,
+      "step": 15776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.796862177126206e-06,
+      "loss": 0.8594,
+      "step": 15777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7946688474573766e-06,
+      "loss": 0.7295,
+      "step": 15778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.792475862510149e-06,
+      "loss": 0.8799,
+      "step": 15779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.79028322233776e-06,
+      "loss": 0.8486,
+      "step": 15780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.788090926993406e-06,
+      "loss": 0.8555,
+      "step": 15781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.785898976530282e-06,
+      "loss": 0.9219,
+      "step": 15782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.783707371001595e-06,
+      "loss": 0.792,
+      "step": 15783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7815161104605215e-06,
+      "loss": 0.7832,
+      "step": 15784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.779325194960237e-06,
+      "loss": 0.8613,
+      "step": 15785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7771346245539015e-06,
+      "loss": 0.7646,
+      "step": 15786
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.774944399294689e-06,
+      "loss": 0.9766,
+      "step": 15787
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.772754519235739e-06,
+      "loss": 0.8623,
+      "step": 15788
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.770564984430198e-06,
+      "loss": 0.8506,
+      "step": 15789
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.768375794931191e-06,
+      "loss": 1.0068,
+      "step": 15790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.766186950791854e-06,
+      "loss": 0.8789,
+      "step": 15791
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.763998452065302e-06,
+      "loss": 0.8398,
+      "step": 15792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.76181029880464e-06,
+      "loss": 0.8721,
+      "step": 15793
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.759622491062964e-06,
+      "loss": 0.8506,
+      "step": 15794
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.757435028893375e-06,
+      "loss": 0.8447,
+      "step": 15795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7552479123489536e-06,
+      "loss": 0.9053,
+      "step": 15796
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.753061141482772e-06,
+      "loss": 0.7842,
+      "step": 15797
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.750874716347892e-06,
+      "loss": 1.0215,
+      "step": 15798
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.748688636997382e-06,
+      "loss": 0.8066,
+      "step": 15799
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.74650290348429e-06,
+      "loss": 0.9502,
+      "step": 15800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.744317515861653e-06,
+      "loss": 0.9707,
+      "step": 15801
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.742132474182498e-06,
+      "loss": 0.9365,
+      "step": 15802
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.739947778499866e-06,
+      "loss": 0.8486,
+      "step": 15803
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.737763428866763e-06,
+      "loss": 0.8232,
+      "step": 15804
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7355794253361975e-06,
+      "loss": 0.9287,
+      "step": 15805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.733395767961161e-06,
+      "loss": 0.9307,
+      "step": 15806
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7312124567946615e-06,
+      "loss": 0.9229,
+      "step": 15807
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.729029491889673e-06,
+      "loss": 0.7305,
+      "step": 15808
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.726846873299168e-06,
+      "loss": 0.9697,
+      "step": 15809
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.724664601076106e-06,
+      "loss": 0.7939,
+      "step": 15810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.722482675273462e-06,
+      "loss": 0.7803,
+      "step": 15811
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.720301095944172e-06,
+      "loss": 0.8271,
+      "step": 15812
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7181198631411785e-06,
+      "loss": 0.8926,
+      "step": 15813
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.71593897691741e-06,
+      "loss": 0.9287,
+      "step": 15814
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7137584373258e-06,
+      "loss": 0.6953,
+      "step": 15815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.71157824441926e-06,
+      "loss": 0.6626,
+      "step": 15816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.709398398250694e-06,
+      "loss": 0.9062,
+      "step": 15817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.707218898872995e-06,
+      "loss": 0.7588,
+      "step": 15818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.705039746339067e-06,
+      "loss": 0.9219,
+      "step": 15819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.702860940701784e-06,
+      "loss": 0.8975,
+      "step": 15820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.700682482014019e-06,
+      "loss": 0.9072,
+      "step": 15821
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.698504370328633e-06,
+      "loss": 0.8828,
+      "step": 15822
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.69632660569849e-06,
+      "loss": 0.8203,
+      "step": 15823
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6941491881764365e-06,
+      "loss": 0.8398,
+      "step": 15824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.69197211781531e-06,
+      "loss": 1.0176,
+      "step": 15825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6897953946679365e-06,
+      "loss": 0.9365,
+      "step": 15826
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.687619018787149e-06,
+      "loss": 0.8877,
+      "step": 15827
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.685442990225758e-06,
+      "loss": 0.833,
+      "step": 15828
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.68326730903656e-06,
+      "loss": 0.7715,
+      "step": 15829
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6810919752723684e-06,
+      "loss": 0.8438,
+      "step": 15830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.678916988985963e-06,
+      "loss": 0.8281,
+      "step": 15831
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.676742350230124e-06,
+      "loss": 0.8691,
+      "step": 15832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.674568059057621e-06,
+      "loss": 0.8477,
+      "step": 15833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.672394115521223e-06,
+      "loss": 0.9072,
+      "step": 15834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6702205196736884e-06,
+      "loss": 0.8301,
+      "step": 15835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.668047271567758e-06,
+      "loss": 0.7334,
+      "step": 15836
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.665874371256173e-06,
+      "loss": 0.8311,
+      "step": 15837
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.663701818791656e-06,
+      "loss": 0.8633,
+      "step": 15838
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.66152961422694e-06,
+      "loss": 0.7393,
+      "step": 15839
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.659357757614732e-06,
+      "loss": 0.8604,
+      "step": 15840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.657186249007738e-06,
+      "loss": 0.915,
+      "step": 15841
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.655015088458644e-06,
+      "loss": 0.9336,
+      "step": 15842
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6528442760201556e-06,
+      "loss": 0.9775,
+      "step": 15843
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6506738117449425e-06,
+      "loss": 0.8564,
+      "step": 15844
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.648503695685676e-06,
+      "loss": 0.8574,
+      "step": 15845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6463339278950135e-06,
+      "loss": 0.9102,
+      "step": 15846
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.644164508425618e-06,
+      "loss": 0.791,
+      "step": 15847
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.641995437330133e-06,
+      "loss": 1.0234,
+      "step": 15848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.639826714661192e-06,
+      "loss": 0.8252,
+      "step": 15849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.637658340471419e-06,
+      "loss": 0.9004,
+      "step": 15850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.635490314813448e-06,
+      "loss": 0.9834,
+      "step": 15851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.63332263773988e-06,
+      "loss": 0.7178,
+      "step": 15852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6311553093033225e-06,
+      "loss": 0.9287,
+      "step": 15853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.628988329556359e-06,
+      "loss": 0.9023,
+      "step": 15854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.626821698551594e-06,
+      "loss": 0.8799,
+      "step": 15855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6246554163415954e-06,
+      "loss": 0.7637,
+      "step": 15856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.622489482978934e-06,
+      "loss": 0.873,
+      "step": 15857
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.620323898516164e-06,
+      "loss": 0.8374,
+      "step": 15858
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.618158663005851e-06,
+      "loss": 0.7451,
+      "step": 15859
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.61599377650053e-06,
+      "loss": 0.7773,
+      "step": 15860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.613829239052739e-06,
+      "loss": 0.6709,
+      "step": 15861
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.611665050714998e-06,
+      "loss": 0.7551,
+      "step": 15862
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.609501211539838e-06,
+      "loss": 0.9072,
+      "step": 15863
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.607337721579762e-06,
+      "loss": 1.0352,
+      "step": 15864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.605174580887274e-06,
+      "loss": 0.812,
+      "step": 15865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.603011789514856e-06,
+      "loss": 0.7783,
+      "step": 15866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6008493475150115e-06,
+      "loss": 1.0713,
+      "step": 15867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.598687254940205e-06,
+      "loss": 0.7822,
+      "step": 15868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.596525511842901e-06,
+      "loss": 0.8105,
+      "step": 15869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.5943641182755684e-06,
+      "loss": 0.8545,
+      "step": 15870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.592203074290654e-06,
+      "loss": 0.9512,
+      "step": 15871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.590042379940599e-06,
+      "loss": 0.915,
+      "step": 15872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.58788203527783e-06,
+      "loss": 1.0166,
+      "step": 15873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.585722040354787e-06,
+      "loss": 1.0371,
+      "step": 15874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.583562395223878e-06,
+      "loss": 0.9219,
+      "step": 15875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.581403099937515e-06,
+      "loss": 0.8608,
+      "step": 15876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.579244154548087e-06,
+      "loss": 0.9199,
+      "step": 15877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.577085559107999e-06,
+      "loss": 0.7471,
+      "step": 15878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.574927313669629e-06,
+      "loss": 0.8838,
+      "step": 15879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.572769418285351e-06,
+      "loss": 0.6978,
+      "step": 15880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.570611873007526e-06,
+      "loss": 1.0156,
+      "step": 15881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.568454677888517e-06,
+      "loss": 0.8984,
+      "step": 15882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.566297832980676e-06,
+      "loss": 0.751,
+      "step": 15883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.564141338336335e-06,
+      "loss": 0.7275,
+      "step": 15884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.561985194007826e-06,
+      "loss": 0.6611,
+      "step": 15885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.559829400047481e-06,
+      "loss": 0.9355,
+      "step": 15886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.5576739565076075e-06,
+      "loss": 0.875,
+      "step": 15887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.555518863440515e-06,
+      "loss": 0.8096,
+      "step": 15888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.553364120898495e-06,
+      "loss": 0.9395,
+      "step": 15889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.551209728933846e-06,
+      "loss": 0.8926,
+      "step": 15890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.549055687598845e-06,
+      "loss": 0.8359,
+      "step": 15891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.546901996945766e-06,
+      "loss": 0.8848,
+      "step": 15892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.544748657026861e-06,
+      "loss": 0.8105,
+      "step": 15893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.542595667894404e-06,
+      "loss": 0.8369,
+      "step": 15894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.540443029600631e-06,
+      "loss": 0.8525,
+      "step": 15895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.53829074219778e-06,
+      "loss": 0.9893,
+      "step": 15896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.5361388057380785e-06,
+      "loss": 0.9668,
+      "step": 15897
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.533987220273756e-06,
+      "loss": 0.9404,
+      "step": 15898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.531835985857022e-06,
+      "loss": 0.7568,
+      "step": 15899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.5296851025400785e-06,
+      "loss": 0.9541,
+      "step": 15900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.527534570375117e-06,
+      "loss": 0.9482,
+      "step": 15901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.525384389414334e-06,
+      "loss": 0.6729,
+      "step": 15902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.523234559709905e-06,
+      "loss": 0.8848,
+      "step": 15903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.521085081313997e-06,
+      "loss": 0.8955,
+      "step": 15904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.518935954278768e-06,
+      "loss": 0.7275,
+      "step": 15905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.516787178656382e-06,
+      "loss": 0.9385,
+      "step": 15906
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.514638754498978e-06,
+      "loss": 0.8555,
+      "step": 15907
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.512490681858691e-06,
+      "loss": 0.8955,
+      "step": 15908
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.510342960787643e-06,
+      "loss": 0.9492,
+      "step": 15909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.5081955913379634e-06,
+      "loss": 0.9844,
+      "step": 15910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.50604857356176e-06,
+      "loss": 0.8496,
+      "step": 15911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.503901907511133e-06,
+      "loss": 0.7812,
+      "step": 15912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.501755593238168e-06,
+      "loss": 0.8018,
+      "step": 15913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.499609630794962e-06,
+      "loss": 0.9062,
+      "step": 15914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.497464020233587e-06,
+      "loss": 0.8672,
+      "step": 15915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.495318761606106e-06,
+      "loss": 0.9336,
+      "step": 15916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.493173854964578e-06,
+      "loss": 0.7305,
+      "step": 15917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.491029300361066e-06,
+      "loss": 0.8994,
+      "step": 15918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.488885097847603e-06,
+      "loss": 0.9727,
+      "step": 15919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.486741247476224e-06,
+      "loss": 0.7676,
+      "step": 15920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.484597749298945e-06,
+      "loss": 1.0146,
+      "step": 15921
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.482454603367797e-06,
+      "loss": 1.0684,
+      "step": 15922
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.480311809734781e-06,
+      "loss": 1.0439,
+      "step": 15923
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.478169368451897e-06,
+      "loss": 0.832,
+      "step": 15924
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.47602727957113e-06,
+      "loss": 0.9424,
+      "step": 15925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4738855431444725e-06,
+      "loss": 0.9268,
+      "step": 15926
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.471744159223893e-06,
+      "loss": 0.874,
+      "step": 15927
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.469603127861358e-06,
+      "loss": 0.9941,
+      "step": 15928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.467462449108816e-06,
+      "loss": 0.9473,
+      "step": 15929
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.465322123018226e-06,
+      "loss": 0.6875,
+      "step": 15930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.463182149641526e-06,
+      "loss": 0.8447,
+      "step": 15931
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.461042529030643e-06,
+      "loss": 0.876,
+      "step": 15932
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.458903261237491e-06,
+      "loss": 0.8604,
+      "step": 15933
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.456764346314003e-06,
+      "loss": 0.7324,
+      "step": 15934
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.454625784312073e-06,
+      "loss": 0.7051,
+      "step": 15935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4524875752836e-06,
+      "loss": 1.0117,
+      "step": 15936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.450349719280463e-06,
+      "loss": 0.9111,
+      "step": 15937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.448212216354556e-06,
+      "loss": 0.8447,
+      "step": 15938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4460750665577435e-06,
+      "loss": 0.8496,
+      "step": 15939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4439382699418866e-06,
+      "loss": 0.8535,
+      "step": 15940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.441801826558835e-06,
+      "loss": 0.9639,
+      "step": 15941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4396657364604446e-06,
+      "loss": 0.9473,
+      "step": 15942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.437529999698547e-06,
+      "loss": 0.9326,
+      "step": 15943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.43539461632497e-06,
+      "loss": 0.9062,
+      "step": 15944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4332595863915264e-06,
+      "loss": 0.9414,
+      "step": 15945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.431124909950038e-06,
+      "loss": 0.9219,
+      "step": 15946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.428990587052303e-06,
+      "loss": 0.7568,
+      "step": 15947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.426856617750116e-06,
+      "loss": 1.126,
+      "step": 15948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.424723002095253e-06,
+      "loss": 0.957,
+      "step": 15949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4225897401395055e-06,
+      "loss": 0.834,
+      "step": 15950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.420456831934633e-06,
+      "loss": 0.7725,
+      "step": 15951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4183242775323985e-06,
+      "loss": 0.7993,
+      "step": 15952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.416192076984543e-06,
+      "loss": 0.8867,
+      "step": 15953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.414060230342823e-06,
+      "loss": 0.75,
+      "step": 15954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.411928737658965e-06,
+      "loss": 0.9629,
+      "step": 15955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.40979759898469e-06,
+      "loss": 0.752,
+      "step": 15956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.407666814371724e-06,
+      "loss": 0.832,
+      "step": 15957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.40553638387177e-06,
+      "loss": 0.8975,
+      "step": 15958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4034063075365275e-06,
+      "loss": 0.7988,
+      "step": 15959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.401276585417678e-06,
+      "loss": 0.9531,
+      "step": 15960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.399147217566923e-06,
+      "loss": 0.9834,
+      "step": 15961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.397018204035922e-06,
+      "loss": 1.0146,
+      "step": 15962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.394889544876345e-06,
+      "loss": 1.002,
+      "step": 15963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.392761240139841e-06,
+      "loss": 1.0068,
+      "step": 15964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.390633289878069e-06,
+      "loss": 1.0127,
+      "step": 15965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.388505694142663e-06,
+      "loss": 0.873,
+      "step": 15966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.38637845298525e-06,
+      "loss": 0.8516,
+      "step": 15967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.384251566457452e-06,
+      "loss": 0.9268,
+      "step": 15968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.38212503461089e-06,
+      "loss": 0.8066,
+      "step": 15969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.379998857497164e-06,
+      "loss": 0.7939,
+      "step": 15970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.377873035167869e-06,
+      "loss": 0.8545,
+      "step": 15971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.375747567674589e-06,
+      "loss": 0.7705,
+      "step": 15972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.373622455068912e-06,
+      "loss": 0.9092,
+      "step": 15973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.371497697402404e-06,
+      "loss": 0.9062,
+      "step": 15974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.369373294726625e-06,
+      "loss": 0.8252,
+      "step": 15975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.367249247093125e-06,
+      "loss": 0.8457,
+      "step": 15976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.365125554553456e-06,
+      "loss": 0.8477,
+      "step": 15977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.36300221715915e-06,
+      "loss": 0.7246,
+      "step": 15978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.360879234961735e-06,
+      "loss": 0.7085,
+      "step": 15979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.358756608012725e-06,
+      "loss": 0.7529,
+      "step": 15980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.356634336363638e-06,
+      "loss": 0.9248,
+      "step": 15981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.354512420065972e-06,
+      "loss": 0.9619,
+      "step": 15982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.352390859171217e-06,
+      "loss": 0.8604,
+      "step": 15983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.350269653730853e-06,
+      "loss": 0.9395,
+      "step": 15984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.348148803796369e-06,
+      "loss": 0.8154,
+      "step": 15985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.346028309419222e-06,
+      "loss": 0.8877,
+      "step": 15986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.343908170650873e-06,
+      "loss": 0.8936,
+      "step": 15987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.341788387542766e-06,
+      "loss": 0.8896,
+      "step": 15988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.339668960146351e-06,
+      "loss": 0.8662,
+      "step": 15989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.337549888513056e-06,
+      "loss": 0.8467,
+      "step": 15990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.335431172694305e-06,
+      "loss": 0.9316,
+      "step": 15991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.333312812741504e-06,
+      "loss": 0.9277,
+      "step": 15992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.331194808706077e-06,
+      "loss": 0.9102,
+      "step": 15993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.329077160639413e-06,
+      "loss": 0.8857,
+      "step": 15994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.326959868592898e-06,
+      "loss": 0.8936,
+      "step": 15995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.324842932617911e-06,
+      "loss": 0.8359,
+      "step": 15996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.322726352765833e-06,
+      "loss": 0.9541,
+      "step": 15997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.320610129088022e-06,
+      "loss": 0.8418,
+      "step": 15998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.318494261635827e-06,
+      "loss": 0.8984,
+      "step": 15999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3163787504606e-06,
+      "loss": 0.7368,
+      "step": 16000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.314263595613685e-06,
+      "loss": 0.9531,
+      "step": 16001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.312148797146399e-06,
+      "loss": 0.8369,
+      "step": 16002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.31003435511007e-06,
+      "loss": 0.96,
+      "step": 16003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.307920269555999e-06,
+      "loss": 0.8809,
+      "step": 16004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.305806540535501e-06,
+      "loss": 0.9277,
+      "step": 16005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.303693168099864e-06,
+      "loss": 0.9199,
+      "step": 16006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.301580152300372e-06,
+      "loss": 0.8906,
+      "step": 16007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.299467493188299e-06,
+      "loss": 0.8574,
+      "step": 16008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.297355190814923e-06,
+      "loss": 0.8545,
+      "step": 16009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.295243245231498e-06,
+      "loss": 0.7891,
+      "step": 16010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.293131656489274e-06,
+      "loss": 0.8643,
+      "step": 16011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2910204246394855e-06,
+      "loss": 0.8779,
+      "step": 16012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.288909549733383e-06,
+      "loss": 0.8271,
+      "step": 16013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2867990318221805e-06,
+      "loss": 0.7944,
+      "step": 16014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.284688870957096e-06,
+      "loss": 0.7842,
+      "step": 16015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.282579067189328e-06,
+      "loss": 0.9316,
+      "step": 16016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.280469620570092e-06,
+      "loss": 0.8682,
+      "step": 16017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2783605311505705e-06,
+      "loss": 0.8652,
+      "step": 16018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.276251798981942e-06,
+      "loss": 0.9531,
+      "step": 16019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.274143424115377e-06,
+      "loss": 0.8052,
+      "step": 16020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.272035406602048e-06,
+      "loss": 0.8213,
+      "step": 16021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.269927746493107e-06,
+      "loss": 0.8672,
+      "step": 16022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2678204438397e-06,
+      "loss": 0.8984,
+      "step": 16023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.265713498692959e-06,
+      "loss": 0.7051,
+      "step": 16024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.263606911104027e-06,
+      "loss": 0.8047,
+      "step": 16025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.261500681124016e-06,
+      "loss": 0.875,
+      "step": 16026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.259394808804039e-06,
+      "loss": 0.876,
+      "step": 16027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.257289294195194e-06,
+      "loss": 0.8545,
+      "step": 16028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.255184137348586e-06,
+      "loss": 0.8135,
+      "step": 16029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.253079338315297e-06,
+      "loss": 0.9199,
+      "step": 16030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.250974897146401e-06,
+      "loss": 0.8721,
+      "step": 16031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.248870813892966e-06,
+      "loss": 0.8359,
+      "step": 16032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.246767088606059e-06,
+      "loss": 0.876,
+      "step": 16033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2446637213367294e-06,
+      "loss": 0.9111,
+      "step": 16034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.242560712136016e-06,
+      "loss": 0.9561,
+      "step": 16035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.240458061054947e-06,
+      "loss": 0.8877,
+      "step": 16036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.238355768144561e-06,
+      "loss": 0.8809,
+      "step": 16037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.236253833455869e-06,
+      "loss": 0.8066,
+      "step": 16038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.23415225703987e-06,
+      "loss": 1.0303,
+      "step": 16039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.232051038947579e-06,
+      "loss": 0.8643,
+      "step": 16040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.229950179229977e-06,
+      "loss": 0.8438,
+      "step": 16041
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.227849677938048e-06,
+      "loss": 0.8018,
+      "step": 16042
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2257495351227575e-06,
+      "loss": 0.7861,
+      "step": 16043
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.223649750835083e-06,
+      "loss": 0.7368,
+      "step": 16044
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.221550325125973e-06,
+      "loss": 0.9619,
+      "step": 16045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.219451258046373e-06,
+      "loss": 0.8613,
+      "step": 16046
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.217352549647219e-06,
+      "loss": 0.7656,
+      "step": 16047
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.215254199979449e-06,
+      "loss": 0.7637,
+      "step": 16048
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.213156209093979e-06,
+      "loss": 0.6738,
+      "step": 16049
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.211058577041723e-06,
+      "loss": 0.7354,
+      "step": 16050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.208961303873577e-06,
+      "loss": 0.7905,
+      "step": 16051
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2068643896404445e-06,
+      "loss": 0.7793,
+      "step": 16052
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.204767834393209e-06,
+      "loss": 0.9131,
+      "step": 16053
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.202671638182748e-06,
+      "loss": 0.9316,
+      "step": 16054
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.200575801059924e-06,
+      "loss": 0.6543,
+      "step": 16055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1984803230756055e-06,
+      "loss": 0.9053,
+      "step": 16056
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.196385204280641e-06,
+      "loss": 0.9297,
+      "step": 16057
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.194290444725872e-06,
+      "loss": 0.9824,
+      "step": 16058
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.192196044462126e-06,
+      "loss": 0.876,
+      "step": 16059
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.190102003540239e-06,
+      "loss": 0.8916,
+      "step": 16060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.188008322011022e-06,
+      "loss": 0.8984,
+      "step": 16061
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.185914999925283e-06,
+      "loss": 0.9102,
+      "step": 16062
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1838220373338144e-06,
+      "loss": 0.832,
+      "step": 16063
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.181729434287417e-06,
+      "loss": 0.8467,
+      "step": 16064
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1796371908368685e-06,
+      "loss": 0.9062,
+      "step": 16065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.17754530703294e-06,
+      "loss": 0.8779,
+      "step": 16066
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.17545378292639e-06,
+      "loss": 0.9131,
+      "step": 16067
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.173362618567983e-06,
+      "loss": 0.8174,
+      "step": 16068
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.171271814008465e-06,
+      "loss": 0.9033,
+      "step": 16069
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.169181369298568e-06,
+      "loss": 0.8447,
+      "step": 16070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1670912844890186e-06,
+      "loss": 0.9043,
+      "step": 16071
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.165001559630547e-06,
+      "loss": 0.9141,
+      "step": 16072
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.16291219477386e-06,
+      "loss": 0.7812,
+      "step": 16073
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1608231899696595e-06,
+      "loss": 0.8379,
+      "step": 16074
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.158734545268634e-06,
+      "loss": 0.8252,
+      "step": 16075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1566462607214785e-06,
+      "loss": 0.7803,
+      "step": 16076
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.154558336378868e-06,
+      "loss": 0.8867,
+      "step": 16077
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.152470772291467e-06,
+      "loss": 0.7959,
+      "step": 16078
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1503835685099295e-06,
+      "loss": 0.8193,
+      "step": 16079
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.148296725084916e-06,
+      "loss": 0.9648,
+      "step": 16080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.146210242067064e-06,
+      "loss": 0.6646,
+      "step": 16081
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.144124119507008e-06,
+      "loss": 0.9229,
+      "step": 16082
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.142038357455357e-06,
+      "loss": 0.8047,
+      "step": 16083
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1399529559627525e-06,
+      "loss": 0.6294,
+      "step": 16084
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.137867915079786e-06,
+      "loss": 0.9219,
+      "step": 16085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.135783234857059e-06,
+      "loss": 0.8525,
+      "step": 16086
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.133698915345151e-06,
+      "loss": 0.9883,
+      "step": 16087
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1316149565946575e-06,
+      "loss": 0.8467,
+      "step": 16088
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.129531358656142e-06,
+      "loss": 0.7729,
+      "step": 16089
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.127448121580167e-06,
+      "loss": 0.8018,
+      "step": 16090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.125365245417284e-06,
+      "loss": 0.6543,
+      "step": 16091
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.123282730218046e-06,
+      "loss": 0.959,
+      "step": 16092
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.121200576032985e-06,
+      "loss": 0.9521,
+      "step": 16093
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.119118782912631e-06,
+      "loss": 1.0166,
+      "step": 16094
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.117037350907492e-06,
+      "loss": 0.8965,
+      "step": 16095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1149562800680955e-06,
+      "loss": 0.915,
+      "step": 16096
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.112875570444935e-06,
+      "loss": 0.7568,
+      "step": 16097
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.110795222088502e-06,
+      "loss": 0.7253,
+      "step": 16098
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1087152350492755e-06,
+      "loss": 0.9238,
+      "step": 16099
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.106635609377744e-06,
+      "loss": 0.7666,
+      "step": 16100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.104556345124363e-06,
+      "loss": 0.873,
+      "step": 16101
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1024774423395975e-06,
+      "loss": 0.8584,
+      "step": 16102
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.100398901073884e-06,
+      "loss": 0.7354,
+      "step": 16103
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.098320721377677e-06,
+      "loss": 0.8359,
+      "step": 16104
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.096242903301403e-06,
+      "loss": 0.9541,
+      "step": 16105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.094165446895483e-06,
+      "loss": 0.6885,
+      "step": 16106
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.092088352210325e-06,
+      "loss": 0.8613,
+      "step": 16107
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.090011619296347e-06,
+      "loss": 0.7715,
+      "step": 16108
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.087935248203937e-06,
+      "loss": 0.9551,
+      "step": 16109
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.085859238983484e-06,
+      "loss": 0.9287,
+      "step": 16110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.08378359168536e-06,
+      "loss": 0.9131,
+      "step": 16111
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.081708306359947e-06,
+      "loss": 0.834,
+      "step": 16112
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.079633383057601e-06,
+      "loss": 0.6533,
+      "step": 16113
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.077558821828674e-06,
+      "loss": 0.6836,
+      "step": 16114
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.075484622723504e-06,
+      "loss": 0.71,
+      "step": 16115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.073410785792434e-06,
+      "loss": 0.7256,
+      "step": 16116
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.071337311085787e-06,
+      "loss": 0.7734,
+      "step": 16117
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0692641986538805e-06,
+      "loss": 0.8828,
+      "step": 16118
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.067191448547017e-06,
+      "loss": 1.0596,
+      "step": 16119
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0651190608155065e-06,
+      "loss": 0.8965,
+      "step": 16120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.063047035509636e-06,
+      "loss": 0.9443,
+      "step": 16121
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.060975372679684e-06,
+      "loss": 0.7192,
+      "step": 16122
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.058904072375921e-06,
+      "loss": 0.8271,
+      "step": 16123
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.056833134648622e-06,
+      "loss": 0.7778,
+      "step": 16124
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.054762559548034e-06,
+      "loss": 0.7441,
+      "step": 16125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.052692347124404e-06,
+      "loss": 0.7803,
+      "step": 16126
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0506224974279746e-06,
+      "loss": 0.7358,
+      "step": 16127
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.048553010508976e-06,
+      "loss": 0.8926,
+      "step": 16128
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.046483886417624e-06,
+      "loss": 0.8447,
+      "step": 16129
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.044415125204125e-06,
+      "loss": 0.9121,
+      "step": 16130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.042346726918694e-06,
+      "loss": 0.6289,
+      "step": 16131
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.04027869161152e-06,
+      "loss": 0.7256,
+      "step": 16132
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.038211019332786e-06,
+      "loss": 0.7676,
+      "step": 16133
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0361437101326635e-06,
+      "loss": 0.8643,
+      "step": 16134
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0340767640613334e-06,
+      "loss": 0.8447,
+      "step": 16135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.032010181168945e-06,
+      "loss": 0.7598,
+      "step": 16136
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.029943961505652e-06,
+      "loss": 0.8271,
+      "step": 16137
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.027878105121587e-06,
+      "loss": 0.8945,
+      "step": 16138
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.025812612066896e-06,
+      "loss": 0.6597,
+      "step": 16139
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.023747482391694e-06,
+      "loss": 0.9062,
+      "step": 16140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.021682716146097e-06,
+      "loss": 0.9092,
+      "step": 16141
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.019618313380206e-06,
+      "loss": 0.7559,
+      "step": 16142
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.017554274144128e-06,
+      "loss": 0.9678,
+      "step": 16143
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0154905984879465e-06,
+      "loss": 0.918,
+      "step": 16144
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.013427286461738e-06,
+      "loss": 0.748,
+      "step": 16145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.011364338115572e-06,
+      "loss": 0.9209,
+      "step": 16146
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.00930175349952e-06,
+      "loss": 0.9141,
+      "step": 16147
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.007239532663626e-06,
+      "loss": 0.999,
+      "step": 16148
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0051776756579375e-06,
+      "loss": 0.9824,
+      "step": 16149
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.003116182532484e-06,
+      "loss": 0.7793,
+      "step": 16150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0010550533373e-06,
+      "loss": 0.9023,
+      "step": 16151
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.998994288122401e-06,
+      "loss": 0.8184,
+      "step": 16152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.996933886937794e-06,
+      "loss": 0.752,
+      "step": 16153
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.994873849833475e-06,
+      "loss": 0.9873,
+      "step": 16154
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.992814176859442e-06,
+      "loss": 0.9072,
+      "step": 16155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.990754868065675e-06,
+      "loss": 1.0674,
+      "step": 16156
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.988695923502149e-06,
+      "loss": 0.8613,
+      "step": 16157
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9866373432188185e-06,
+      "loss": 0.7949,
+      "step": 16158
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9845791272656514e-06,
+      "loss": 0.8242,
+      "step": 16159
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.982521275692594e-06,
+      "loss": 0.9072,
+      "step": 16160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9804637885495765e-06,
+      "loss": 0.7275,
+      "step": 16161
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.978406665886528e-06,
+      "loss": 0.7812,
+      "step": 16162
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.976349907753379e-06,
+      "loss": 0.8486,
+      "step": 16163
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.974293514200034e-06,
+      "loss": 0.9004,
+      "step": 16164
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.972237485276399e-06,
+      "loss": 0.8154,
+      "step": 16165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.970181821032349e-06,
+      "loss": 0.8262,
+      "step": 16166
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9681265215178e-06,
+      "loss": 0.7949,
+      "step": 16167
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9660715867826125e-06,
+      "loss": 0.7524,
+      "step": 16168
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.964017016876655e-06,
+      "loss": 0.8643,
+      "step": 16169
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9619628118497765e-06,
+      "loss": 0.8691,
+      "step": 16170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9599089717518436e-06,
+      "loss": 0.9229,
+      "step": 16171
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.957855496632689e-06,
+      "loss": 0.7549,
+      "step": 16172
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9558023865421435e-06,
+      "loss": 0.7861,
+      "step": 16173
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.953749641530024e-06,
+      "loss": 0.8359,
+      "step": 16174
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.951697261646158e-06,
+      "loss": 0.8457,
+      "step": 16175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.949645246940345e-06,
+      "loss": 0.9473,
+      "step": 16176
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.947593597462379e-06,
+      "loss": 0.5889,
+      "step": 16177
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9455423132620415e-06,
+      "loss": 0.6982,
+      "step": 16178
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.943491394389126e-06,
+      "loss": 0.8701,
+      "step": 16179
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.941440840893394e-06,
+      "loss": 1.0547,
+      "step": 16180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9393906528246095e-06,
+      "loss": 0.9531,
+      "step": 16181
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.937340830232515e-06,
+      "loss": 0.9531,
+      "step": 16182
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.935291373166866e-06,
+      "loss": 0.9316,
+      "step": 16183
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.933242281677393e-06,
+      "loss": 1.0371,
+      "step": 16184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.931193555813818e-06,
+      "loss": 0.8018,
+      "step": 16185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.929145195625855e-06,
+      "loss": 0.873,
+      "step": 16186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9270972011632225e-06,
+      "loss": 0.7881,
+      "step": 16187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.925049572475613e-06,
+      "loss": 0.7686,
+      "step": 16188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9230023096127165e-06,
+      "loss": 0.7324,
+      "step": 16189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.920955412624206e-06,
+      "loss": 0.9277,
+      "step": 16190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.918908881559772e-06,
+      "loss": 0.8076,
+      "step": 16191
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.916862716469064e-06,
+      "loss": 0.8701,
+      "step": 16192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.914816917401741e-06,
+      "loss": 0.8213,
+      "step": 16193
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.912771484407442e-06,
+      "loss": 0.9053,
+      "step": 16194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.910726417535816e-06,
+      "loss": 0.7773,
+      "step": 16195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.908681716836481e-06,
+      "loss": 0.9141,
+      "step": 16196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.906637382359061e-06,
+      "loss": 0.8799,
+      "step": 16197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9045934141531584e-06,
+      "loss": 0.8164,
+      "step": 16198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.902549812268382e-06,
+      "loss": 0.9424,
+      "step": 16199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.900506576754324e-06,
+      "loss": 0.9004,
+      "step": 16200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8984637076605675e-06,
+      "loss": 0.7109,
+      "step": 16201
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8964212050366765e-06,
+      "loss": 1.0029,
+      "step": 16202
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.894379068932231e-06,
+      "loss": 0.752,
+      "step": 16203
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.892337299396781e-06,
+      "loss": 0.8418,
+      "step": 16204
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8902958964798754e-06,
+      "loss": 0.6812,
+      "step": 16205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.888254860231046e-06,
+      "loss": 0.918,
+      "step": 16206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.886214190699836e-06,
+      "loss": 0.9189,
+      "step": 16207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.884173887935759e-06,
+      "loss": 0.751,
+      "step": 16208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.882133951988326e-06,
+      "loss": 0.7168,
+      "step": 16209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.880094382907039e-06,
+      "loss": 0.7744,
+      "step": 16210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.878055180741399e-06,
+      "loss": 0.8193,
+      "step": 16211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.876016345540888e-06,
+      "loss": 0.8311,
+      "step": 16212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.873977877354978e-06,
+      "loss": 0.7998,
+      "step": 16213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.871939776233145e-06,
+      "loss": 0.8779,
+      "step": 16214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.869902042224846e-06,
+      "loss": 0.6667,
+      "step": 16215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.867864675379525e-06,
+      "loss": 0.9268,
+      "step": 16216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.865827675746622e-06,
+      "loss": 0.7764,
+      "step": 16217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.863791043375578e-06,
+      "loss": 0.8232,
+      "step": 16218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.861754778315814e-06,
+      "loss": 0.8184,
+      "step": 16219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.859718880616739e-06,
+      "loss": 0.7656,
+      "step": 16220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.857683350327758e-06,
+      "loss": 0.7842,
+      "step": 16221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.855648187498274e-06,
+      "loss": 0.8232,
+      "step": 16222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.853613392177669e-06,
+      "loss": 0.8286,
+      "step": 16223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.851578964415326e-06,
+      "loss": 0.7686,
+      "step": 16224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.849544904260603e-06,
+      "loss": 0.9629,
+      "step": 16225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.847511211762878e-06,
+      "loss": 0.8496,
+      "step": 16226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.845477886971493e-06,
+      "loss": 0.7803,
+      "step": 16227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.84344492993579e-06,
+      "loss": 0.7666,
+      "step": 16228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.841412340705101e-06,
+      "loss": 0.8271,
+      "step": 16229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.839380119328761e-06,
+      "loss": 0.7212,
+      "step": 16230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.837348265856079e-06,
+      "loss": 0.6958,
+      "step": 16231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.835316780336361e-06,
+      "loss": 0.8008,
+      "step": 16232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.833285662818903e-06,
+      "loss": 0.7988,
+      "step": 16233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.831254913353003e-06,
+      "loss": 0.7129,
+      "step": 16234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.829224531987937e-06,
+      "loss": 0.874,
+      "step": 16235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.827194518772978e-06,
+      "loss": 0.7051,
+      "step": 16236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.825164873757378e-06,
+      "loss": 0.8477,
+      "step": 16237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.823135596990405e-06,
+      "loss": 0.8828,
+      "step": 16238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8211066885213e-06,
+      "loss": 0.8096,
+      "step": 16239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.819078148399292e-06,
+      "loss": 0.7812,
+      "step": 16240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.817049976673609e-06,
+      "loss": 0.8926,
+      "step": 16241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.815022173393478e-06,
+      "loss": 0.9629,
+      "step": 16242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.812994738608099e-06,
+      "loss": 0.8389,
+      "step": 16243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.810967672366675e-06,
+      "loss": 0.8809,
+      "step": 16244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.808940974718393e-06,
+      "loss": 0.877,
+      "step": 16245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.806914645712441e-06,
+      "loss": 0.877,
+      "step": 16246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.80488868539799e-06,
+      "loss": 0.6255,
+      "step": 16247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.802863093824204e-06,
+      "loss": 0.749,
+      "step": 16248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.800837871040231e-06,
+      "loss": 0.8574,
+      "step": 16249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.798813017095225e-06,
+      "loss": 0.7373,
+      "step": 16250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7967885320383256e-06,
+      "loss": 0.793,
+      "step": 16251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.79476441591866e-06,
+      "loss": 0.9082,
+      "step": 16252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.792740668785338e-06,
+      "loss": 0.9482,
+      "step": 16253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.790717290687484e-06,
+      "loss": 0.8232,
+      "step": 16254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7886942816741886e-06,
+      "loss": 0.957,
+      "step": 16255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.786671641794551e-06,
+      "loss": 0.9033,
+      "step": 16256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.784649371097647e-06,
+      "loss": 0.7939,
+      "step": 16257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.782627469632561e-06,
+      "loss": 0.9355,
+      "step": 16258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.780605937448353e-06,
+      "loss": 0.9619,
+      "step": 16259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.778584774594081e-06,
+      "loss": 0.9629,
+      "step": 16260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.776563981118784e-06,
+      "loss": 0.8848,
+      "step": 16261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7745435570715185e-06,
+      "loss": 0.9209,
+      "step": 16262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.772523502501301e-06,
+      "loss": 0.8652,
+      "step": 16263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.770503817457157e-06,
+      "loss": 0.9297,
+      "step": 16264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.768484501988091e-06,
+      "loss": 0.8643,
+      "step": 16265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.766465556143118e-06,
+      "loss": 0.8193,
+      "step": 16266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7644469799712265e-06,
+      "loss": 0.8955,
+      "step": 16267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.762428773521399e-06,
+      "loss": 0.9404,
+      "step": 16268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.760410936842608e-06,
+      "loss": 0.9199,
+      "step": 16269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7583934699838304e-06,
+      "loss": 0.9375,
+      "step": 16270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7563763729940204e-06,
+      "loss": 0.8525,
+      "step": 16271
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7543596459221264e-06,
+      "loss": 0.9434,
+      "step": 16272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.752343288817081e-06,
+      "loss": 0.8926,
+      "step": 16273
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.750327301727828e-06,
+      "loss": 0.9062,
+      "step": 16274
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.748311684703286e-06,
+      "loss": 0.9248,
+      "step": 16275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.746296437792364e-06,
+      "loss": 0.8125,
+      "step": 16276
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.744281561043962e-06,
+      "loss": 0.9336,
+      "step": 16277
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.742267054506986e-06,
+      "loss": 0.7822,
+      "step": 16278
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.740252918230319e-06,
+      "loss": 0.9443,
+      "step": 16279
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7382391522628355e-06,
+      "loss": 0.8271,
+      "step": 16280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.736225756653399e-06,
+      "loss": 1.0459,
+      "step": 16281
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.73421273145088e-06,
+      "loss": 0.7012,
+      "step": 16282
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.732200076704123e-06,
+      "loss": 0.793,
+      "step": 16283
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.730187792461969e-06,
+      "loss": 0.8271,
+      "step": 16284
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.728175878773244e-06,
+      "loss": 0.9238,
+      "step": 16285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.726164335686785e-06,
+      "loss": 0.8975,
+      "step": 16286
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.724153163251397e-06,
+      "loss": 1.0508,
+      "step": 16287
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.722142361515889e-06,
+      "loss": 0.957,
+      "step": 16288
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.720131930529049e-06,
+      "loss": 0.9092,
+      "step": 16289
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7181218703396735e-06,
+      "loss": 0.7188,
+      "step": 16290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.716112180996541e-06,
+      "loss": 0.6655,
+      "step": 16291
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.714102862548417e-06,
+      "loss": 0.853,
+      "step": 16292
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.712093915044056e-06,
+      "loss": 0.8662,
+      "step": 16293
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.710085338532222e-06,
+      "loss": 0.7871,
+      "step": 16294
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.70807713306165e-06,
+      "loss": 0.8623,
+      "step": 16295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.706069298681069e-06,
+      "loss": 0.8037,
+      "step": 16296
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.704061835439213e-06,
+      "loss": 0.8047,
+      "step": 16297
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7020547433847945e-06,
+      "loss": 0.9346,
+      "step": 16298
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.700048022566515e-06,
+      "loss": 0.8633,
+      "step": 16299
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.698041673033071e-06,
+      "loss": 0.915,
+      "step": 16300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.696035694833159e-06,
+      "loss": 0.8467,
+      "step": 16301
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6940300880154534e-06,
+      "loss": 0.832,
+      "step": 16302
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.692024852628625e-06,
+      "loss": 0.873,
+      "step": 16303
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.690019988721326e-06,
+      "loss": 0.7578,
+      "step": 16304
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.688015496342226e-06,
+      "loss": 0.8477,
+      "step": 16305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.686011375539956e-06,
+      "loss": 0.9883,
+      "step": 16306
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.684007626363156e-06,
+      "loss": 0.9004,
+      "step": 16307
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.682004248860441e-06,
+      "loss": 0.8594,
+      "step": 16308
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.680001243080439e-06,
+      "loss": 0.8447,
+      "step": 16309
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.677998609071754e-06,
+      "loss": 0.7354,
+      "step": 16310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.675996346882983e-06,
+      "loss": 0.8428,
+      "step": 16311
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.673994456562707e-06,
+      "loss": 0.7881,
+      "step": 16312
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.671992938159517e-06,
+      "loss": 0.9297,
+      "step": 16313
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6699917917219835e-06,
+      "loss": 0.8574,
+      "step": 16314
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6679910172986655e-06,
+      "loss": 0.8457,
+      "step": 16315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.665990614938109e-06,
+      "loss": 0.8896,
+      "step": 16316
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.663990584688871e-06,
+      "loss": 0.8174,
+      "step": 16317
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.661990926599478e-06,
+      "loss": 0.9229,
+      "step": 16318
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.65999164071846e-06,
+      "loss": 0.8237,
+      "step": 16319
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6579927270943244e-06,
+      "loss": 0.9658,
+      "step": 16320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.655994185775592e-06,
+      "loss": 0.9775,
+      "step": 16321
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.653996016810756e-06,
+      "loss": 0.9551,
+      "step": 16322
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6519982202483064e-06,
+      "loss": 0.8633,
+      "step": 16323
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.650000796136716e-06,
+      "loss": 1.0381,
+      "step": 16324
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6480037445244695e-06,
+      "loss": 0.8516,
+      "step": 16325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.646007065460025e-06,
+      "loss": 0.7227,
+      "step": 16326
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.644010758991833e-06,
+      "loss": 0.8145,
+      "step": 16327
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.642014825168335e-06,
+      "loss": 0.7656,
+      "step": 16328
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.640019264037976e-06,
+      "loss": 0.875,
+      "step": 16329
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.638024075649177e-06,
+      "loss": 0.791,
+      "step": 16330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.636029260050356e-06,
+      "loss": 1.0439,
+      "step": 16331
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.634034817289914e-06,
+      "loss": 0.8975,
+      "step": 16332
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6320407474162595e-06,
+      "loss": 0.7725,
+      "step": 16333
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.630047050477786e-06,
+      "loss": 0.7686,
+      "step": 16334
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.628053726522867e-06,
+      "loss": 0.9795,
+      "step": 16335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.626060775599874e-06,
+      "loss": 0.7568,
+      "step": 16336
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.624068197757176e-06,
+      "loss": 0.918,
+      "step": 16337
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.622075993043127e-06,
+      "loss": 0.7705,
+      "step": 16338
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.620084161506066e-06,
+      "loss": 0.7534,
+      "step": 16339
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.618092703194328e-06,
+      "loss": 0.9805,
+      "step": 16340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.616101618156248e-06,
+      "loss": 0.9258,
+      "step": 16341
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.614110906440139e-06,
+      "loss": 0.8838,
+      "step": 16342
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6121205680943115e-06,
+      "loss": 0.8662,
+      "step": 16343
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6101306031670555e-06,
+      "loss": 0.7803,
+      "step": 16344
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.608141011706677e-06,
+      "loss": 0.8838,
+      "step": 16345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.60615179376145e-06,
+      "loss": 0.9893,
+      "step": 16346
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6041629493796496e-06,
+      "loss": 0.791,
+      "step": 16347
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.602174478609529e-06,
+      "loss": 0.8877,
+      "step": 16348
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.600186381499354e-06,
+      "loss": 0.873,
+      "step": 16349
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.59819865809737e-06,
+      "loss": 0.9287,
+      "step": 16350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5962113084518075e-06,
+      "loss": 0.9775,
+      "step": 16351
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.59422433261089e-06,
+      "loss": 0.8682,
+      "step": 16352
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.592237730622848e-06,
+      "loss": 1.0146,
+      "step": 16353
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5902515025358825e-06,
+      "loss": 0.6621,
+      "step": 16354
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.588265648398196e-06,
+      "loss": 0.8564,
+      "step": 16355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.58628016825797e-06,
+      "loss": 0.8008,
+      "step": 16356
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5842950621634015e-06,
+      "loss": 0.8057,
+      "step": 16357
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.582310330162656e-06,
+      "loss": 0.915,
+      "step": 16358
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.580325972303898e-06,
+      "loss": 0.9072,
+      "step": 16359
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.578341988635273e-06,
+      "loss": 0.8369,
+      "step": 16360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.576358379204941e-06,
+      "loss": 1.0068,
+      "step": 16361
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5743751440610315e-06,
+      "loss": 0.876,
+      "step": 16362
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.572392283251674e-06,
+      "loss": 0.8965,
+      "step": 16363
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.570409796824977e-06,
+      "loss": 0.8447,
+      "step": 16364
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.568427684829064e-06,
+      "loss": 0.8691,
+      "step": 16365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.566445947312028e-06,
+      "loss": 0.9336,
+      "step": 16366
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.564464584321962e-06,
+      "loss": 0.8779,
+      "step": 16367
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.562483595906939e-06,
+      "loss": 0.7773,
+      "step": 16368
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.560502982115045e-06,
+      "loss": 0.8447,
+      "step": 16369
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.558522742994338e-06,
+      "loss": 0.7227,
+      "step": 16370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.556542878592871e-06,
+      "loss": 0.9395,
+      "step": 16371
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.554563388958686e-06,
+      "loss": 0.8887,
+      "step": 16372
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5525842741398305e-06,
+      "loss": 0.8389,
+      "step": 16373
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.550605534184325e-06,
+      "loss": 0.7207,
+      "step": 16374
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.548627169140187e-06,
+      "loss": 0.8193,
+      "step": 16375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.546649179055423e-06,
+      "loss": 0.8516,
+      "step": 16376
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.544671563978043e-06,
+      "loss": 0.8501,
+      "step": 16377
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5426943239560296e-06,
+      "loss": 0.8594,
+      "step": 16378
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.540717459037367e-06,
+      "loss": 0.9043,
+      "step": 16379
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5387409692700215e-06,
+      "loss": 0.9492,
+      "step": 16380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.536764854701969e-06,
+      "loss": 0.769,
+      "step": 16381
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.534789115381158e-06,
+      "loss": 0.9033,
+      "step": 16382
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.532813751355527e-06,
+      "loss": 0.7588,
+      "step": 16383
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.530838762673027e-06,
+      "loss": 0.8232,
+      "step": 16384
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.528864149381573e-06,
+      "loss": 0.8721,
+      "step": 16385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.526889911529091e-06,
+      "loss": 0.7939,
+      "step": 16386
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.524916049163479e-06,
+      "loss": 0.877,
+      "step": 16387
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.522942562332648e-06,
+      "loss": 0.7861,
+      "step": 16388
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.520969451084487e-06,
+      "loss": 0.9746,
+      "step": 16389
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.518996715466875e-06,
+      "loss": 0.8838,
+      "step": 16390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5170243555276774e-06,
+      "loss": 0.873,
+      "step": 16391
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.515052371314772e-06,
+      "loss": 0.9238,
+      "step": 16392
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.513080762876005e-06,
+      "loss": 0.8896,
+      "step": 16393
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.511109530259224e-06,
+      "loss": 0.8086,
+      "step": 16394
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5091386735122565e-06,
+      "loss": 0.9648,
+      "step": 16395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.507168192682943e-06,
+      "loss": 0.9922,
+      "step": 16396
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.505198087819094e-06,
+      "loss": 0.7788,
+      "step": 16397
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.503228358968519e-06,
+      "loss": 1.0039,
+      "step": 16398
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.50125900617901e-06,
+      "loss": 0.8037,
+      "step": 16399
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4992900294983734e-06,
+      "loss": 0.9082,
+      "step": 16400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.49732142897438e-06,
+      "loss": 0.9814,
+      "step": 16401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.495353204654804e-06,
+      "loss": 0.8145,
+      "step": 16402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.493385356587403e-06,
+      "loss": 0.8984,
+      "step": 16403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.491417884819942e-06,
+      "loss": 0.7334,
+      "step": 16404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.489450789400158e-06,
+      "loss": 0.9385,
+      "step": 16405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.487484070375789e-06,
+      "loss": 0.8779,
+      "step": 16406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.485517727794557e-06,
+      "loss": 0.8203,
+      "step": 16407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.483551761704188e-06,
+      "loss": 0.9629,
+      "step": 16408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.481586172152385e-06,
+      "loss": 0.8936,
+      "step": 16409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.479620959186848e-06,
+      "loss": 0.7363,
+      "step": 16410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.477656122855263e-06,
+      "loss": 0.7773,
+      "step": 16411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4756916632053175e-06,
+      "loss": 0.8789,
+      "step": 16412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.473727580284679e-06,
+      "loss": 0.917,
+      "step": 16413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.471763874141013e-06,
+      "loss": 0.8574,
+      "step": 16414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.469800544821965e-06,
+      "loss": 0.8706,
+      "step": 16415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.467837592375184e-06,
+      "loss": 0.9912,
+      "step": 16416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4658750168483135e-06,
+      "loss": 0.7168,
+      "step": 16417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4639128182889715e-06,
+      "loss": 0.8809,
+      "step": 16418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.461950996744775e-06,
+      "loss": 0.8604,
+      "step": 16419
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.459989552263326e-06,
+      "loss": 0.8281,
+      "step": 16420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4580284848922365e-06,
+      "loss": 0.7773,
+      "step": 16421
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.456067794679086e-06,
+      "loss": 0.876,
+      "step": 16422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.454107481671452e-06,
+      "loss": 0.8394,
+      "step": 16423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.452147545916918e-06,
+      "loss": 0.5605,
+      "step": 16424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.450187987463037e-06,
+      "loss": 0.6279,
+      "step": 16425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.448228806357365e-06,
+      "loss": 0.7578,
+      "step": 16426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.446270002647435e-06,
+      "loss": 0.9219,
+      "step": 16427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.444311576380798e-06,
+      "loss": 0.7231,
+      "step": 16428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.44235352760497e-06,
+      "loss": 0.9355,
+      "step": 16429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.44039585636747e-06,
+      "loss": 0.7441,
+      "step": 16430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.438438562715797e-06,
+      "loss": 0.9785,
+      "step": 16431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.436481646697459e-06,
+      "loss": 1.043,
+      "step": 16432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.434525108359944e-06,
+      "loss": 0.8633,
+      "step": 16433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.432568947750726e-06,
+      "loss": 0.8809,
+      "step": 16434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.430613164917273e-06,
+      "loss": 0.7793,
+      "step": 16435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.428657759907056e-06,
+      "loss": 0.8779,
+      "step": 16436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.426702732767521e-06,
+      "loss": 0.8799,
+      "step": 16437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.424748083546111e-06,
+      "loss": 0.7832,
+      "step": 16438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4227938122902556e-06,
+      "loss": 0.9346,
+      "step": 16439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.420839919047389e-06,
+      "loss": 0.8105,
+      "step": 16440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.418886403864919e-06,
+      "loss": 0.8467,
+      "step": 16441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.416933266790256e-06,
+      "loss": 0.8564,
+      "step": 16442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.414980507870788e-06,
+      "loss": 0.9482,
+      "step": 16443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4130281271539135e-06,
+      "loss": 0.8574,
+      "step": 16444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.411076124687008e-06,
+      "loss": 0.9922,
+      "step": 16445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.409124500517439e-06,
+      "loss": 0.8135,
+      "step": 16446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.407173254692563e-06,
+      "loss": 0.8232,
+      "step": 16447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.405222387259737e-06,
+      "loss": 0.9561,
+      "step": 16448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.403271898266306e-06,
+      "loss": 0.9287,
+      "step": 16449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.401321787759594e-06,
+      "loss": 0.8506,
+      "step": 16450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3993720557869255e-06,
+      "loss": 0.8955,
+      "step": 16451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.397422702395622e-06,
+      "loss": 0.791,
+      "step": 16452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3954737276329816e-06,
+      "loss": 0.8701,
+      "step": 16453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.393525131546306e-06,
+      "loss": 0.9131,
+      "step": 16454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.391576914182871e-06,
+      "loss": 1.0391,
+      "step": 16455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.389629075589967e-06,
+      "loss": 0.9014,
+      "step": 16456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.387681615814859e-06,
+      "loss": 0.9326,
+      "step": 16457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.385734534904802e-06,
+      "loss": 0.9316,
+      "step": 16458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.383787832907043e-06,
+      "loss": 0.874,
+      "step": 16459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.381841509868832e-06,
+      "loss": 0.9082,
+      "step": 16460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.379895565837397e-06,
+      "loss": 0.8584,
+      "step": 16461
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3779500008599605e-06,
+      "loss": 0.8545,
+      "step": 16462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.376004814983727e-06,
+      "loss": 0.8828,
+      "step": 16463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.374060008255916e-06,
+      "loss": 0.9092,
+      "step": 16464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.372115580723713e-06,
+      "loss": 0.752,
+      "step": 16465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.370171532434307e-06,
+      "loss": 0.75,
+      "step": 16466
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.368227863434864e-06,
+      "loss": 0.7803,
+      "step": 16467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.366284573772566e-06,
+      "loss": 0.7666,
+      "step": 16468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.364341663494565e-06,
+      "loss": 0.9072,
+      "step": 16469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.362399132648001e-06,
+      "loss": 0.8174,
+      "step": 16470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.360456981280028e-06,
+      "loss": 0.8408,
+      "step": 16471
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.358515209437772e-06,
+      "loss": 0.9277,
+      "step": 16472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3565738171683505e-06,
+      "loss": 0.8589,
+      "step": 16473
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.35463280451887e-06,
+      "loss": 0.8506,
+      "step": 16474
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3526921715364475e-06,
+      "loss": 0.7881,
+      "step": 16475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.350751918268168e-06,
+      "loss": 0.7334,
+      "step": 16476
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.348812044761117e-06,
+      "loss": 0.9922,
+      "step": 16477
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.346872551062364e-06,
+      "loss": 0.877,
+      "step": 16478
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.344933437218986e-06,
+      "loss": 0.9541,
+      "step": 16479
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3429947032780315e-06,
+      "loss": 0.8682,
+      "step": 16480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.34105634928655e-06,
+      "loss": 0.8418,
+      "step": 16481
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.339118375291576e-06,
+      "loss": 0.5776,
+      "step": 16482
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.337180781340147e-06,
+      "loss": 0.7061,
+      "step": 16483
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.335243567479277e-06,
+      "loss": 0.9814,
+      "step": 16484
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.333306733755977e-06,
+      "loss": 0.8887,
+      "step": 16485
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.331370280217244e-06,
+      "loss": 1.0117,
+      "step": 16486
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3294342069100795e-06,
+      "loss": 0.8906,
+      "step": 16487
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.327498513881461e-06,
+      "loss": 0.7734,
+      "step": 16488
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.325563201178364e-06,
+      "loss": 1.0361,
+      "step": 16489
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.323628268847744e-06,
+      "loss": 0.8242,
+      "step": 16490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.32169371693657e-06,
+      "loss": 0.9238,
+      "step": 16491
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3197595454917795e-06,
+      "loss": 0.9209,
+      "step": 16492
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.317825754560312e-06,
+      "loss": 0.832,
+      "step": 16493
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.31589234418909e-06,
+      "loss": 0.8486,
+      "step": 16494
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.313959314425038e-06,
+      "loss": 0.9111,
+      "step": 16495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.312026665315066e-06,
+      "loss": 0.8267,
+      "step": 16496
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.310094396906066e-06,
+      "loss": 0.8945,
+      "step": 16497
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.308162509244929e-06,
+      "loss": 0.8271,
+      "step": 16498
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.306231002378547e-06,
+      "loss": 0.6953,
+      "step": 16499
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.304299876353777e-06,
+      "loss": 0.877,
+      "step": 16500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3023691312174965e-06,
+      "loss": 0.6982,
+      "step": 16501
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.30043876701655e-06,
+      "loss": 0.9033,
+      "step": 16502
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.298508783797781e-06,
+      "loss": 0.9561,
+      "step": 16503
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.296579181608032e-06,
+      "loss": 1.04,
+      "step": 16504
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.294649960494124e-06,
+      "loss": 0.9287,
+      "step": 16505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.292721120502874e-06,
+      "loss": 0.9795,
+      "step": 16506
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.290792661681082e-06,
+      "loss": 0.5947,
+      "step": 16507
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.28886458407556e-06,
+      "loss": 0.9258,
+      "step": 16508
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.286936887733091e-06,
+      "loss": 0.8008,
+      "step": 16509
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.285009572700445e-06,
+      "loss": 0.8994,
+      "step": 16510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.283082639024407e-06,
+      "loss": 0.9404,
+      "step": 16511
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.281156086751732e-06,
+      "loss": 0.8506,
+      "step": 16512
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.27922991592917e-06,
+      "loss": 0.7661,
+      "step": 16513
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.277304126603461e-06,
+      "loss": 0.9277,
+      "step": 16514
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.275378718821348e-06,
+      "loss": 0.7451,
+      "step": 16515
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.273453692629548e-06,
+      "loss": 0.7451,
+      "step": 16516
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.271529048074776e-06,
+      "loss": 0.6938,
+      "step": 16517
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.269604785203734e-06,
+      "loss": 0.999,
+      "step": 16518
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.267680904063128e-06,
+      "loss": 0.8037,
+      "step": 16519
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.26575740469964e-06,
+      "loss": 0.8887,
+      "step": 16520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2638342871599445e-06,
+      "loss": 0.9707,
+      "step": 16521
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.26191155149071e-06,
+      "loss": 0.9414,
+      "step": 16522
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.259989197738601e-06,
+      "loss": 0.8218,
+      "step": 16523
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.258067225950264e-06,
+      "loss": 0.8984,
+      "step": 16524
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.256145636172342e-06,
+      "loss": 0.9404,
+      "step": 16525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.254224428451459e-06,
+      "loss": 0.8066,
+      "step": 16526
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.252303602834247e-06,
+      "loss": 0.7578,
+      "step": 16527
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.250383159367315e-06,
+      "loss": 0.9658,
+      "step": 16528
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.248463098097264e-06,
+      "loss": 0.7656,
+      "step": 16529
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.246543419070687e-06,
+      "loss": 0.8623,
+      "step": 16530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.244624122334176e-06,
+      "loss": 0.9092,
+      "step": 16531
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.242705207934304e-06,
+      "loss": 0.8262,
+      "step": 16532
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.240786675917634e-06,
+      "loss": 0.7715,
+      "step": 16533
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.238868526330722e-06,
+      "loss": 0.9648,
+      "step": 16534
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.236950759220122e-06,
+      "loss": 0.96,
+      "step": 16535
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.235033374632371e-06,
+      "loss": 0.7539,
+      "step": 16536
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.233116372613997e-06,
+      "loss": 0.832,
+      "step": 16537
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.231199753211515e-06,
+      "loss": 0.8906,
+      "step": 16538
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.229283516471445e-06,
+      "loss": 0.7695,
+      "step": 16539
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2273676624402845e-06,
+      "loss": 0.7339,
+      "step": 16540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2254521911645255e-06,
+      "loss": 0.9424,
+      "step": 16541
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.223537102690642e-06,
+      "loss": 0.7588,
+      "step": 16542
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.221622397065126e-06,
+      "loss": 0.8906,
+      "step": 16543
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.21970807433443e-06,
+      "loss": 0.9727,
+      "step": 16544
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.21779413454501e-06,
+      "loss": 0.9141,
+      "step": 16545
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2158805777433094e-06,
+      "loss": 0.667,
+      "step": 16546
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.213967403975772e-06,
+      "loss": 0.9707,
+      "step": 16547
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.212054613288821e-06,
+      "loss": 0.8516,
+      "step": 16548
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.210142205728871e-06,
+      "loss": 0.7266,
+      "step": 16549
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2082301813423325e-06,
+      "loss": 0.998,
+      "step": 16550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.206318540175607e-06,
+      "loss": 0.7949,
+      "step": 16551
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.204407282275085e-06,
+      "loss": 0.8174,
+      "step": 16552
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.20249640768714e-06,
+      "loss": 0.8936,
+      "step": 16553
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.200585916458153e-06,
+      "loss": 0.9277,
+      "step": 16554
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.198675808634482e-06,
+      "loss": 0.8652,
+      "step": 16555
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.196766084262478e-06,
+      "loss": 0.8057,
+      "step": 16556
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.19485674338848e-06,
+      "loss": 0.8125,
+      "step": 16557
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1929477860588345e-06,
+      "loss": 0.7563,
+      "step": 16558
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.191039212319859e-06,
+      "loss": 0.8994,
+      "step": 16559
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.189131022217869e-06,
+      "loss": 0.8281,
+      "step": 16560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.187223215799166e-06,
+      "loss": 0.6504,
+      "step": 16561
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.185315793110056e-06,
+      "loss": 0.8115,
+      "step": 16562
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.183408754196825e-06,
+      "loss": 0.8262,
+      "step": 16563
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.181502099105745e-06,
+      "loss": 0.8262,
+      "step": 16564
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.179595827883085e-06,
+      "loss": 0.8301,
+      "step": 16565
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1776899405751135e-06,
+      "loss": 0.7144,
+      "step": 16566
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.175784437228076e-06,
+      "loss": 0.8496,
+      "step": 16567
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.173879317888212e-06,
+      "loss": 0.8965,
+      "step": 16568
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1719745826017495e-06,
+      "loss": 0.9619,
+      "step": 16569
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.17007023141492e-06,
+      "loss": 0.6885,
+      "step": 16570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.168166264373932e-06,
+      "loss": 0.7715,
+      "step": 16571
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.166262681524988e-06,
+      "loss": 0.916,
+      "step": 16572
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.164359482914279e-06,
+      "loss": 0.9072,
+      "step": 16573
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.162456668588e-06,
+      "loss": 0.9297,
+      "step": 16574
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.160554238592322e-06,
+      "loss": 0.689,
+      "step": 16575
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.158652192973409e-06,
+      "loss": 0.8545,
+      "step": 16576
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.156750531777414e-06,
+      "loss": 0.7373,
+      "step": 16577
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.154849255050497e-06,
+      "loss": 0.7246,
+      "step": 16578
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.152948362838789e-06,
+      "loss": 0.749,
+      "step": 16579
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.151047855188419e-06,
+      "loss": 0.8486,
+      "step": 16580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.149147732145502e-06,
+      "loss": 0.832,
+      "step": 16581
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1472479937561605e-06,
+      "loss": 0.7754,
+      "step": 16582
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.145348640066482e-06,
+      "loss": 0.8369,
+      "step": 16583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.143449671122573e-06,
+      "loss": 0.7637,
+      "step": 16584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.141551086970506e-06,
+      "loss": 0.8916,
+      "step": 16585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.139652887656352e-06,
+      "loss": 0.834,
+      "step": 16586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.137755073226182e-06,
+      "loss": 0.7686,
+      "step": 16587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.135857643726049e-06,
+      "loss": 1.043,
+      "step": 16588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.133960599201996e-06,
+      "loss": 0.8975,
+      "step": 16589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.132063939700053e-06,
+      "loss": 0.877,
+      "step": 16590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.130167665266258e-06,
+      "loss": 0.9482,
+      "step": 16591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.128271775946622e-06,
+      "loss": 0.8203,
+      "step": 16592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.126376271787149e-06,
+      "loss": 0.6689,
+      "step": 16593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.124481152833846e-06,
+      "loss": 0.8994,
+      "step": 16594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.122586419132695e-06,
+      "loss": 0.9336,
+      "step": 16595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1206920707296795e-06,
+      "loss": 0.8486,
+      "step": 16596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.118798107670763e-06,
+      "loss": 0.9248,
+      "step": 16597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.116904530001917e-06,
+      "loss": 0.9248,
+      "step": 16598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.115011337769086e-06,
+      "loss": 0.7959,
+      "step": 16599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.113118531018216e-06,
+      "loss": 0.9473,
+      "step": 16600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1112261097952295e-06,
+      "loss": 0.9814,
+      "step": 16601
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.109334074146065e-06,
+      "loss": 0.8584,
+      "step": 16602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.10744242411663e-06,
+      "loss": 0.7446,
+      "step": 16603
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.10555115975283e-06,
+      "loss": 0.8008,
+      "step": 16604
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1036602811005526e-06,
+      "loss": 0.8408,
+      "step": 16605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.101769788205696e-06,
+      "loss": 0.9287,
+      "step": 16606
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0998796811141316e-06,
+      "loss": 0.8267,
+      "step": 16607
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.097989959871729e-06,
+      "loss": 0.9668,
+      "step": 16608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.096100624524335e-06,
+      "loss": 0.8203,
+      "step": 16609
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.094211675117814e-06,
+      "loss": 0.8633,
+      "step": 16610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.092323111698e-06,
+      "loss": 0.8125,
+      "step": 16611
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0904349343107185e-06,
+      "loss": 0.9111,
+      "step": 16612
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.088547143001791e-06,
+      "loss": 0.9434,
+      "step": 16613
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.086659737817034e-06,
+      "loss": 0.9619,
+      "step": 16614
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.084772718802243e-06,
+      "loss": 0.8555,
+      "step": 16615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.082886086003217e-06,
+      "loss": 0.9707,
+      "step": 16616
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.080999839465729e-06,
+      "loss": 0.7739,
+      "step": 16617
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.079113979235563e-06,
+      "loss": 0.7043,
+      "step": 16618
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.07722850535848e-06,
+      "loss": 0.9375,
+      "step": 16619
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.075343417880233e-06,
+      "loss": 0.7007,
+      "step": 16620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0734587168465635e-06,
+      "loss": 0.7842,
+      "step": 16621
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.071574402303218e-06,
+      "loss": 0.9111,
+      "step": 16622
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.069690474295918e-06,
+      "loss": 0.8926,
+      "step": 16623
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.067806932870382e-06,
+      "loss": 0.9189,
+      "step": 16624
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0659237780723095e-06,
+      "loss": 0.6631,
+      "step": 16625
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.064041009947413e-06,
+      "loss": 0.8652,
+      "step": 16626
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.062158628541375e-06,
+      "loss": 0.8604,
+      "step": 16627
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.060276633899875e-06,
+      "loss": 0.8066,
+      "step": 16628
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.05839502606858e-06,
+      "loss": 0.8906,
+      "step": 16629
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.056513805093161e-06,
+      "loss": 0.7524,
+      "step": 16630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.054632971019261e-06,
+      "loss": 0.9443,
+      "step": 16631
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.052752523892527e-06,
+      "loss": 0.7314,
+      "step": 16632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.050872463758586e-06,
+      "loss": 0.5728,
+      "step": 16633
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.048992790663068e-06,
+      "loss": 0.6543,
+      "step": 16634
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.047113504651587e-06,
+      "loss": 0.8232,
+      "step": 16635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.045234605769745e-06,
+      "loss": 1.0439,
+      "step": 16636
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.043356094063131e-06,
+      "loss": 0.8877,
+      "step": 16637
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0414779695773455e-06,
+      "loss": 0.8779,
+      "step": 16638
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.039600232357956e-06,
+      "loss": 0.9414,
+      "step": 16639
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.037722882450525e-06,
+      "loss": 0.8276,
+      "step": 16640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.035845919900623e-06,
+      "loss": 0.8301,
+      "step": 16641
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.033969344753792e-06,
+      "loss": 0.9268,
+      "step": 16642
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.03209315705557e-06,
+      "loss": 0.915,
+      "step": 16643
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0302173568514805e-06,
+      "loss": 0.9502,
+      "step": 16644
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.028341944187058e-06,
+      "loss": 0.875,
+      "step": 16645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.026466919107803e-06,
+      "loss": 0.917,
+      "step": 16646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.024592281659221e-06,
+      "loss": 0.8984,
+      "step": 16647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.022718031886796e-06,
+      "loss": 0.8994,
+      "step": 16648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0208441698360225e-06,
+      "loss": 0.9072,
+      "step": 16649
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.018970695552369e-06,
+      "loss": 0.876,
+      "step": 16650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0170976090812975e-06,
+      "loss": 1.0879,
+      "step": 16651
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0152249104682585e-06,
+      "loss": 0.9727,
+      "step": 16652
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.013352599758706e-06,
+      "loss": 0.8574,
+      "step": 16653
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.011480676998069e-06,
+      "loss": 0.8857,
+      "step": 16654
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.009609142231779e-06,
+      "loss": 0.9404,
+      "step": 16655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.007737995505241e-06,
+      "loss": 0.9619,
+      "step": 16656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.005867236863876e-06,
+      "loss": 0.9727,
+      "step": 16657
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.003996866353077e-06,
+      "loss": 0.8955,
+      "step": 16658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.002126884018231e-06,
+      "loss": 0.7666,
+      "step": 16659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.000257289904714e-06,
+      "loss": 0.8145,
+      "step": 16660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.998388084057902e-06,
+      "loss": 0.8555,
+      "step": 16661
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.996519266523153e-06,
+      "loss": 0.9844,
+      "step": 16662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.994650837345817e-06,
+      "loss": 0.8232,
+      "step": 16663
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.992782796571229e-06,
+      "loss": 0.6616,
+      "step": 16664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.990915144244733e-06,
+      "loss": 0.6572,
+      "step": 16665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9890478804116405e-06,
+      "loss": 0.9189,
+      "step": 16666
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9871810051172736e-06,
+      "loss": 0.9375,
+      "step": 16667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.985314518406933e-06,
+      "loss": 0.6997,
+      "step": 16668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.983448420325904e-06,
+      "loss": 0.5757,
+      "step": 16669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.981582710919487e-06,
+      "loss": 0.79,
+      "step": 16670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.979717390232947e-06,
+      "loss": 0.8711,
+      "step": 16671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.977852458311553e-06,
+      "loss": 1.0742,
+      "step": 16672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9759879152005546e-06,
+      "loss": 0.8682,
+      "step": 16673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9741237609452096e-06,
+      "loss": 0.9521,
+      "step": 16674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.972259995590753e-06,
+      "loss": 0.7822,
+      "step": 16675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.970396619182408e-06,
+      "loss": 0.9512,
+      "step": 16676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.96853363176539e-06,
+      "loss": 0.8389,
+      "step": 16677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.966671033384921e-06,
+      "loss": 0.9766,
+      "step": 16678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.964808824086193e-06,
+      "loss": 0.957,
+      "step": 16679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.96294700391439e-06,
+      "loss": 0.8906,
+      "step": 16680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9610855729147065e-06,
+      "loss": 0.7979,
+      "step": 16681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9592245311323085e-06,
+      "loss": 0.9648,
+      "step": 16682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.957363878612357e-06,
+      "loss": 0.8701,
+      "step": 16683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.955503615399998e-06,
+      "loss": 0.8721,
+      "step": 16684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.953643741540385e-06,
+      "loss": 0.9355,
+      "step": 16685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.95178425707865e-06,
+      "loss": 0.8359,
+      "step": 16686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.949925162059915e-06,
+      "loss": 0.7939,
+      "step": 16687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9480664565292895e-06,
+      "loss": 0.8154,
+      "step": 16688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.946208140531888e-06,
+      "loss": 0.7607,
+      "step": 16689
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.944350214112802e-06,
+      "loss": 0.8555,
+      "step": 16690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.94249267731712e-06,
+      "loss": 0.8623,
+      "step": 16691
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9406355301899115e-06,
+      "loss": 0.9834,
+      "step": 16692
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.938778772776255e-06,
+      "loss": 0.7471,
+      "step": 16693
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.936922405121204e-06,
+      "loss": 0.7207,
+      "step": 16694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.935066427269805e-06,
+      "loss": 0.9424,
+      "step": 16695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.933210839267094e-06,
+      "loss": 0.8301,
+      "step": 16696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.93135564115811e-06,
+      "loss": 0.7617,
+      "step": 16697
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.929500832987869e-06,
+      "loss": 0.9883,
+      "step": 16698
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.927646414801381e-06,
+      "loss": 0.7979,
+      "step": 16699
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.925792386643643e-06,
+      "loss": 0.6865,
+      "step": 16700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.923938748559654e-06,
+      "loss": 0.8877,
+      "step": 16701
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.922085500594397e-06,
+      "loss": 0.9629,
+      "step": 16702
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.920232642792839e-06,
+      "loss": 0.7686,
+      "step": 16703
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.918380175199941e-06,
+      "loss": 0.876,
+      "step": 16704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.916528097860668e-06,
+      "loss": 0.832,
+      "step": 16705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.914676410819957e-06,
+      "loss": 0.8262,
+      "step": 16706
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.912825114122745e-06,
+      "loss": 0.9619,
+      "step": 16707
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.910974207813951e-06,
+      "loss": 0.7256,
+      "step": 16708
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.909123691938503e-06,
+      "loss": 0.8252,
+      "step": 16709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9072735665413005e-06,
+      "loss": 0.8174,
+      "step": 16710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.905423831667243e-06,
+      "loss": 0.6919,
+      "step": 16711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.90357448736121e-06,
+      "loss": 0.8926,
+      "step": 16712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.901725533668093e-06,
+      "loss": 0.7305,
+      "step": 16713
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.899876970632752e-06,
+      "loss": 0.8906,
+      "step": 16714
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.898028798300049e-06,
+      "loss": 0.8184,
+      "step": 16715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.896181016714829e-06,
+      "loss": 0.9014,
+      "step": 16716
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.89433362592194e-06,
+      "loss": 0.7764,
+      "step": 16717
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.892486625966208e-06,
+      "loss": 0.8711,
+      "step": 16718
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.890640016892455e-06,
+      "loss": 0.7773,
+      "step": 16719
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8887937987454895e-06,
+      "loss": 0.707,
+      "step": 16720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8869479715701206e-06,
+      "loss": 0.7329,
+      "step": 16721
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.88510253541114e-06,
+      "loss": 0.9277,
+      "step": 16722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.883257490313328e-06,
+      "loss": 0.9346,
+      "step": 16723
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.881412836321452e-06,
+      "loss": 0.9238,
+      "step": 16724
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.87956857348029e-06,
+      "loss": 0.7764,
+      "step": 16725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.877724701834589e-06,
+      "loss": 0.8779,
+      "step": 16726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.875881221429092e-06,
+      "loss": 0.8789,
+      "step": 16727
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.874038132308544e-06,
+      "loss": 0.8008,
+      "step": 16728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.872195434517665e-06,
+      "loss": 1.0283,
+      "step": 16729
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.870353128101172e-06,
+      "loss": 0.8643,
+      "step": 16730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.868511213103767e-06,
+      "loss": 0.7646,
+      "step": 16731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.866669689570159e-06,
+      "loss": 0.8506,
+      "step": 16732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.864828557545035e-06,
+      "loss": 0.7617,
+      "step": 16733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.862987817073067e-06,
+      "loss": 0.8613,
+      "step": 16734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.861147468198922e-06,
+      "loss": 0.8809,
+      "step": 16735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.859307510967272e-06,
+      "loss": 0.8965,
+      "step": 16736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.857467945422759e-06,
+      "loss": 0.7871,
+      "step": 16737
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8556287716100275e-06,
+      "loss": 0.8311,
+      "step": 16738
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8537899895737e-06,
+      "loss": 0.8789,
+      "step": 16739
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.85195159935841e-06,
+      "loss": 0.7744,
+      "step": 16740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.850113601008767e-06,
+      "loss": 0.9854,
+      "step": 16741
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8482759945693705e-06,
+      "loss": 0.9023,
+      "step": 16742
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.846438780084809e-06,
+      "loss": 0.8486,
+      "step": 16743
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.844601957599679e-06,
+      "loss": 0.7715,
+      "step": 16744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8427655271585475e-06,
+      "loss": 0.8428,
+      "step": 16745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.84092948880598e-06,
+      "loss": 0.9492,
+      "step": 16746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.839093842586528e-06,
+      "loss": 0.9199,
+      "step": 16747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.837258588544744e-06,
+      "loss": 0.8345,
+      "step": 16748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.835423726725162e-06,
+      "loss": 0.8252,
+      "step": 16749
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.833589257172303e-06,
+      "loss": 0.8105,
+      "step": 16750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.831755179930694e-06,
+      "loss": 0.749,
+      "step": 16751
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.829921495044833e-06,
+      "loss": 0.7168,
+      "step": 16752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8280882025592256e-06,
+      "loss": 0.8779,
+      "step": 16753
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.82625530251836e-06,
+      "loss": 1.0195,
+      "step": 16754
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8244227949667116e-06,
+      "loss": 0.7676,
+      "step": 16755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8225906799487474e-06,
+      "loss": 0.9453,
+      "step": 16756
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.820758957508936e-06,
+      "loss": 0.9521,
+      "step": 16757
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.818927627691724e-06,
+      "loss": 0.8252,
+      "step": 16758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.81709669054155e-06,
+      "loss": 0.8877,
+      "step": 16759
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.815266146102843e-06,
+      "loss": 0.8545,
+      "step": 16760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.813435994420034e-06,
+      "loss": 0.79,
+      "step": 16761
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8116062355375304e-06,
+      "loss": 0.7139,
+      "step": 16762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.809776869499735e-06,
+      "loss": 0.9766,
+      "step": 16763
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8079478963510364e-06,
+      "loss": 0.7998,
+      "step": 16764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8061193161358277e-06,
+      "loss": 0.8818,
+      "step": 16765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8042911288984806e-06,
+      "loss": 0.9248,
+      "step": 16766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.802463334683351e-06,
+      "loss": 1.002,
+      "step": 16767
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8006359335348063e-06,
+      "loss": 0.8828,
+      "step": 16768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.798808925497186e-06,
+      "loss": 0.7969,
+      "step": 16769
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7969823106148297e-06,
+      "loss": 0.8018,
+      "step": 16770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7951560889320548e-06,
+      "loss": 0.9385,
+      "step": 16771
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7933302604931886e-06,
+      "loss": 0.9521,
+      "step": 16772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.791504825342538e-06,
+      "loss": 0.9258,
+      "step": 16773
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.789679783524396e-06,
+      "loss": 0.8809,
+      "step": 16774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.787855135083047e-06,
+      "loss": 0.8232,
+      "step": 16775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.786030880062781e-06,
+      "loss": 0.8115,
+      "step": 16776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.784207018507864e-06,
+      "loss": 0.8828,
+      "step": 16777
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.782383550462552e-06,
+      "loss": 0.8555,
+      "step": 16778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7805604759710912e-06,
+      "loss": 0.8604,
+      "step": 16779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.778737795077736e-06,
+      "loss": 0.7617,
+      "step": 16780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.776915507826706e-06,
+      "loss": 0.8242,
+      "step": 16781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7750936142622287e-06,
+      "loss": 0.8906,
+      "step": 16782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.773272114428508e-06,
+      "loss": 0.8701,
+      "step": 16783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7714510083697553e-06,
+      "loss": 0.9668,
+      "step": 16784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7696302961301624e-06,
+      "loss": 0.8926,
+      "step": 16785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.767809977753909e-06,
+      "loss": 0.7236,
+      "step": 16786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7659900532851666e-06,
+      "loss": 0.8467,
+      "step": 16787
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7641705227681067e-06,
+      "loss": 0.9883,
+      "step": 16788
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7623513862468807e-06,
+      "loss": 0.918,
+      "step": 16789
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7605326437656354e-06,
+      "loss": 0.8887,
+      "step": 16790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7587142953684974e-06,
+      "loss": 0.7676,
+      "step": 16791
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7568963410996028e-06,
+      "loss": 0.7754,
+      "step": 16792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7550787810030674e-06,
+      "loss": 0.8662,
+      "step": 16793
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7532616151229938e-06,
+      "loss": 0.7856,
+      "step": 16794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.751444843503478e-06,
+      "loss": 0.7676,
+      "step": 16795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.749628466188613e-06,
+      "loss": 0.8672,
+      "step": 16796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.747812483222477e-06,
+      "loss": 0.7266,
+      "step": 16797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7459968946491353e-06,
+      "loss": 0.7549,
+      "step": 16798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7441817005126415e-06,
+      "loss": 0.9189,
+      "step": 16799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7423669008570573e-06,
+      "loss": 0.7783,
+      "step": 16800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.740552495726415e-06,
+      "loss": 0.9243,
+      "step": 16801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.738738485164748e-06,
+      "loss": 0.8398,
+      "step": 16802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7369248692160697e-06,
+      "loss": 0.9023,
+      "step": 16803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.735111647924401e-06,
+      "loss": 0.8496,
+      "step": 16804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7332988213337417e-06,
+      "loss": 0.9414,
+      "step": 16805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7314863894880793e-06,
+      "loss": 0.8105,
+      "step": 16806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.729674352431394e-06,
+      "loss": 0.9834,
+      "step": 16807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7278627102076658e-06,
+      "loss": 0.834,
+      "step": 16808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.726051462860858e-06,
+      "loss": 0.8848,
+      "step": 16809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7242406104349192e-06,
+      "loss": 0.9072,
+      "step": 16810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7224301529737904e-06,
+      "loss": 0.8799,
+      "step": 16811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7206200905214185e-06,
+      "loss": 0.8066,
+      "step": 16812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.718810423121719e-06,
+      "loss": 1.1074,
+      "step": 16813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7170011508186043e-06,
+      "loss": 0.96,
+      "step": 16814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.715192273655992e-06,
+      "loss": 0.8721,
+      "step": 16815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.713383791677769e-06,
+      "loss": 0.835,
+      "step": 16816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7115757049278256e-06,
+      "loss": 0.6548,
+      "step": 16817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7097680134500324e-06,
+      "loss": 0.7295,
+      "step": 16818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.707960717288266e-06,
+      "loss": 0.7646,
+      "step": 16819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.70615381648638e-06,
+      "loss": 0.9033,
+      "step": 16820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.704347311088221e-06,
+      "loss": 0.8037,
+      "step": 16821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.702541201137626e-06,
+      "loss": 0.7783,
+      "step": 16822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.700735486678431e-06,
+      "loss": 0.7441,
+      "step": 16823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6989301677544532e-06,
+      "loss": 0.8789,
+      "step": 16824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6971252444094983e-06,
+      "loss": 0.6123,
+      "step": 16825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6953207166873628e-06,
+      "loss": 0.8486,
+      "step": 16826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.693516584631851e-06,
+      "loss": 1.002,
+      "step": 16827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.691712848286735e-06,
+      "loss": 0.834,
+      "step": 16828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6899095076957858e-06,
+      "loss": 0.7344,
+      "step": 16829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6881065629027624e-06,
+      "loss": 0.8271,
+      "step": 16830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.686304013951427e-06,
+      "loss": 0.7593,
+      "step": 16831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6845018608855166e-06,
+      "loss": 0.8799,
+      "step": 16832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6827001037487576e-06,
+      "loss": 0.8594,
+      "step": 16833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.680898742584884e-06,
+      "loss": 0.8701,
+      "step": 16834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6790977774376013e-06,
+      "loss": 0.9307,
+      "step": 16835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.677297208350623e-06,
+      "loss": 0.8545,
+      "step": 16836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6754970353676387e-06,
+      "loss": 0.835,
+      "step": 16837
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6736972585323315e-06,
+      "loss": 0.6646,
+      "step": 16838
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.671897877888373e-06,
+      "loss": 0.9033,
+      "step": 16839
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.670098893479439e-06,
+      "loss": 0.8906,
+      "step": 16840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6683003053491794e-06,
+      "loss": 0.9941,
+      "step": 16841
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6665021135412417e-06,
+      "loss": 0.8848,
+      "step": 16842
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.664704318099259e-06,
+      "loss": 0.874,
+      "step": 16843
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6629069190668664e-06,
+      "loss": 0.8369,
+      "step": 16844
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6611099164876775e-06,
+      "loss": 0.8154,
+      "step": 16845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659313310405299e-06,
+      "loss": 0.9307,
+      "step": 16846
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657517100863326e-06,
+      "loss": 0.7456,
+      "step": 16847
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655721287905356e-06,
+      "loss": 0.877,
+      "step": 16848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6539258715749636e-06,
+      "loss": 0.7744,
+      "step": 16849
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6521308519157184e-06,
+      "loss": 0.9844,
+      "step": 16850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6503362289711764e-06,
+      "loss": 0.8516,
+      "step": 16851
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6485420027848938e-06,
+      "loss": 0.7261,
+      "step": 16852
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.646748173400412e-06,
+      "loss": 0.8828,
+      "step": 16853
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644954740861251e-06,
+      "loss": 0.8887,
+      "step": 16854
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.643161705210947e-06,
+      "loss": 0.8447,
+      "step": 16855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6413690664930035e-06,
+      "loss": 0.7695,
+      "step": 16856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.639576824750923e-06,
+      "loss": 0.8193,
+      "step": 16857
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6377849800281963e-06,
+      "loss": 0.71,
+      "step": 16858
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6359935323683114e-06,
+      "loss": 0.9854,
+      "step": 16859
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6342024818147393e-06,
+      "loss": 0.8994,
+      "step": 16860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.632411828410942e-06,
+      "loss": 0.7646,
+      "step": 16861
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6306215722003702e-06,
+      "loss": 0.8857,
+      "step": 16862
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.628831713226477e-06,
+      "loss": 0.8594,
+      "step": 16863
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.627042251532693e-06,
+      "loss": 0.8857,
+      "step": 16864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.625253187162443e-06,
+      "loss": 0.7798,
+      "step": 16865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6234645201591344e-06,
+      "loss": 0.8848,
+      "step": 16866
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.621676250566186e-06,
+      "loss": 0.8125,
+      "step": 16867
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.61988837842699e-06,
+      "loss": 0.7646,
+      "step": 16868
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.618100903784929e-06,
+      "loss": 0.9482,
+      "step": 16869
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6163138266833777e-06,
+      "loss": 0.7979,
+      "step": 16870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6145271471657095e-06,
+      "loss": 0.8467,
+      "step": 16871
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6127408652752815e-06,
+      "loss": 0.7324,
+      "step": 16872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.610954981055441e-06,
+      "loss": 0.9316,
+      "step": 16873
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6091694945495183e-06,
+      "loss": 0.7197,
+      "step": 16874
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.607384405800853e-06,
+      "loss": 1.0107,
+      "step": 16875
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6055997148527587e-06,
+      "loss": 0.8535,
+      "step": 16876
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6038154217485468e-06,
+      "loss": 0.916,
+      "step": 16877
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6020315265315087e-06,
+      "loss": 0.8135,
+      "step": 16878
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.600248029244946e-06,
+      "loss": 0.9248,
+      "step": 16879
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5984649299321327e-06,
+      "loss": 0.9482,
+      "step": 16880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5966822286363434e-06,
+      "loss": 1.0059,
+      "step": 16881
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5948999254008276e-06,
+      "loss": 0.9404,
+      "step": 16882
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.593118020268851e-06,
+      "loss": 0.9102,
+      "step": 16883
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5913365132836498e-06,
+      "loss": 0.7549,
+      "step": 16884
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5895554044884563e-06,
+      "loss": 0.8203,
+      "step": 16885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5877746939264844e-06,
+      "loss": 0.8271,
+      "step": 16886
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5859943816409603e-06,
+      "loss": 0.8174,
+      "step": 16887
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5842144676750823e-06,
+      "loss": 0.7891,
+      "step": 16888
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.58243495207204e-06,
+      "loss": 0.9697,
+      "step": 16889
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.580655834875013e-06,
+      "loss": 0.8887,
+      "step": 16890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5788771161271885e-06,
+      "loss": 0.8525,
+      "step": 16891
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.577098795871723e-06,
+      "loss": 0.6807,
+      "step": 16892
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.575320874151771e-06,
+      "loss": 0.7847,
+      "step": 16893
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5735433510104757e-06,
+      "loss": 0.8438,
+      "step": 16894
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.571766226490976e-06,
+      "loss": 0.8525,
+      "step": 16895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5699895006363993e-06,
+      "loss": 0.8623,
+      "step": 16896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5682131734898517e-06,
+      "loss": 0.835,
+      "step": 16897
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.566437245094452e-06,
+      "loss": 0.9189,
+      "step": 16898
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5646617154932917e-06,
+      "loss": 0.8467,
+      "step": 16899
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.562886584729455e-06,
+      "loss": 0.9082,
+      "step": 16900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5611118528460177e-06,
+      "loss": 0.6699,
+      "step": 16901
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5593375198860547e-06,
+      "loss": 0.958,
+      "step": 16902
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5575635858926207e-06,
+      "loss": 0.8633,
+      "step": 16903
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.555790050908763e-06,
+      "loss": 0.9033,
+      "step": 16904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.554016914977514e-06,
+      "loss": 0.8916,
+      "step": 16905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.552244178141915e-06,
+      "loss": 0.668,
+      "step": 16906
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.550471840444978e-06,
+      "loss": 0.7505,
+      "step": 16907
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.548699901929713e-06,
+      "loss": 0.8525,
+      "step": 16908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5469283626391148e-06,
+      "loss": 0.6602,
+      "step": 16909
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5451572226161847e-06,
+      "loss": 0.9717,
+      "step": 16910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5433864819038945e-06,
+      "loss": 0.8486,
+      "step": 16911
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.541616140545219e-06,
+      "loss": 0.7148,
+      "step": 16912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5398461985831124e-06,
+      "loss": 0.8857,
+      "step": 16913
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.538076656060534e-06,
+      "loss": 0.7969,
+      "step": 16914
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.536307513020425e-06,
+      "loss": 0.9863,
+      "step": 16915
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.534538769505709e-06,
+      "loss": 0.7344,
+      "step": 16916
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5327704255593175e-06,
+      "loss": 0.916,
+      "step": 16917
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.531002481224155e-06,
+      "loss": 0.7832,
+      "step": 16918
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5292349365431354e-06,
+      "loss": 1.0059,
+      "step": 16919
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.527467791559143e-06,
+      "loss": 0.8467,
+      "step": 16920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5257010463150642e-06,
+      "loss": 0.834,
+      "step": 16921
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.523934700853766e-06,
+      "loss": 0.8623,
+      "step": 16922
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.522168755218125e-06,
+      "loss": 0.7734,
+      "step": 16923
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5204032094509886e-06,
+      "loss": 0.8896,
+      "step": 16924
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.518638063595199e-06,
+      "loss": 0.9277,
+      "step": 16925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.516873317693592e-06,
+      "loss": 0.9238,
+      "step": 16926
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5151089717889964e-06,
+      "loss": 0.9844,
+      "step": 16927
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5133450259242286e-06,
+      "loss": 0.8345,
+      "step": 16928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.51158148014209e-06,
+      "loss": 0.8477,
+      "step": 16929
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5098183344853732e-06,
+      "loss": 0.7324,
+      "step": 16930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5080555889968724e-06,
+      "loss": 0.792,
+      "step": 16931
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.506293243719365e-06,
+      "loss": 0.9629,
+      "step": 16932
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5045312986956126e-06,
+      "loss": 0.7036,
+      "step": 16933
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.502769753968369e-06,
+      "loss": 0.9717,
+      "step": 16934
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.501008609580392e-06,
+      "loss": 0.6592,
+      "step": 16935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4992478655744156e-06,
+      "loss": 0.8271,
+      "step": 16936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.497487521993159e-06,
+      "loss": 0.8291,
+      "step": 16937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4957275788793554e-06,
+      "loss": 0.6787,
+      "step": 16938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4939680362757054e-06,
+      "loss": 0.7771,
+      "step": 16939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.492208894224909e-06,
+      "loss": 0.8711,
+      "step": 16940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4904501527696488e-06,
+      "loss": 0.8564,
+      "step": 16941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.488691811952618e-06,
+      "loss": 0.915,
+      "step": 16942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.486933871816476e-06,
+      "loss": 0.9238,
+      "step": 16943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.485176332403888e-06,
+      "loss": 0.7822,
+      "step": 16944
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4834191937574957e-06,
+      "loss": 0.7754,
+      "step": 16945
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.481662455919952e-06,
+      "loss": 0.8008,
+      "step": 16946
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.47990611893388e-06,
+      "loss": 0.9492,
+      "step": 16947
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.478150182841904e-06,
+      "loss": 0.7803,
+      "step": 16948
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4763946476866295e-06,
+      "loss": 0.7734,
+      "step": 16949
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.474639513510667e-06,
+      "loss": 0.8779,
+      "step": 16950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4728847803566048e-06,
+      "loss": 1.0293,
+      "step": 16951
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4711304482670238e-06,
+      "loss": 0.793,
+      "step": 16952
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.469376517284493e-06,
+      "loss": 0.7988,
+      "step": 16953
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4676229874515843e-06,
+      "loss": 0.793,
+      "step": 16954
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.465869858810844e-06,
+      "loss": 0.8477,
+      "step": 16955
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4641171314048182e-06,
+      "loss": 0.8076,
+      "step": 16956
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.462364805276035e-06,
+      "loss": 0.874,
+      "step": 16957
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.460612880467027e-06,
+      "loss": 0.9043,
+      "step": 16958
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.458861357020302e-06,
+      "loss": 0.7871,
+      "step": 16959
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.457110234978369e-06,
+      "loss": 0.9199,
+      "step": 16960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.455359514383714e-06,
+      "loss": 0.8672,
+      "step": 16961
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.453609195278831e-06,
+      "loss": 0.957,
+      "step": 16962
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4518592777061933e-06,
+      "loss": 0.9326,
+      "step": 16963
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4501097617082648e-06,
+      "loss": 0.8281,
+      "step": 16964
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4483606473274955e-06,
+      "loss": 0.8613,
+      "step": 16965
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.446611934606341e-06,
+      "loss": 0.7637,
+      "step": 16966
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4448636235872337e-06,
+      "loss": 0.8887,
+      "step": 16967
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4431157143125972e-06,
+      "loss": 0.873,
+      "step": 16968
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.441368206824849e-06,
+      "loss": 0.8848,
+      "step": 16969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4396211011663995e-06,
+      "loss": 0.8203,
+      "step": 16970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4378743973796435e-06,
+      "loss": 0.8135,
+      "step": 16971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.43612809550697e-06,
+      "loss": 0.9941,
+      "step": 16972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4343821955907487e-06,
+      "loss": 0.75,
+      "step": 16973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4326366976733594e-06,
+      "loss": 0.8652,
+      "step": 16974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.430891601797155e-06,
+      "loss": 0.7461,
+      "step": 16975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4291469080044836e-06,
+      "loss": 0.8857,
+      "step": 16976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4274026163376783e-06,
+      "loss": 0.8193,
+      "step": 16977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4256587268390785e-06,
+      "loss": 1.0078,
+      "step": 16978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.423915239550997e-06,
+      "loss": 0.8535,
+      "step": 16979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.422172154515744e-06,
+      "loss": 0.9336,
+      "step": 16980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4204294717756147e-06,
+      "loss": 0.9609,
+      "step": 16981
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.418687191372909e-06,
+      "loss": 0.7451,
+      "step": 16982
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4169453133498998e-06,
+      "loss": 0.8984,
+      "step": 16983
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.415203837748855e-06,
+      "loss": 0.9209,
+      "step": 16984
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.413462764612043e-06,
+      "loss": 0.877,
+      "step": 16985
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.411722093981713e-06,
+      "loss": 0.7891,
+      "step": 16986
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4099818259001017e-06,
+      "loss": 0.7402,
+      "step": 16987
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4082419604094354e-06,
+      "loss": 0.8613,
+      "step": 16988
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.406502497551949e-06,
+      "loss": 0.9043,
+      "step": 16989
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.404763437369847e-06,
+      "loss": 0.9375,
+      "step": 16990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4030247799053305e-06,
+      "loss": 0.7539,
+      "step": 16991
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.401286525200589e-06,
+      "loss": 0.8623,
+      "step": 16992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3995486732978143e-06,
+      "loss": 0.9648,
+      "step": 16993
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3978112242391716e-06,
+      "loss": 0.8076,
+      "step": 16994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.396074178066826e-06,
+      "loss": 0.9697,
+      "step": 16995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3943375348229224e-06,
+      "loss": 0.8188,
+      "step": 16996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.392601294549618e-06,
+      "loss": 0.7871,
+      "step": 16997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.390865457289041e-06,
+      "loss": 0.9717,
+      "step": 16998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3891300230833112e-06,
+      "loss": 0.8857,
+      "step": 16999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.387394991974542e-06,
+      "loss": 0.8525,
+      "step": 17000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.385660364004839e-06,
+      "loss": 0.7969,
+      "step": 17001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3839261392163025e-06,
+      "loss": 0.708,
+      "step": 17002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3821923176510142e-06,
+      "loss": 0.9336,
+      "step": 17003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3804588993510466e-06,
+      "loss": 0.6318,
+      "step": 17004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3787258843584603e-06,
+      "loss": 0.8037,
+      "step": 17005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3769932727153187e-06,
+      "loss": 0.834,
+      "step": 17006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.375261064463666e-06,
+      "loss": 0.8047,
+      "step": 17007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.373529259645534e-06,
+      "loss": 0.9639,
+      "step": 17008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3717978583029474e-06,
+      "loss": 0.6113,
+      "step": 17009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3700668604779275e-06,
+      "loss": 0.9443,
+      "step": 17010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3683362662124795e-06,
+      "loss": 0.8711,
+      "step": 17011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3666060755485976e-06,
+      "loss": 0.8916,
+      "step": 17012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.364876288528265e-06,
+      "loss": 0.8643,
+      "step": 17013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3631469051934664e-06,
+      "loss": 0.8633,
+      "step": 17014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.361417925586168e-06,
+      "loss": 0.8359,
+      "step": 17015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3596893497483206e-06,
+      "loss": 0.8438,
+      "step": 17016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3579611777218736e-06,
+      "loss": 0.917,
+      "step": 17017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.356233409548768e-06,
+      "loss": 0.8574,
+      "step": 17018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3545060452709333e-06,
+      "loss": 0.8848,
+      "step": 17019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.352779084930282e-06,
+      "loss": 0.7812,
+      "step": 17020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3510525285687214e-06,
+      "loss": 0.8174,
+      "step": 17021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.349326376228155e-06,
+      "loss": 0.9521,
+      "step": 17022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3476006279504715e-06,
+      "loss": 0.9531,
+      "step": 17023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3458752837775443e-06,
+      "loss": 0.832,
+      "step": 17024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.344150343751249e-06,
+      "loss": 0.8164,
+      "step": 17025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.342425807913443e-06,
+      "loss": 0.7432,
+      "step": 17026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.340701676305973e-06,
+      "loss": 0.6133,
+      "step": 17027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3389779489706762e-06,
+      "loss": 0.8096,
+      "step": 17028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.337254625949391e-06,
+      "loss": 0.7568,
+      "step": 17029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3355317072839322e-06,
+      "loss": 0.7529,
+      "step": 17030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.333809193016111e-06,
+      "loss": 0.9609,
+      "step": 17031
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.332087083187723e-06,
+      "loss": 0.752,
+      "step": 17032
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3303653778405654e-06,
+      "loss": 0.7285,
+      "step": 17033
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3286440770164184e-06,
+      "loss": 0.8164,
+      "step": 17034
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.326923180757051e-06,
+      "loss": 0.9756,
+      "step": 17035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3252026891042188e-06,
+      "loss": 0.9199,
+      "step": 17036
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3234826020996837e-06,
+      "loss": 0.8037,
+      "step": 17037
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.321762919785181e-06,
+      "loss": 0.8223,
+      "step": 17038
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.320043642202444e-06,
+      "loss": 0.9336,
+      "step": 17039
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3183247693931885e-06,
+      "loss": 0.8799,
+      "step": 17040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3166063013991368e-06,
+      "loss": 0.8467,
+      "step": 17041
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3148882382619883e-06,
+      "loss": 0.9453,
+      "step": 17042
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3131705800234305e-06,
+      "loss": 0.876,
+      "step": 17043
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3114533267251448e-06,
+      "loss": 0.7876,
+      "step": 17044
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3097364784088114e-06,
+      "loss": 0.9238,
+      "step": 17045
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.308020035116091e-06,
+      "loss": 0.9248,
+      "step": 17046
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3063039968886334e-06,
+      "loss": 0.8262,
+      "step": 17047
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3045883637680796e-06,
+      "loss": 0.9004,
+      "step": 17048
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3028731357960698e-06,
+      "loss": 0.876,
+      "step": 17049
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3011583130142277e-06,
+      "loss": 0.833,
+      "step": 17050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2994438954641607e-06,
+      "loss": 0.9209,
+      "step": 17051
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.297729883187473e-06,
+      "loss": 0.9189,
+      "step": 17052
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.296016276225764e-06,
+      "loss": 0.9512,
+      "step": 17053
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.294303074620617e-06,
+      "loss": 0.9053,
+      "step": 17054
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2925902784136055e-06,
+      "loss": 0.8857,
+      "step": 17055
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2908778876462866e-06,
+      "loss": 0.8516,
+      "step": 17056
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2891659023602274e-06,
+      "loss": 1.0156,
+      "step": 17057
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.287454322596968e-06,
+      "loss": 1.0293,
+      "step": 17058
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.285743148398042e-06,
+      "loss": 0.9932,
+      "step": 17059
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.284032379804969e-06,
+      "loss": 0.7373,
+      "step": 17060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.282322016859274e-06,
+      "loss": 0.9453,
+      "step": 17061
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2806120596024617e-06,
+      "loss": 0.8691,
+      "step": 17062
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2789025080760227e-06,
+      "loss": 0.7168,
+      "step": 17063
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.27719336232144e-06,
+      "loss": 0.6553,
+      "step": 17064
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2754846223801985e-06,
+      "loss": 0.8076,
+      "step": 17065
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.273776288293762e-06,
+      "loss": 0.7754,
+      "step": 17066
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.272068360103584e-06,
+      "loss": 0.8623,
+      "step": 17067
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.270360837851108e-06,
+      "loss": 0.7578,
+      "step": 17068
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2686537215777793e-06,
+      "loss": 0.9121,
+      "step": 17069
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2669470113250192e-06,
+      "loss": 0.708,
+      "step": 17070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2652407071342383e-06,
+      "loss": 0.9688,
+      "step": 17071
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2635348090468576e-06,
+      "loss": 0.834,
+      "step": 17072
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2618293171042657e-06,
+      "loss": 0.8213,
+      "step": 17073
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.26012423134785e-06,
+      "loss": 0.9707,
+      "step": 17074
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2584195518189855e-06,
+      "loss": 0.8125,
+      "step": 17075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2567152785590483e-06,
+      "loss": 0.8652,
+      "step": 17076
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.255011411609392e-06,
+      "loss": 0.8838,
+      "step": 17077
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.253307951011362e-06,
+      "loss": 0.793,
+      "step": 17078
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.251604896806293e-06,
+      "loss": 0.8525,
+      "step": 17079
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2499022490355237e-06,
+      "loss": 0.7422,
+      "step": 17080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2482000077403655e-06,
+      "loss": 0.9189,
+      "step": 17081
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.246498172962127e-06,
+      "loss": 0.9443,
+      "step": 17082
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2447967447421045e-06,
+      "loss": 0.8496,
+      "step": 17083
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2430957231215896e-06,
+      "loss": 0.7134,
+      "step": 17084
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.241395108141865e-06,
+      "loss": 0.8394,
+      "step": 17085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.239694899844197e-06,
+      "loss": 0.9717,
+      "step": 17086
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.237995098269844e-06,
+      "loss": 0.9658,
+      "step": 17087
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2362957034600486e-06,
+      "loss": 0.7344,
+      "step": 17088
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.23459671545606e-06,
+      "loss": 0.8359,
+      "step": 17089
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.232898134299105e-06,
+      "loss": 0.7773,
+      "step": 17090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.231199960030402e-06,
+      "loss": 0.8838,
+      "step": 17091
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.229502192691156e-06,
+      "loss": 0.8076,
+      "step": 17092
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2278048323225764e-06,
+      "loss": 0.8975,
+      "step": 17093
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2261078789658474e-06,
+      "loss": 0.9355,
+      "step": 17094
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.224411332662152e-06,
+      "loss": 0.8115,
+      "step": 17095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2227151934526523e-06,
+      "loss": 0.8135,
+      "step": 17096
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.221019461378518e-06,
+      "loss": 0.644,
+      "step": 17097
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2193241364808993e-06,
+      "loss": 0.834,
+      "step": 17098
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2176292188009304e-06,
+      "loss": 0.916,
+      "step": 17099
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.215934708379742e-06,
+      "loss": 0.8867,
+      "step": 17100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.214240605258463e-06,
+      "loss": 0.7568,
+      "step": 17101
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.212546909478198e-06,
+      "loss": 0.7412,
+      "step": 17102
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.210853621080052e-06,
+      "loss": 0.7861,
+      "step": 17103
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.209160740105106e-06,
+      "loss": 0.835,
+      "step": 17104
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.207468266594456e-06,
+      "loss": 0.8062,
+      "step": 17105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2057762005891634e-06,
+      "loss": 0.9473,
+      "step": 17106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.204084542130295e-06,
+      "loss": 0.8091,
+      "step": 17107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.202393291258894e-06,
+      "loss": 0.873,
+      "step": 17108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2007024480160107e-06,
+      "loss": 0.6104,
+      "step": 17109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.199012012442677e-06,
+      "loss": 0.9043,
+      "step": 17110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.197321984579904e-06,
+      "loss": 0.8242,
+      "step": 17111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1956323644687193e-06,
+      "loss": 0.9082,
+      "step": 17112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.193943152150116e-06,
+      "loss": 0.9189,
+      "step": 17113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.192254347665087e-06,
+      "loss": 0.8018,
+      "step": 17114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1905659510546114e-06,
+      "loss": 0.7021,
+      "step": 17115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.188877962359669e-06,
+      "loss": 0.8818,
+      "step": 17116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.187190381621219e-06,
+      "loss": 0.9053,
+      "step": 17117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1855032088802142e-06,
+      "loss": 0.7969,
+      "step": 17118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.183816444177592e-06,
+      "loss": 0.793,
+      "step": 17119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.182130087554296e-06,
+      "loss": 0.7373,
+      "step": 17120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.180444139051242e-06,
+      "loss": 0.8828,
+      "step": 17121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.178758598709344e-06,
+      "loss": 0.9434,
+      "step": 17122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1770734665695e-06,
+      "loss": 0.875,
+      "step": 17123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.175388742672616e-06,
+      "loss": 0.7617,
+      "step": 17124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1737044270595653e-06,
+      "loss": 0.8096,
+      "step": 17125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.172020519771224e-06,
+      "loss": 0.7881,
+      "step": 17126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.170337020848451e-06,
+      "loss": 0.7373,
+      "step": 17127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.168653930332111e-06,
+      "loss": 0.7881,
+      "step": 17128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.166971248263038e-06,
+      "loss": 0.791,
+      "step": 17129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1652889746820703e-06,
+      "loss": 0.874,
+      "step": 17130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1636071096300247e-06,
+      "loss": 0.8809,
+      "step": 17131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1619256531477238e-06,
+      "loss": 0.9131,
+      "step": 17132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1602446052759707e-06,
+      "loss": 0.8906,
+      "step": 17133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1585639660555565e-06,
+      "loss": 0.8643,
+      "step": 17134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.156883735527261e-06,
+      "loss": 0.8438,
+      "step": 17135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.155203913731868e-06,
+      "loss": 0.7871,
+      "step": 17136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1535245007101366e-06,
+      "loss": 0.8857,
+      "step": 17137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.151845496502821e-06,
+      "loss": 0.8008,
+      "step": 17138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1501669011506643e-06,
+      "loss": 0.9785,
+      "step": 17139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1484887146944064e-06,
+      "loss": 0.8701,
+      "step": 17140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.146810937174771e-06,
+      "loss": 0.6611,
+      "step": 17141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.145133568632468e-06,
+      "loss": 0.75,
+      "step": 17142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1434566091082018e-06,
+      "loss": 0.7432,
+      "step": 17143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.141780058642676e-06,
+      "loss": 0.8818,
+      "step": 17144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1401039172765668e-06,
+      "loss": 0.7539,
+      "step": 17145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.138428185050555e-06,
+      "loss": 0.959,
+      "step": 17146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.136752862005299e-06,
+      "loss": 0.9336,
+      "step": 17147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1350779481814598e-06,
+      "loss": 0.9375,
+      "step": 17148
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1334034436196824e-06,
+      "loss": 0.8193,
+      "step": 17149
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1317293483606016e-06,
+      "loss": 0.9326,
+      "step": 17150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1300556624448352e-06,
+      "loss": 0.9648,
+      "step": 17151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1283823859130112e-06,
+      "loss": 0.8545,
+      "step": 17152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1267095188057283e-06,
+      "loss": 0.8584,
+      "step": 17153
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1250370611635784e-06,
+      "loss": 0.6836,
+      "step": 17154
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1233650130271555e-06,
+      "loss": 0.8262,
+      "step": 17155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1216933744370313e-06,
+      "loss": 0.8799,
+      "step": 17156
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.120022145433772e-06,
+      "loss": 0.9131,
+      "step": 17157
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.118351326057929e-06,
+      "loss": 0.8027,
+      "step": 17158
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1166809163500567e-06,
+      "loss": 0.9688,
+      "step": 17159
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.115010916350687e-06,
+      "loss": 0.8145,
+      "step": 17160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.113341326100343e-06,
+      "loss": 0.9141,
+      "step": 17161
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1116721456395415e-06,
+      "loss": 0.7373,
+      "step": 17162
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.110003375008792e-06,
+      "loss": 0.9512,
+      "step": 17163
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.108335014248591e-06,
+      "loss": 0.7441,
+      "step": 17164
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.106667063399422e-06,
+      "loss": 0.8877,
+      "step": 17165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.104999522501757e-06,
+      "loss": 0.7451,
+      "step": 17166
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.103332391596068e-06,
+      "loss": 0.916,
+      "step": 17167
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.101665670722813e-06,
+      "loss": 0.6777,
+      "step": 17168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0999993599224386e-06,
+      "loss": 0.8916,
+      "step": 17169
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0983334592353765e-06,
+      "loss": 0.8213,
+      "step": 17170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0966679687020516e-06,
+      "loss": 0.8154,
+      "step": 17171
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.09500288836289e-06,
+      "loss": 0.9854,
+      "step": 17172
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0933382182582904e-06,
+      "loss": 0.8662,
+      "step": 17173
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.091673958428654e-06,
+      "loss": 0.8955,
+      "step": 17174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0900101089143584e-06,
+      "loss": 0.8711,
+      "step": 17175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.088346669755793e-06,
+      "loss": 0.9961,
+      "step": 17176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0866836409933176e-06,
+      "loss": 0.6597,
+      "step": 17177
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0850210226672895e-06,
+      "loss": 0.8643,
+      "step": 17178
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0833588148180516e-06,
+      "loss": 0.7939,
+      "step": 17179
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0816970174859493e-06,
+      "loss": 0.875,
+      "step": 17180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0800356307113065e-06,
+      "loss": 0.9131,
+      "step": 17181
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0783746545344395e-06,
+      "loss": 0.7637,
+      "step": 17182
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.076714088995649e-06,
+      "loss": 0.9268,
+      "step": 17183
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0750539341352436e-06,
+      "loss": 0.8867,
+      "step": 17184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0733941899935037e-06,
+      "loss": 0.8203,
+      "step": 17185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0717348566107086e-06,
+      "loss": 0.7842,
+      "step": 17186
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0700759340271192e-06,
+      "loss": 0.749,
+      "step": 17187
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.068417422283003e-06,
+      "loss": 0.8086,
+      "step": 17188
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.066759321418602e-06,
+      "loss": 0.877,
+      "step": 17189
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0651016314741523e-06,
+      "loss": 0.8877,
+      "step": 17190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0634443524898795e-06,
+      "loss": 0.917,
+      "step": 17191
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0617874845060048e-06,
+      "loss": 0.8428,
+      "step": 17192
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0601310275627383e-06,
+      "loss": 0.8213,
+      "step": 17193
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0584749817002656e-06,
+      "loss": 0.8232,
+      "step": 17194
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.056819346958788e-06,
+      "loss": 0.916,
+      "step": 17195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.055164123378478e-06,
+      "loss": 0.8525,
+      "step": 17196
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0535093109994986e-06,
+      "loss": 0.7559,
+      "step": 17197
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0518549098620087e-06,
+      "loss": 0.707,
+      "step": 17198
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.050200920006161e-06,
+      "loss": 0.9346,
+      "step": 17199
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0485473414720877e-06,
+      "loss": 0.8633,
+      "step": 17200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0468941742999215e-06,
+      "loss": 0.9141,
+      "step": 17201
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0452414185297697e-06,
+      "loss": 0.5664,
+      "step": 17202
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.043589074201749e-06,
+      "loss": 0.8838,
+      "step": 17203
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.041937141355957e-06,
+      "loss": 0.9541,
+      "step": 17204
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0402856200324793e-06,
+      "loss": 0.8154,
+      "step": 17205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0386345102713877e-06,
+      "loss": 0.8491,
+      "step": 17206
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.036983812112759e-06,
+      "loss": 0.7422,
+      "step": 17207
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0353335255966466e-06,
+      "loss": 1.0283,
+      "step": 17208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0336836507630997e-06,
+      "loss": 0.8389,
+      "step": 17209
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0320341876521484e-06,
+      "loss": 0.835,
+      "step": 17210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0303851363038304e-06,
+      "loss": 0.9971,
+      "step": 17211
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0287364967581624e-06,
+      "loss": 0.7666,
+      "step": 17212
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.027088269055145e-06,
+      "loss": 0.8867,
+      "step": 17213
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.025440453234778e-06,
+      "loss": 0.9121,
+      "step": 17214
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0237930493370537e-06,
+      "loss": 0.8184,
+      "step": 17215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.022146057401949e-06,
+      "loss": 0.6777,
+      "step": 17216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0204994774694276e-06,
+      "loss": 0.8323,
+      "step": 17217
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0188533095794437e-06,
+      "loss": 0.7715,
+      "step": 17218
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.017207553771957e-06,
+      "loss": 0.832,
+      "step": 17219
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0155622100868953e-06,
+      "loss": 0.7324,
+      "step": 17220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0139172785641937e-06,
+      "loss": 0.7158,
+      "step": 17221
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0122727592437574e-06,
+      "loss": 0.9414,
+      "step": 17222
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0106286521655083e-06,
+      "loss": 0.7695,
+      "step": 17223
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0089849573693363e-06,
+      "loss": 0.7939,
+      "step": 17224
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0073416748951323e-06,
+      "loss": 0.8672,
+      "step": 17225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.005698804782766e-06,
+      "loss": 0.9473,
+      "step": 17226
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0040563470721176e-06,
+      "loss": 0.9199,
+      "step": 17227
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0024143018030384e-06,
+      "loss": 0.9834,
+      "step": 17228
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.000772669015375e-06,
+      "loss": 0.8379,
+      "step": 17229
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9991314487489596e-06,
+      "loss": 0.7949,
+      "step": 17230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9974906410436344e-06,
+      "loss": 0.8506,
+      "step": 17231
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.995850245939207e-06,
+      "loss": 0.9365,
+      "step": 17232
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9942102634754855e-06,
+      "loss": 0.8994,
+      "step": 17233
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.992570693692267e-06,
+      "loss": 0.9854,
+      "step": 17234
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.990931536629342e-06,
+      "loss": 0.8066,
+      "step": 17235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.98929279232649e-06,
+      "loss": 0.9268,
+      "step": 17236
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9876544608234725e-06,
+      "loss": 0.8398,
+      "step": 17237
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9860165421600464e-06,
+      "loss": 0.9668,
+      "step": 17238
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9843790363759682e-06,
+      "loss": 0.9775,
+      "step": 17239
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.982741943510967e-06,
+      "loss": 0.8779,
+      "step": 17240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9811052636047712e-06,
+      "loss": 0.9082,
+      "step": 17241
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9794689966971037e-06,
+      "loss": 0.793,
+      "step": 17242
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9778331428276662e-06,
+      "loss": 0.8975,
+      "step": 17243
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.976197702036159e-06,
+      "loss": 0.9551,
+      "step": 17244
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9745626743622648e-06,
+      "loss": 0.8643,
+      "step": 17245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9729280598456677e-06,
+      "loss": 0.8887,
+      "step": 17246
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9712938585260322e-06,
+      "loss": 0.9326,
+      "step": 17247
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9696600704430147e-06,
+      "loss": 0.958,
+      "step": 17248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.968026695636259e-06,
+      "loss": 0.5571,
+      "step": 17249
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9663937341454076e-06,
+      "loss": 0.8555,
+      "step": 17250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9647611860100877e-06,
+      "loss": 0.9287,
+      "step": 17251
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9631290512699173e-06,
+      "loss": 0.8115,
+      "step": 17252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9614973299645e-06,
+      "loss": 0.8721,
+      "step": 17253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9598660221334285e-06,
+      "loss": 0.8652,
+      "step": 17254
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9582351278163025e-06,
+      "loss": 0.9287,
+      "step": 17255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.95660464705269e-06,
+      "loss": 0.75,
+      "step": 17256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.954974579882162e-06,
+      "loss": 0.7617,
+      "step": 17257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.953344926344266e-06,
+      "loss": 0.9258,
+      "step": 17258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.951715686478562e-06,
+      "loss": 0.8154,
+      "step": 17259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9500868603245815e-06,
+      "loss": 0.6274,
+      "step": 17260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9484584479218513e-06,
+      "loss": 0.8701,
+      "step": 17261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9468304493098833e-06,
+      "loss": 0.8301,
+      "step": 17262
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9452028645281915e-06,
+      "loss": 0.8438,
+      "step": 17263
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.943575693616272e-06,
+      "loss": 0.8008,
+      "step": 17264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.941948936613608e-06,
+      "loss": 0.8516,
+      "step": 17265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9403225935596725e-06,
+      "loss": 0.8164,
+      "step": 17266
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9386966644939406e-06,
+      "loss": 0.7495,
+      "step": 17267
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.937071149455868e-06,
+      "loss": 0.8027,
+      "step": 17268
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9354460484848955e-06,
+      "loss": 0.7656,
+      "step": 17269
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9338213616204614e-06,
+      "loss": 0.9297,
+      "step": 17270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9321970889019936e-06,
+      "loss": 0.9492,
+      "step": 17271
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9305732303689096e-06,
+      "loss": 0.8398,
+      "step": 17272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9289497860606154e-06,
+      "loss": 0.9385,
+      "step": 17273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9273267560164998e-06,
+      "loss": 0.9736,
+      "step": 17274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.925704140275958e-06,
+      "loss": 0.9053,
+      "step": 17275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9240819388783647e-06,
+      "loss": 0.8281,
+      "step": 17276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9224601518630846e-06,
+      "loss": 0.8408,
+      "step": 17277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.920838779269468e-06,
+      "loss": 0.7432,
+      "step": 17278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9192178211368705e-06,
+      "loss": 0.793,
+      "step": 17279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9175972775046245e-06,
+      "loss": 0.7476,
+      "step": 17280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9159771484120504e-06,
+      "loss": 0.96,
+      "step": 17281
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9143574338984737e-06,
+      "loss": 0.7305,
+      "step": 17282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.912738134003192e-06,
+      "loss": 0.8486,
+      "step": 17283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.911119248765508e-06,
+      "loss": 0.8398,
+      "step": 17284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9095007782246966e-06,
+      "loss": 0.7588,
+      "step": 17285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9078827224200457e-06,
+      "loss": 0.8203,
+      "step": 17286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9062650813908133e-06,
+      "loss": 0.9473,
+      "step": 17287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9046478551762567e-06,
+      "loss": 0.8145,
+      "step": 17288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9030310438156182e-06,
+      "loss": 0.8486,
+      "step": 17289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9014146473481396e-06,
+      "loss": 0.8672,
+      "step": 17290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8997986658130405e-06,
+      "loss": 0.7705,
+      "step": 17291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.898183099249541e-06,
+      "loss": 0.7593,
+      "step": 17292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8965679476968356e-06,
+      "loss": 0.7578,
+      "step": 17293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8949532111941336e-06,
+      "loss": 0.9277,
+      "step": 17294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8933388897806127e-06,
+      "loss": 0.9688,
+      "step": 17295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8917249834954477e-06,
+      "loss": 0.9756,
+      "step": 17296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8901114923777986e-06,
+      "loss": 0.7441,
+      "step": 17297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8884984164668296e-06,
+      "loss": 0.8223,
+      "step": 17298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8868857558016806e-06,
+      "loss": 0.8691,
+      "step": 17299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.885273510421487e-06,
+      "loss": 0.8169,
+      "step": 17300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8836616803653684e-06,
+      "loss": 0.8477,
+      "step": 17301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.882050265672449e-06,
+      "loss": 1.001,
+      "step": 17302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.880439266381827e-06,
+      "loss": 0.7822,
+      "step": 17303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8788286825325973e-06,
+      "loss": 0.8193,
+      "step": 17304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8772185141638375e-06,
+      "loss": 1.0293,
+      "step": 17305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8756087613146346e-06,
+      "loss": 0.8975,
+      "step": 17306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8739994240240455e-06,
+      "loss": 0.9111,
+      "step": 17307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8723905023311237e-06,
+      "loss": 1.0078,
+      "step": 17308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8707819962749118e-06,
+      "loss": 0.8281,
+      "step": 17309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.869173905894447e-06,
+      "loss": 0.8726,
+      "step": 17310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8675662312287534e-06,
+      "loss": 0.7163,
+      "step": 17311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8659589723168424e-06,
+      "loss": 0.8936,
+      "step": 17312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.864352129197714e-06,
+      "loss": 0.8193,
+      "step": 17313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8627457019103675e-06,
+      "loss": 0.752,
+      "step": 17314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.861139690493786e-06,
+      "loss": 0.9561,
+      "step": 17315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.85953409498694e-06,
+      "loss": 0.8389,
+      "step": 17316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8579289154287894e-06,
+      "loss": 0.8115,
+      "step": 17317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.856324151858296e-06,
+      "loss": 0.9043,
+      "step": 17318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8547198043143964e-06,
+      "loss": 0.8721,
+      "step": 17319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.853115872836025e-06,
+      "loss": 0.958,
+      "step": 17320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8515123574620984e-06,
+      "loss": 0.8545,
+      "step": 17321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8499092582315423e-06,
+      "loss": 0.9326,
+      "step": 17322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8483065751832506e-06,
+      "loss": 0.9775,
+      "step": 17323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8467043083561165e-06,
+      "loss": 0.7207,
+      "step": 17324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8451024577890175e-06,
+      "loss": 0.8779,
+      "step": 17325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8435010235208384e-06,
+      "loss": 0.7153,
+      "step": 17326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.841900005590432e-06,
+      "loss": 0.8906,
+      "step": 17327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8402994040366483e-06,
+      "loss": 0.8848,
+      "step": 17328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8386992188983377e-06,
+      "loss": 0.7734,
+      "step": 17329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8370994502143267e-06,
+      "loss": 0.8994,
+      "step": 17330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.835500098023438e-06,
+      "loss": 0.8525,
+      "step": 17331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8339011623644786e-06,
+      "loss": 0.8652,
+      "step": 17332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8323026432762546e-06,
+      "loss": 0.9707,
+      "step": 17333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.830704540797562e-06,
+      "loss": 0.6973,
+      "step": 17334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8291068549671763e-06,
+      "loss": 0.75,
+      "step": 17335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8275095858238687e-06,
+      "loss": 0.6694,
+      "step": 17336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8259127334063973e-06,
+      "loss": 0.8828,
+      "step": 17337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.824316297753522e-06,
+      "loss": 1.0293,
+      "step": 17338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8227202789039787e-06,
+      "loss": 0.8906,
+      "step": 17339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.821124676896496e-06,
+      "loss": 0.8047,
+      "step": 17340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8195294917697922e-06,
+      "loss": 0.8193,
+      "step": 17341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8179347235625853e-06,
+      "loss": 0.8525,
+      "step": 17342
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.816340372313573e-06,
+      "loss": 0.9736,
+      "step": 17343
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8147464380614463e-06,
+      "loss": 0.8408,
+      "step": 17344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.813152920844877e-06,
+      "loss": 0.79,
+      "step": 17345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.811559820702545e-06,
+      "loss": 0.9033,
+      "step": 17346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8099671376731086e-06,
+      "loss": 0.8359,
+      "step": 17347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8083748717952162e-06,
+      "loss": 0.8018,
+      "step": 17348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8067830231075e-06,
+      "loss": 0.9219,
+      "step": 17349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.805191591648604e-06,
+      "loss": 0.8555,
+      "step": 17350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8036005774571373e-06,
+      "loss": 0.7959,
+      "step": 17351
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.802009980571714e-06,
+      "loss": 0.918,
+      "step": 17352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.800419801030925e-06,
+      "loss": 0.6182,
+      "step": 17353
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7988300388733703e-06,
+      "loss": 0.916,
+      "step": 17354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7972406941376218e-06,
+      "loss": 0.9189,
+      "step": 17355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7956517668622528e-06,
+      "loss": 0.9307,
+      "step": 17356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.794063257085813e-06,
+      "loss": 0.8613,
+      "step": 17357
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.792475164846864e-06,
+      "loss": 0.9053,
+      "step": 17358
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7908874901839357e-06,
+      "loss": 0.9229,
+      "step": 17359
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7893002331355566e-06,
+      "loss": 0.9014,
+      "step": 17360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7877133937402435e-06,
+      "loss": 0.9277,
+      "step": 17361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.786126972036509e-06,
+      "loss": 0.5737,
+      "step": 17362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.784540968062852e-06,
+      "loss": 0.8379,
+      "step": 17363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.782955381857755e-06,
+      "loss": 0.7246,
+      "step": 17364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7813702134596933e-06,
+      "loss": 0.8511,
+      "step": 17365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.779785462907141e-06,
+      "loss": 0.7324,
+      "step": 17366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7782011302385537e-06,
+      "loss": 0.9062,
+      "step": 17367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7766172154923744e-06,
+      "loss": 0.752,
+      "step": 17368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7750337187070474e-06,
+      "loss": 0.7786,
+      "step": 17369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7734506399209937e-06,
+      "loss": 0.8213,
+      "step": 17370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7718679791726334e-06,
+      "loss": 0.8682,
+      "step": 17371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7702857365003643e-06,
+      "loss": 0.8887,
+      "step": 17372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7687039119425962e-06,
+      "loss": 0.873,
+      "step": 17373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.767122505537707e-06,
+      "loss": 0.834,
+      "step": 17374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.765541517324075e-06,
+      "loss": 0.8154,
+      "step": 17375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.763960947340063e-06,
+      "loss": 0.875,
+      "step": 17376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.762380795624031e-06,
+      "loss": 0.9111,
+      "step": 17377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7608010622143266e-06,
+      "loss": 0.8936,
+      "step": 17378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.759221747149279e-06,
+      "loss": 0.7505,
+      "step": 17379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7576428504672127e-06,
+      "loss": 0.7788,
+      "step": 17380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7560643722064504e-06,
+      "loss": 0.8936,
+      "step": 17381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7544863124052934e-06,
+      "loss": 0.7988,
+      "step": 17382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.752908671102037e-06,
+      "loss": 0.8496,
+      "step": 17383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7513314483349596e-06,
+      "loss": 0.8682,
+      "step": 17384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7497546441423483e-06,
+      "loss": 0.9082,
+      "step": 17385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7481782585624573e-06,
+      "loss": 0.8975,
+      "step": 17386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.746602291633547e-06,
+      "loss": 0.958,
+      "step": 17387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7450267433938526e-06,
+      "loss": 0.8042,
+      "step": 17388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7434516138816205e-06,
+      "loss": 1.1113,
+      "step": 17389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7418769031350655e-06,
+      "loss": 0.8779,
+      "step": 17390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.740302611192405e-06,
+      "loss": 0.8271,
+      "step": 17391
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7387287380918357e-06,
+      "loss": 0.9404,
+      "step": 17392
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.737155283871562e-06,
+      "loss": 0.8672,
+      "step": 17393
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7355822485697615e-06,
+      "loss": 0.8008,
+      "step": 17394
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.734009632224606e-06,
+      "loss": 0.8848,
+      "step": 17395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7324374348742557e-06,
+      "loss": 0.6772,
+      "step": 17396
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7308656565568716e-06,
+      "loss": 0.7998,
+      "step": 17397
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7292942973105894e-06,
+      "loss": 0.7822,
+      "step": 17398
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.727723357173546e-06,
+      "loss": 0.8086,
+      "step": 17399
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.726152836183855e-06,
+      "loss": 0.7227,
+      "step": 17400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.724582734379637e-06,
+      "loss": 0.9189,
+      "step": 17401
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7230130517989927e-06,
+      "loss": 0.7183,
+      "step": 17402
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7214437884800114e-06,
+      "loss": 0.8076,
+      "step": 17403
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7198749444607696e-06,
+      "loss": 0.8066,
+      "step": 17404
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7183065197793503e-06,
+      "loss": 0.7969,
+      "step": 17405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.716738514473807e-06,
+      "loss": 0.8086,
+      "step": 17406
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7151709285821914e-06,
+      "loss": 0.8164,
+      "step": 17407
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7136037621425404e-06,
+      "loss": 0.606,
+      "step": 17408
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.712037015192894e-06,
+      "loss": 0.7383,
+      "step": 17409
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.710470687771265e-06,
+      "loss": 1.0078,
+      "step": 17410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7089047799156685e-06,
+      "loss": 0.8984,
+      "step": 17411
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7073392916640952e-06,
+      "loss": 0.9258,
+      "step": 17412
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.705774223054547e-06,
+      "loss": 0.9297,
+      "step": 17413
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7042095741249963e-06,
+      "loss": 0.7451,
+      "step": 17414
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7026453449134106e-06,
+      "loss": 0.8574,
+      "step": 17415
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7010815354577526e-06,
+      "loss": 0.7695,
+      "step": 17416
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.699518145795976e-06,
+      "loss": 0.8188,
+      "step": 17417
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.697955175966016e-06,
+      "loss": 0.8281,
+      "step": 17418
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6963926260057995e-06,
+      "loss": 0.7559,
+      "step": 17419
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.694830495953242e-06,
+      "loss": 0.8301,
+      "step": 17420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.693268785846259e-06,
+      "loss": 0.8682,
+      "step": 17421
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6917074957227463e-06,
+      "loss": 0.8916,
+      "step": 17422
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.690146625620591e-06,
+      "loss": 1.0195,
+      "step": 17423
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6885861755776654e-06,
+      "loss": 0.999,
+      "step": 17424
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.687026145631848e-06,
+      "loss": 0.957,
+      "step": 17425
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6854665358209886e-06,
+      "loss": 0.7715,
+      "step": 17426
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6839073461829367e-06,
+      "loss": 0.7744,
+      "step": 17427
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6823485767555248e-06,
+      "loss": 0.7266,
+      "step": 17428
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6807902275765885e-06,
+      "loss": 0.8643,
+      "step": 17429
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.679232298683938e-06,
+      "loss": 0.7139,
+      "step": 17430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.677674790115381e-06,
+      "loss": 0.959,
+      "step": 17431
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6761177019087116e-06,
+      "loss": 0.8564,
+      "step": 17432
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.674561034101719e-06,
+      "loss": 0.8613,
+      "step": 17433
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6730047867321806e-06,
+      "loss": 0.8643,
+      "step": 17434
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.671448959837859e-06,
+      "loss": 0.8857,
+      "step": 17435
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.669893553456504e-06,
+      "loss": 0.9053,
+      "step": 17436
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.668338567625872e-06,
+      "loss": 0.7769,
+      "step": 17437
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.666784002383691e-06,
+      "loss": 0.7432,
+      "step": 17438
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.665229857767688e-06,
+      "loss": 0.7285,
+      "step": 17439
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6636761338155713e-06,
+      "loss": 0.8223,
+      "step": 17440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6621228305650547e-06,
+      "loss": 0.8389,
+      "step": 17441
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6605699480538305e-06,
+      "loss": 0.8125,
+      "step": 17442
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6590174863195796e-06,
+      "loss": 0.8867,
+      "step": 17443
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6574654453999693e-06,
+      "loss": 0.7363,
+      "step": 17444
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6559138253326768e-06,
+      "loss": 0.9678,
+      "step": 17445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.654362626155349e-06,
+      "loss": 0.7764,
+      "step": 17446
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6528118479056274e-06,
+      "loss": 0.9639,
+      "step": 17447
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.651261490621142e-06,
+      "loss": 0.7119,
+      "step": 17448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6497115543395226e-06,
+      "loss": 0.7822,
+      "step": 17449
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6481620390983807e-06,
+      "loss": 0.9268,
+      "step": 17450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6466129449353096e-06,
+      "loss": 0.8564,
+      "step": 17451
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6450642718879137e-06,
+      "loss": 0.8818,
+      "step": 17452
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6435160199937682e-06,
+      "loss": 0.7822,
+      "step": 17453
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.641968189290447e-06,
+      "loss": 0.8799,
+      "step": 17454
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6404207798155024e-06,
+      "loss": 0.8525,
+      "step": 17455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6388737916065e-06,
+      "loss": 0.9189,
+      "step": 17456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6373272247009717e-06,
+      "loss": 0.8291,
+      "step": 17457
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.63578107913645e-06,
+      "loss": 0.9023,
+      "step": 17458
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6342353549504517e-06,
+      "loss": 0.8516,
+      "step": 17459
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.632690052180493e-06,
+      "loss": 0.8467,
+      "step": 17460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6311451708640734e-06,
+      "loss": 0.915,
+      "step": 17461
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.62960071103868e-06,
+      "loss": 0.8662,
+      "step": 17462
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6280566727417877e-06,
+      "loss": 0.9385,
+      "step": 17463
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.626513056010875e-06,
+      "loss": 0.8096,
+      "step": 17464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.624969860883397e-06,
+      "loss": 1.041,
+      "step": 17465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.623427087396804e-06,
+      "loss": 0.7075,
+      "step": 17466
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.621884735588527e-06,
+      "loss": 0.9434,
+      "step": 17467
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6203428054960054e-06,
+      "loss": 0.6372,
+      "step": 17468
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.618801297156652e-06,
+      "loss": 0.7173,
+      "step": 17469
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6172602106078747e-06,
+      "loss": 1.0391,
+      "step": 17470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.615719545887068e-06,
+      "loss": 0.7812,
+      "step": 17471
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6141793030316255e-06,
+      "loss": 0.8247,
+      "step": 17472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6126394820789223e-06,
+      "loss": 1.0312,
+      "step": 17473
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6111000830663225e-06,
+      "loss": 0.9551,
+      "step": 17474
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.609561106031182e-06,
+      "loss": 0.8633,
+      "step": 17475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.608022551010856e-06,
+      "loss": 0.8896,
+      "step": 17476
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.606484418042672e-06,
+      "loss": 0.9043,
+      "step": 17477
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.60494670716396e-06,
+      "loss": 0.7251,
+      "step": 17478
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6034094184120286e-06,
+      "loss": 0.8047,
+      "step": 17479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.601872551824196e-06,
+      "loss": 0.8867,
+      "step": 17480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.600336107437749e-06,
+      "loss": 1.0049,
+      "step": 17481
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5988000852899743e-06,
+      "loss": 0.8945,
+      "step": 17482
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5972644854181407e-06,
+      "loss": 1.0938,
+      "step": 17483
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5957293078595247e-06,
+      "loss": 0.8125,
+      "step": 17484
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5941945526513745e-06,
+      "loss": 0.9541,
+      "step": 17485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5926602198309313e-06,
+      "loss": 0.8633,
+      "step": 17486
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.59112630943543e-06,
+      "loss": 0.9863,
+      "step": 17487
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.589592821502098e-06,
+      "loss": 0.7783,
+      "step": 17488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5880597560681463e-06,
+      "loss": 0.8486,
+      "step": 17489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.586527113170778e-06,
+      "loss": 1.0693,
+      "step": 17490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.584994892847181e-06,
+      "loss": 0.8213,
+      "step": 17491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5834630951345486e-06,
+      "loss": 0.7842,
+      "step": 17492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5819317200700455e-06,
+      "loss": 0.7734,
+      "step": 17493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.580400767690834e-06,
+      "loss": 0.9629,
+      "step": 17494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.578870238034061e-06,
+      "loss": 0.7109,
+      "step": 17495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.577340131136881e-06,
+      "loss": 0.9434,
+      "step": 17496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5758104470364155e-06,
+      "loss": 0.9326,
+      "step": 17497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5742811857697847e-06,
+      "loss": 0.8252,
+      "step": 17498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5727523473741054e-06,
+      "loss": 0.877,
+      "step": 17499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.571223931886473e-06,
+      "loss": 0.9004,
+      "step": 17500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5696959393439813e-06,
+      "loss": 0.8525,
+      "step": 17501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.56816836978371e-06,
+      "loss": 0.8076,
+      "step": 17502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.566641223242723e-06,
+      "loss": 0.8848,
+      "step": 17503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5651144997580877e-06,
+      "loss": 0.8594,
+      "step": 17504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5635881993668487e-06,
+      "loss": 0.7939,
+      "step": 17505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5620623221060446e-06,
+      "loss": 0.9473,
+      "step": 17506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.560536868012702e-06,
+      "loss": 0.833,
+      "step": 17507
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.559011837123846e-06,
+      "loss": 0.7959,
+      "step": 17508
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.557487229476481e-06,
+      "loss": 0.9365,
+      "step": 17509
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5559630451076056e-06,
+      "loss": 0.7778,
+      "step": 17510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5544392840541998e-06,
+      "loss": 0.8496,
+      "step": 17511
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5529159463532515e-06,
+      "loss": 0.8662,
+      "step": 17512
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.551393032041725e-06,
+      "loss": 0.9014,
+      "step": 17513
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5498705411565716e-06,
+      "loss": 0.7627,
+      "step": 17514
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5483484737347386e-06,
+      "loss": 1.0049,
+      "step": 17515
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.546826829813169e-06,
+      "loss": 0.8105,
+      "step": 17516
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5453056094287854e-06,
+      "loss": 0.7725,
+      "step": 17517
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.543784812618502e-06,
+      "loss": 0.6802,
+      "step": 17518
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5422644394192195e-06,
+      "loss": 0.7021,
+      "step": 17519
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.540744489867841e-06,
+      "loss": 0.9746,
+      "step": 17520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.53922496400125e-06,
+      "loss": 0.8545,
+      "step": 17521
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5377058618563167e-06,
+      "loss": 0.8525,
+      "step": 17522
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.536187183469905e-06,
+      "loss": 0.9854,
+      "step": 17523
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5346689288788738e-06,
+      "loss": 0.8564,
+      "step": 17524
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5331510981200635e-06,
+      "loss": 0.8252,
+      "step": 17525
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5316336912303086e-06,
+      "loss": 0.9238,
+      "step": 17526
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5301167082464283e-06,
+      "loss": 0.8457,
+      "step": 17527
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5286001492052427e-06,
+      "loss": 0.7856,
+      "step": 17528
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.527084014143548e-06,
+      "loss": 0.8379,
+      "step": 17529
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5255683030981405e-06,
+      "loss": 0.9072,
+      "step": 17530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.524053016105794e-06,
+      "loss": 0.8799,
+      "step": 17531
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.522538153203291e-06,
+      "loss": 0.7354,
+      "step": 17532
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5210237144273865e-06,
+      "loss": 0.8789,
+      "step": 17533
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.519509699814835e-06,
+      "loss": 0.8096,
+      "step": 17534
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5179961094023674e-06,
+      "loss": 0.9102,
+      "step": 17535
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.516482943226728e-06,
+      "loss": 0.9424,
+      "step": 17536
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.514970201324629e-06,
+      "loss": 0.8115,
+      "step": 17537
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5134578837327796e-06,
+      "loss": 0.8262,
+      "step": 17538
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5119459904878827e-06,
+      "loss": 0.7939,
+      "step": 17539
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.510434521626628e-06,
+      "loss": 0.9551,
+      "step": 17540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.508923477185692e-06,
+      "loss": 0.8564,
+      "step": 17541
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5074128572017385e-06,
+      "loss": 0.8867,
+      "step": 17542
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5059026617114368e-06,
+      "loss": 0.8271,
+      "step": 17543
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5043928907514283e-06,
+      "loss": 0.915,
+      "step": 17544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.502883544358352e-06,
+      "loss": 0.8857,
+      "step": 17545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.501374622568831e-06,
+      "loss": 0.8906,
+      "step": 17546
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.49986612541949e-06,
+      "loss": 0.6865,
+      "step": 17547
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.498358052946932e-06,
+      "loss": 0.8193,
+      "step": 17548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.496850405187754e-06,
+      "loss": 0.8398,
+      "step": 17549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.495343182178538e-06,
+      "loss": 0.6846,
+      "step": 17550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.493836383955868e-06,
+      "loss": 0.7412,
+      "step": 17551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.492330010556303e-06,
+      "loss": 0.9131,
+      "step": 17552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.490824062016404e-06,
+      "loss": 0.7988,
+      "step": 17553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.489318538372705e-06,
+      "loss": 0.7402,
+      "step": 17554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4878134396617524e-06,
+      "loss": 0.9541,
+      "step": 17555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.486308765920067e-06,
+      "loss": 0.8555,
+      "step": 17556
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.484804517184163e-06,
+      "loss": 0.917,
+      "step": 17557
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.483300693490538e-06,
+      "loss": 1.0547,
+      "step": 17558
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4817972948756942e-06,
+      "loss": 0.9795,
+      "step": 17559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.480294321376111e-06,
+      "loss": 0.8154,
+      "step": 17560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4787917730282617e-06,
+      "loss": 0.8398,
+      "step": 17561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.477289649868602e-06,
+      "loss": 0.8564,
+      "step": 17562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4757879519335947e-06,
+      "loss": 0.8936,
+      "step": 17563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4742866792596785e-06,
+      "loss": 0.8594,
+      "step": 17564
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.472785831883282e-06,
+      "loss": 0.8418,
+      "step": 17565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4712854098408245e-06,
+      "loss": 0.9541,
+      "step": 17566
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.469785413168724e-06,
+      "loss": 0.7373,
+      "step": 17567
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.468285841903377e-06,
+      "loss": 0.834,
+      "step": 17568
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4667866960811736e-06,
+      "loss": 0.875,
+      "step": 17569
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4652879757384883e-06,
+      "loss": 0.8057,
+      "step": 17570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4637896809117033e-06,
+      "loss": 0.8906,
+      "step": 17571
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4622918116371676e-06,
+      "loss": 0.7837,
+      "step": 17572
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4607943679512335e-06,
+      "loss": 0.8311,
+      "step": 17573
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4592973498902375e-06,
+      "loss": 0.8164,
+      "step": 17574
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.457800757490512e-06,
+      "loss": 0.6729,
+      "step": 17575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.456304590788372e-06,
+      "loss": 0.9062,
+      "step": 17576
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.454808849820125e-06,
+      "loss": 0.7871,
+      "step": 17577
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4533135346220636e-06,
+      "loss": 0.9668,
+      "step": 17578
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.451818645230486e-06,
+      "loss": 0.8701,
+      "step": 17579
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4503241816816625e-06,
+      "loss": 0.6567,
+      "step": 17580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.448830144011858e-06,
+      "loss": 0.7881,
+      "step": 17581
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.447336532257327e-06,
+      "loss": 0.8027,
+      "step": 17582
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4458433464543174e-06,
+      "loss": 0.8867,
+      "step": 17583
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.444350586639068e-06,
+      "loss": 0.834,
+      "step": 17584
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4428582528478017e-06,
+      "loss": 0.7129,
+      "step": 17585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.441366345116729e-06,
+      "loss": 0.8838,
+      "step": 17586
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.439874863482059e-06,
+      "loss": 0.8296,
+      "step": 17587
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.438383807979985e-06,
+      "loss": 0.8779,
+      "step": 17588
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4368931786466863e-06,
+      "loss": 0.748,
+      "step": 17589
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4354029755183375e-06,
+      "loss": 0.9316,
+      "step": 17590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.433913198631106e-06,
+      "loss": 0.8438,
+      "step": 17591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4324238480211414e-06,
+      "loss": 0.792,
+      "step": 17592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.430934923724586e-06,
+      "loss": 0.9277,
+      "step": 17593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4294464257775683e-06,
+      "loss": 0.8154,
+      "step": 17594
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4279583542162155e-06,
+      "loss": 0.7441,
+      "step": 17595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.426470709076636e-06,
+      "loss": 0.7502,
+      "step": 17596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.42498349039493e-06,
+      "loss": 0.7559,
+      "step": 17597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.423496698207184e-06,
+      "loss": 0.6914,
+      "step": 17598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.422010332549487e-06,
+      "loss": 1.0098,
+      "step": 17599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4205243934579037e-06,
+      "loss": 0.9385,
+      "step": 17600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4190388809684963e-06,
+      "loss": 0.9414,
+      "step": 17601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4175537951173045e-06,
+      "loss": 0.8672,
+      "step": 17602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4160691359403798e-06,
+      "loss": 0.8398,
+      "step": 17603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4145849034737444e-06,
+      "loss": 0.9033,
+      "step": 17604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.413101097753416e-06,
+      "loss": 0.8379,
+      "step": 17605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4116177188154e-06,
+      "loss": 0.9619,
+      "step": 17606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4101347666956996e-06,
+      "loss": 0.8159,
+      "step": 17607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.408652241430298e-06,
+      "loss": 0.7158,
+      "step": 17608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.407170143055173e-06,
+      "loss": 0.8594,
+      "step": 17609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4056884716062866e-06,
+      "loss": 0.8887,
+      "step": 17610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.404207227119599e-06,
+      "loss": 0.707,
+      "step": 17611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4027264096310576e-06,
+      "loss": 0.8311,
+      "step": 17612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.401246019176593e-06,
+      "loss": 0.9072,
+      "step": 17613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3997660557921275e-06,
+      "loss": 0.8291,
+      "step": 17614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.398286519513584e-06,
+      "loss": 0.7759,
+      "step": 17615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3968074103768624e-06,
+      "loss": 0.9355,
+      "step": 17616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.395328728417856e-06,
+      "loss": 0.8096,
+      "step": 17617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3938504736724432e-06,
+      "loss": 0.8721,
+      "step": 17618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3923726461765063e-06,
+      "loss": 0.8057,
+      "step": 17619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3908952459659007e-06,
+      "loss": 0.8721,
+      "step": 17620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3894182730764825e-06,
+      "loss": 0.6855,
+      "step": 17621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3879417275440877e-06,
+      "loss": 0.8711,
+      "step": 17622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3864656094045578e-06,
+      "loss": 0.9434,
+      "step": 17623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3849899186937053e-06,
+      "loss": 0.8613,
+      "step": 17624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.383514655447341e-06,
+      "loss": 0.8179,
+      "step": 17625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3820398197012716e-06,
+      "loss": 0.9033,
+      "step": 17626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3805654114912847e-06,
+      "loss": 0.7803,
+      "step": 17627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3790914308531556e-06,
+      "loss": 0.5657,
+      "step": 17628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3776178778226535e-06,
+      "loss": 0.8711,
+      "step": 17629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3761447524355428e-06,
+      "loss": 0.8975,
+      "step": 17630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3746720547275716e-06,
+      "loss": 0.8789,
+      "step": 17631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3731997847344746e-06,
+      "loss": 0.8457,
+      "step": 17632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.371727942491975e-06,
+      "loss": 0.7656,
+      "step": 17633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3702565280357993e-06,
+      "loss": 1.0527,
+      "step": 17634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3687855414016525e-06,
+      "loss": 0.7783,
+      "step": 17635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.367314982625226e-06,
+      "loss": 0.9307,
+      "step": 17636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.365844851742207e-06,
+      "loss": 0.7109,
+      "step": 17637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3643751487882783e-06,
+      "loss": 0.9336,
+      "step": 17638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3629058737990997e-06,
+      "loss": 0.791,
+      "step": 17639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.361437026810325e-06,
+      "loss": 0.7637,
+      "step": 17640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3599686078575967e-06,
+      "loss": 0.5962,
+      "step": 17641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.358500616976558e-06,
+      "loss": 0.8232,
+      "step": 17642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3570330542028287e-06,
+      "loss": 0.7754,
+      "step": 17643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3555659195720184e-06,
+      "loss": 0.751,
+      "step": 17644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3540992131197294e-06,
+      "loss": 0.9268,
+      "step": 17645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.352632934881562e-06,
+      "loss": 0.8379,
+      "step": 17646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3511670848930955e-06,
+      "loss": 0.833,
+      "step": 17647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3497016631899005e-06,
+      "loss": 0.8174,
+      "step": 17648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.348236669807533e-06,
+      "loss": 0.6987,
+      "step": 17649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.346772104781554e-06,
+      "loss": 0.8369,
+      "step": 17650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.345307968147499e-06,
+      "loss": 0.7871,
+      "step": 17651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3438442599408995e-06,
+      "loss": 0.8291,
+      "step": 17652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3423809801972717e-06,
+      "loss": 0.7026,
+      "step": 17653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.340918128952132e-06,
+      "loss": 0.9199,
+      "step": 17654
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.339455706240976e-06,
+      "loss": 0.9629,
+      "step": 17655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3379937120992913e-06,
+      "loss": 0.7471,
+      "step": 17656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3365321465625535e-06,
+      "loss": 0.7271,
+      "step": 17657
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.335071009666239e-06,
+      "loss": 0.9404,
+      "step": 17658
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.333610301445799e-06,
+      "loss": 0.8833,
+      "step": 17659
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.332150021936683e-06,
+      "loss": 0.7314,
+      "step": 17660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3306901711743236e-06,
+      "loss": 1.0771,
+      "step": 17661
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3292307491941536e-06,
+      "loss": 0.9443,
+      "step": 17662
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.327771756031585e-06,
+      "loss": 0.9414,
+      "step": 17663
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.326313191722023e-06,
+      "loss": 1.0439,
+      "step": 17664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.324855056300859e-06,
+      "loss": 0.834,
+      "step": 17665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3233973498034824e-06,
+      "loss": 0.8096,
+      "step": 17666
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.321940072265272e-06,
+      "loss": 0.8389,
+      "step": 17667
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.320483223721588e-06,
+      "loss": 0.7676,
+      "step": 17668
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.319026804207776e-06,
+      "loss": 0.7979,
+      "step": 17669
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3175708137591913e-06,
+      "loss": 0.8809,
+      "step": 17670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3161152524111595e-06,
+      "loss": 0.7949,
+      "step": 17671
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.314660120199006e-06,
+      "loss": 0.7949,
+      "step": 17672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3132054171580354e-06,
+      "loss": 1.0059,
+      "step": 17673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3117511433235572e-06,
+      "loss": 0.8008,
+      "step": 17674
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3102972987308614e-06,
+      "loss": 0.9375,
+      "step": 17675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.308843883415226e-06,
+      "loss": 0.6533,
+      "step": 17676
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3073908974119164e-06,
+      "loss": 0.7656,
+      "step": 17677
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3059383407562043e-06,
+      "loss": 0.7061,
+      "step": 17678
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.30448621348333e-06,
+      "loss": 0.8535,
+      "step": 17679
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.303034515628535e-06,
+      "loss": 0.7769,
+      "step": 17680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.301583247227044e-06,
+      "loss": 0.6123,
+      "step": 17681
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.300132408314082e-06,
+      "loss": 0.8301,
+      "step": 17682
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2986819989248543e-06,
+      "loss": 0.8047,
+      "step": 17683
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.297232019094555e-06,
+      "loss": 0.8193,
+      "step": 17684
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2957824688583697e-06,
+      "loss": 0.7607,
+      "step": 17685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2943333482514807e-06,
+      "loss": 0.9951,
+      "step": 17686
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2928846573090512e-06,
+      "loss": 0.9912,
+      "step": 17687
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2914363960662376e-06,
+      "loss": 1.0352,
+      "step": 17688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2899885645581787e-06,
+      "loss": 0.8809,
+      "step": 17689
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.288541162820017e-06,
+      "loss": 0.6372,
+      "step": 17690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.287094190886876e-06,
+      "loss": 0.6885,
+      "step": 17691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.285647648793865e-06,
+      "loss": 0.8232,
+      "step": 17692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.284201536576087e-06,
+      "loss": 0.8242,
+      "step": 17693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2827558542686413e-06,
+      "loss": 0.8662,
+      "step": 17694
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2813106019066055e-06,
+      "loss": 0.8926,
+      "step": 17695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2798657795250525e-06,
+      "loss": 0.75,
+      "step": 17696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.278421387159042e-06,
+      "loss": 1.1406,
+      "step": 17697
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2769774248436294e-06,
+      "loss": 0.9814,
+      "step": 17698
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2755338926138526e-06,
+      "loss": 0.9736,
+      "step": 17699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2740907905047438e-06,
+      "loss": 0.8545,
+      "step": 17700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.272648118551317e-06,
+      "loss": 0.8994,
+      "step": 17701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2712058767885916e-06,
+      "loss": 0.7297,
+      "step": 17702
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2697640652515585e-06,
+      "loss": 0.7422,
+      "step": 17703
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2683226839752104e-06,
+      "loss": 0.8037,
+      "step": 17704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2668817329945192e-06,
+      "loss": 0.8594,
+      "step": 17705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2654412123444616e-06,
+      "loss": 0.75,
+      "step": 17706
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.264001122059989e-06,
+      "loss": 0.8701,
+      "step": 17707
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2625614621760515e-06,
+      "loss": 0.7686,
+      "step": 17708
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.261122232727577e-06,
+      "loss": 0.7578,
+      "step": 17709
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2596834337495044e-06,
+      "loss": 0.7529,
+      "step": 17710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2582450652767406e-06,
+      "loss": 1.0205,
+      "step": 17711
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2568071273441896e-06,
+      "loss": 0.8579,
+      "step": 17712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2553696199867515e-06,
+      "loss": 0.9668,
+      "step": 17713
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2539325432393077e-06,
+      "loss": 0.8496,
+      "step": 17714
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2524958971367326e-06,
+      "loss": 0.9795,
+      "step": 17715
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2510596817138854e-06,
+      "loss": 0.9746,
+      "step": 17716
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.249623897005624e-06,
+      "loss": 0.793,
+      "step": 17717
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2481885430467896e-06,
+      "loss": 0.75,
+      "step": 17718
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.246753619872213e-06,
+      "loss": 0.9336,
+      "step": 17719
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.245319127516712e-06,
+      "loss": 0.6924,
+      "step": 17720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2438850660151035e-06,
+      "loss": 0.9043,
+      "step": 17721
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.242451435402189e-06,
+      "loss": 0.9404,
+      "step": 17722
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2410182357127507e-06,
+      "loss": 0.6685,
+      "step": 17723
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2395854669815707e-06,
+      "loss": 0.8096,
+      "step": 17724
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2381531292434256e-06,
+      "loss": 0.8223,
+      "step": 17725
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.236721222533067e-06,
+      "loss": 0.791,
+      "step": 17726
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.235289746885243e-06,
+      "loss": 0.9727,
+      "step": 17727
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.23385870233469e-06,
+      "loss": 0.8789,
+      "step": 17728
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.232428088916141e-06,
+      "loss": 0.8252,
+      "step": 17729
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2309979066643116e-06,
+      "loss": 0.9814,
+      "step": 17730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2295681556139036e-06,
+      "loss": 0.8438,
+      "step": 17731
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2281388357996138e-06,
+      "loss": 0.8232,
+      "step": 17732
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.226709947256134e-06,
+      "loss": 0.6968,
+      "step": 17733
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2252814900181317e-06,
+      "loss": 0.8877,
+      "step": 17734
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.223853464120276e-06,
+      "loss": 0.7754,
+      "step": 17735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2224258695972135e-06,
+      "loss": 0.7373,
+      "step": 17736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2209987064835992e-06,
+      "loss": 0.916,
+      "step": 17737
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.219571974814059e-06,
+      "loss": 0.9131,
+      "step": 17738
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2181456746232154e-06,
+      "loss": 0.9233,
+      "step": 17739
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2167198059456797e-06,
+      "loss": 0.9111,
+      "step": 17740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2152943688160566e-06,
+      "loss": 0.835,
+      "step": 17741
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.213869363268939e-06,
+      "loss": 0.8145,
+      "step": 17742
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2124447893389054e-06,
+      "loss": 0.6934,
+      "step": 17743
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.21102064706052e-06,
+      "loss": 0.8516,
+      "step": 17744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2095969364683524e-06,
+      "loss": 0.9453,
+      "step": 17745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.208173657596946e-06,
+      "loss": 0.8096,
+      "step": 17746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2067508104808422e-06,
+      "loss": 0.8477,
+      "step": 17747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2053283951545645e-06,
+      "loss": 0.9111,
+      "step": 17748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2039064116526387e-06,
+      "loss": 0.7783,
+      "step": 17749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2024848600095637e-06,
+      "loss": 0.9854,
+      "step": 17750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.201063740259843e-06,
+      "loss": 1.0098,
+      "step": 17751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1996430524379588e-06,
+      "loss": 0.8105,
+      "step": 17752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.198222796578391e-06,
+      "loss": 0.9658,
+      "step": 17753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1968029727156036e-06,
+      "loss": 0.7998,
+      "step": 17754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1953835808840495e-06,
+      "loss": 0.7988,
+      "step": 17755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.193964621118172e-06,
+      "loss": 0.8594,
+      "step": 17756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.19254609345241e-06,
+      "loss": 0.6958,
+      "step": 17757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.191127997921185e-06,
+      "loss": 0.918,
+      "step": 17758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.189710334558912e-06,
+      "loss": 0.8447,
+      "step": 17759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1882931033999833e-06,
+      "loss": 0.9375,
+      "step": 17760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.186876304478804e-06,
+      "loss": 0.7891,
+      "step": 17761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1854599378297504e-06,
+      "loss": 0.9316,
+      "step": 17762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1840440034871937e-06,
+      "loss": 0.9482,
+      "step": 17763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.182628501485491e-06,
+      "loss": 0.8848,
+      "step": 17764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1812134318589974e-06,
+      "loss": 0.9033,
+      "step": 17765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1797987946420518e-06,
+      "loss": 0.8496,
+      "step": 17766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1783845898689826e-06,
+      "loss": 0.7441,
+      "step": 17767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1769708175741045e-06,
+      "loss": 0.8574,
+      "step": 17768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.175557477791732e-06,
+      "loss": 0.8467,
+      "step": 17769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.174144570556158e-06,
+      "loss": 0.8096,
+      "step": 17770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.172732095901675e-06,
+      "loss": 0.7617,
+      "step": 17771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.171320053862549e-06,
+      "loss": 0.9336,
+      "step": 17772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.169908444473059e-06,
+      "loss": 0.7969,
+      "step": 17773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1684972677674534e-06,
+      "loss": 0.7715,
+      "step": 17774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1670865237799775e-06,
+      "loss": 1.0723,
+      "step": 17775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.165676212544865e-06,
+      "loss": 0.9014,
+      "step": 17776
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1642663340963456e-06,
+      "loss": 0.7666,
+      "step": 17777
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1628568884686273e-06,
+      "loss": 0.8125,
+      "step": 17778
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1614478756959166e-06,
+      "loss": 0.8945,
+      "step": 17779
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1600392958124016e-06,
+      "loss": 0.709,
+      "step": 17780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1586311488522705e-06,
+      "loss": 0.8848,
+      "step": 17781
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1572234348496913e-06,
+      "loss": 0.8477,
+      "step": 17782
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1558161538388254e-06,
+      "loss": 0.8848,
+      "step": 17783
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1544093058538218e-06,
+      "loss": 0.7148,
+      "step": 17784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.153002890928826e-06,
+      "loss": 0.8809,
+      "step": 17785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1515969090979615e-06,
+      "loss": 0.7646,
+      "step": 17786
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1501913603953507e-06,
+      "loss": 0.8447,
+      "step": 17787
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.148786244855099e-06,
+      "loss": 1.0957,
+      "step": 17788
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1473815625113083e-06,
+      "loss": 0.8643,
+      "step": 17789
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1459773133980666e-06,
+      "loss": 0.7764,
+      "step": 17790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.144573497549447e-06,
+      "loss": 0.8662,
+      "step": 17791
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.143170114999513e-06,
+      "loss": 0.8545,
+      "step": 17792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1417671657823313e-06,
+      "loss": 0.8262,
+      "step": 17793
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1403646499319407e-06,
+      "loss": 0.9082,
+      "step": 17794
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.138962567482372e-06,
+      "loss": 0.8809,
+      "step": 17795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1375609184676583e-06,
+      "loss": 0.8691,
+      "step": 17796
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1361597029218096e-06,
+      "loss": 0.8369,
+      "step": 17797
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.13475892087883e-06,
+      "loss": 0.8799,
+      "step": 17798
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.133358572372708e-06,
+      "loss": 0.9805,
+      "step": 17799
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.131958657437432e-06,
+      "loss": 0.8232,
+      "step": 17800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1305591761069723e-06,
+      "loss": 0.8389,
+      "step": 17801
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1291601284152906e-06,
+      "loss": 0.8701,
+      "step": 17802
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1277615143963294e-06,
+      "loss": 0.791,
+      "step": 17803
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.126363334084043e-06,
+      "loss": 0.7988,
+      "step": 17804
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.124965587512351e-06,
+      "loss": 0.8945,
+      "step": 17805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1235682747151776e-06,
+      "loss": 0.9238,
+      "step": 17806
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1221713957264246e-06,
+      "loss": 0.667,
+      "step": 17807
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.12077495058e-06,
+      "loss": 0.8799,
+      "step": 17808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.119378939309786e-06,
+      "loss": 0.8301,
+      "step": 17809
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1179833619496603e-06,
+      "loss": 0.9697,
+      "step": 17810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1165882185334842e-06,
+      "loss": 0.7954,
+      "step": 17811
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1151935090951237e-06,
+      "loss": 0.6953,
+      "step": 17812
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1137992336684188e-06,
+      "loss": 0.7988,
+      "step": 17813
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1124053922872047e-06,
+      "loss": 0.9355,
+      "step": 17814
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1110119849853027e-06,
+      "loss": 0.7451,
+      "step": 17815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1096190117965355e-06,
+      "loss": 0.9023,
+      "step": 17816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1082264727547e-06,
+      "loss": 0.918,
+      "step": 17817
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.106834367893591e-06,
+      "loss": 0.835,
+      "step": 17818
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.105442697246984e-06,
+      "loss": 0.8145,
+      "step": 17819
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.104051460848664e-06,
+      "loss": 0.8867,
+      "step": 17820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.102660658732383e-06,
+      "loss": 0.8613,
+      "step": 17821
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.101270290931892e-06,
+      "loss": 0.9014,
+      "step": 17822
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0998803574809303e-06,
+      "loss": 0.7432,
+      "step": 17823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0984908584132356e-06,
+      "loss": 0.9307,
+      "step": 17824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0971017937625192e-06,
+      "loss": 0.8887,
+      "step": 17825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.095713163562492e-06,
+      "loss": 0.8877,
+      "step": 17826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.094324967846848e-06,
+      "loss": 0.9424,
+      "step": 17827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0929372066492838e-06,
+      "loss": 0.8379,
+      "step": 17828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0915498800034672e-06,
+      "loss": 0.8936,
+      "step": 17829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0901629879430717e-06,
+      "loss": 0.9141,
+      "step": 17830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0887765305017438e-06,
+      "loss": 0.8223,
+      "step": 17831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0873905077131383e-06,
+      "loss": 0.7832,
+      "step": 17832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0860049196108825e-06,
+      "loss": 1.0645,
+      "step": 17833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.084619766228608e-06,
+      "loss": 0.8701,
+      "step": 17834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.083235047599923e-06,
+      "loss": 0.8555,
+      "step": 17835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.081850763758433e-06,
+      "loss": 0.9395,
+      "step": 17836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.080466914737731e-06,
+      "loss": 0.9033,
+      "step": 17837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.079083500571397e-06,
+      "loss": 0.9082,
+      "step": 17838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.077700521292998e-06,
+      "loss": 0.7559,
+      "step": 17839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0763179769361043e-06,
+      "loss": 0.8242,
+      "step": 17840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0749358675342622e-06,
+      "loss": 0.9521,
+      "step": 17841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0735541931210123e-06,
+      "loss": 0.707,
+      "step": 17842
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0721729537298763e-06,
+      "loss": 0.8828,
+      "step": 17843
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0707921493943827e-06,
+      "loss": 0.9316,
+      "step": 17844
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0694117801480364e-06,
+      "loss": 0.917,
+      "step": 17845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0680318460243345e-06,
+      "loss": 0.873,
+      "step": 17846
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.066652347056759e-06,
+      "loss": 0.9229,
+      "step": 17847
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0652732832787947e-06,
+      "loss": 0.75,
+      "step": 17848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.063894654723906e-06,
+      "loss": 0.7529,
+      "step": 17849
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0625164614255435e-06,
+      "loss": 0.7402,
+      "step": 17850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0611387034171514e-06,
+      "loss": 0.9199,
+      "step": 17851
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0597613807321705e-06,
+      "loss": 0.9414,
+      "step": 17852
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.05838449340402e-06,
+      "loss": 0.7891,
+      "step": 17853
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.057008041466113e-06,
+      "loss": 0.8311,
+      "step": 17854
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0556320249518503e-06,
+      "loss": 0.9883,
+      "step": 17855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0542564438946265e-06,
+      "loss": 0.7446,
+      "step": 17856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0528812983278246e-06,
+      "loss": 0.8613,
+      "step": 17857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.051506588284813e-06,
+      "loss": 0.9775,
+      "step": 17858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0501323137989493e-06,
+      "loss": 0.8613,
+      "step": 17859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0487584749035873e-06,
+      "loss": 0.8086,
+      "step": 17860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0473850716320643e-06,
+      "loss": 0.7686,
+      "step": 17861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0460121040177096e-06,
+      "loss": 0.677,
+      "step": 17862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0446395720938384e-06,
+      "loss": 0.9355,
+      "step": 17863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.04326747589376e-06,
+      "loss": 0.686,
+      "step": 17864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.041895815450774e-06,
+      "loss": 0.7461,
+      "step": 17865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.040524590798163e-06,
+      "loss": 0.9385,
+      "step": 17866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0391538019691983e-06,
+      "loss": 0.8643,
+      "step": 17867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.037783448997155e-06,
+      "loss": 0.8398,
+      "step": 17868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0364135319152824e-06,
+      "loss": 0.793,
+      "step": 17869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0350440507568247e-06,
+      "loss": 0.9385,
+      "step": 17870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.033675005555009e-06,
+      "loss": 0.9121,
+      "step": 17871
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.032306396343069e-06,
+      "loss": 0.8594,
+      "step": 17872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.030938223154213e-06,
+      "loss": 0.6069,
+      "step": 17873
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.029570486021639e-06,
+      "loss": 0.7451,
+      "step": 17874
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0282031849785386e-06,
+      "loss": 0.8008,
+      "step": 17875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0268363200580966e-06,
+      "loss": 0.9043,
+      "step": 17876
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0254698912934788e-06,
+      "loss": 0.8633,
+      "step": 17877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0241038987178462e-06,
+      "loss": 0.8604,
+      "step": 17878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0227383423643433e-06,
+      "loss": 0.7783,
+      "step": 17879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0213732222661143e-06,
+      "loss": 0.8691,
+      "step": 17880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.020008538456286e-06,
+      "loss": 0.8516,
+      "step": 17881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.018644290967968e-06,
+      "loss": 0.7754,
+      "step": 17882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.017280479834276e-06,
+      "loss": 0.9004,
+      "step": 17883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.015917105088301e-06,
+      "loss": 0.9033,
+      "step": 17884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.01455416676313e-06,
+      "loss": 0.9463,
+      "step": 17885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0131916648918294e-06,
+      "loss": 0.8604,
+      "step": 17886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0118295995074755e-06,
+      "loss": 0.8975,
+      "step": 17887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0104679706431175e-06,
+      "loss": 0.79,
+      "step": 17888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.009106778331793e-06,
+      "loss": 0.6875,
+      "step": 17889
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.007746022606536e-06,
+      "loss": 0.792,
+      "step": 17890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.006385703500373e-06,
+      "loss": 0.9697,
+      "step": 17891
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0050258210463114e-06,
+      "loss": 0.8096,
+      "step": 17892
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0036663752773534e-06,
+      "loss": 0.8984,
+      "step": 17893
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0023073662264815e-06,
+      "loss": 0.9014,
+      "step": 17894
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0009487939266846e-06,
+      "loss": 0.832,
+      "step": 17895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999590658410928e-06,
+      "loss": 0.8262,
+      "step": 17896
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9982329597121696e-06,
+      "loss": 0.7842,
+      "step": 17897
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9968756978633495e-06,
+      "loss": 0.958,
+      "step": 17898
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9955188728974175e-06,
+      "loss": 0.9023,
+      "step": 17899
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9941624848472927e-06,
+      "loss": 0.832,
+      "step": 17900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.99280653374589e-06,
+      "loss": 0.6919,
+      "step": 17901
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9914510196261117e-06,
+      "loss": 0.71,
+      "step": 17902
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9900959425208597e-06,
+      "loss": 0.8994,
+      "step": 17903
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9887413024630155e-06,
+      "loss": 0.793,
+      "step": 17904
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9873870994854495e-06,
+      "loss": 0.7432,
+      "step": 17905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9860333336210202e-06,
+      "loss": 0.7939,
+      "step": 17906
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9846800049025904e-06,
+      "loss": 0.915,
+      "step": 17907
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.983327113362996e-06,
+      "loss": 0.8984,
+      "step": 17908
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9819746590350662e-06,
+      "loss": 0.7227,
+      "step": 17909
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9806226419516195e-06,
+      "loss": 0.7383,
+      "step": 17910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9792710621454712e-06,
+      "loss": 0.7739,
+      "step": 17911
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.977919919649416e-06,
+      "loss": 0.8994,
+      "step": 17912
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9765692144962446e-06,
+      "loss": 0.9824,
+      "step": 17913
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9752189467187312e-06,
+      "loss": 0.6465,
+      "step": 17914
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.973869116349647e-06,
+      "loss": 0.9209,
+      "step": 17915
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9725197234217418e-06,
+      "loss": 0.7627,
+      "step": 17916
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971170767967772e-06,
+      "loss": 0.8027,
+      "step": 17917
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9698222500204654e-06,
+      "loss": 0.8232,
+      "step": 17918
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9684741696125442e-06,
+      "loss": 0.8164,
+      "step": 17919
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.96712652677673e-06,
+      "loss": 0.7539,
+      "step": 17920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9657793215457223e-06,
+      "loss": 0.9199,
+      "step": 17921
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.964432553952211e-06,
+      "loss": 1.0605,
+      "step": 17922
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.963086224028883e-06,
+      "loss": 0.8379,
+      "step": 17923
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9617403318084083e-06,
+      "loss": 0.6665,
+      "step": 17924
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9603948773234483e-06,
+      "loss": 0.8887,
+      "step": 17925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9590498606066457e-06,
+      "loss": 0.8027,
+      "step": 17926
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9577052816906515e-06,
+      "loss": 0.7646,
+      "step": 17927
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.956361140608092e-06,
+      "loss": 0.9492,
+      "step": 17928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.95501743739158e-06,
+      "loss": 0.9023,
+      "step": 17929
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9536741720737227e-06,
+      "loss": 0.6943,
+      "step": 17930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.952331344687126e-06,
+      "loss": 0.7148,
+      "step": 17931
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9509889552643724e-06,
+      "loss": 0.7749,
+      "step": 17932
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.949647003838033e-06,
+      "loss": 0.8955,
+      "step": 17933
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9483054904406764e-06,
+      "loss": 0.8809,
+      "step": 17934
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9469644151048596e-06,
+      "loss": 0.7705,
+      "step": 17935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.945623777863124e-06,
+      "loss": 0.9473,
+      "step": 17936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.944283578748005e-06,
+      "loss": 0.7402,
+      "step": 17937
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.942943817792018e-06,
+      "loss": 0.9043,
+      "step": 17938
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9416044950276845e-06,
+      "loss": 0.8086,
+      "step": 17939
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.940265610487504e-06,
+      "loss": 0.8779,
+      "step": 17940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.938927164203963e-06,
+      "loss": 0.7373,
+      "step": 17941
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9375891562095428e-06,
+      "loss": 0.6289,
+      "step": 17942
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9362515865367147e-06,
+      "loss": 0.834,
+      "step": 17943
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.93491445521794e-06,
+      "loss": 0.9561,
+      "step": 17944
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.933577762285663e-06,
+      "loss": 0.7559,
+      "step": 17945
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.932241507772317e-06,
+      "loss": 0.8604,
+      "step": 17946
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9309056917103388e-06,
+      "loss": 0.8428,
+      "step": 17947
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9295703141321408e-06,
+      "loss": 0.8477,
+      "step": 17948
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.928235375070129e-06,
+      "loss": 0.8682,
+      "step": 17949
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9269008745566918e-06,
+      "loss": 0.8066,
+      "step": 17950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9255668126242245e-06,
+      "loss": 0.9297,
+      "step": 17951
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.924233189305094e-06,
+      "loss": 0.7754,
+      "step": 17952
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.922900004631667e-06,
+      "loss": 0.8691,
+      "step": 17953
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.92156725863629e-06,
+      "loss": 0.7861,
+      "step": 17954
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9202349513513117e-06,
+      "loss": 0.8281,
+      "step": 17955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9189030828090604e-06,
+      "loss": 0.7998,
+      "step": 17956
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9175716530418566e-06,
+      "loss": 0.9316,
+      "step": 17957
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9162406620820075e-06,
+      "loss": 0.8066,
+      "step": 17958
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9149101099618184e-06,
+      "loss": 0.8535,
+      "step": 17959
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.913579996713575e-06,
+      "loss": 0.877,
+      "step": 17960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.912250322369553e-06,
+      "loss": 0.7852,
+      "step": 17961
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9109210869620188e-06,
+      "loss": 0.7969,
+      "step": 17962
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9095922905232365e-06,
+      "loss": 0.8613,
+      "step": 17963
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9082639330854457e-06,
+      "loss": 0.6787,
+      "step": 17964
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9069360146808846e-06,
+      "loss": 0.792,
+      "step": 17965
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9056085353417741e-06,
+      "loss": 0.7046,
+      "step": 17966
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.904281495100333e-06,
+      "loss": 0.8564,
+      "step": 17967
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.902954893988762e-06,
+      "loss": 0.9277,
+      "step": 17968
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9016287320392513e-06,
+      "loss": 0.8096,
+      "step": 17969
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.900303009283988e-06,
+      "loss": 0.7344,
+      "step": 17970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8989777257551423e-06,
+      "loss": 0.8438,
+      "step": 17971
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8976528814848748e-06,
+      "loss": 0.8848,
+      "step": 17972
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8963284765053313e-06,
+      "loss": 0.9385,
+      "step": 17973
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895004510848657e-06,
+      "loss": 0.9648,
+      "step": 17974
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8936809845469794e-06,
+      "loss": 0.9502,
+      "step": 17975
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8923578976324153e-06,
+      "loss": 0.8535,
+      "step": 17976
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8910352501370677e-06,
+      "loss": 0.8721,
+      "step": 17977
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.889713042093042e-06,
+      "loss": 0.8945,
+      "step": 17978
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8883912735324216e-06,
+      "loss": 1.0293,
+      "step": 17979
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8870699444872808e-06,
+      "loss": 0.7397,
+      "step": 17980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8857490549896784e-06,
+      "loss": 0.9072,
+      "step": 17981
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8844286050716798e-06,
+      "loss": 0.792,
+      "step": 17982
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8831085947653217e-06,
+      "loss": 0.8623,
+      "step": 17983
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8817890241026404e-06,
+      "loss": 0.6719,
+      "step": 17984
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8804698931156484e-06,
+      "loss": 0.874,
+      "step": 17985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8791512018363712e-06,
+      "loss": 0.8213,
+      "step": 17986
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8778329502968028e-06,
+      "loss": 0.8682,
+      "step": 17987
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8765151385289314e-06,
+      "loss": 0.9414,
+      "step": 17988
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.875197766564736e-06,
+      "loss": 0.9033,
+      "step": 17989
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8738808344361903e-06,
+      "loss": 0.8564,
+      "step": 17990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8725643421752514e-06,
+      "loss": 0.8945,
+      "step": 17991
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8712482898138652e-06,
+      "loss": 0.8408,
+      "step": 17992
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8699326773839631e-06,
+      "loss": 0.7451,
+      "step": 17993
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8686175049174804e-06,
+      "loss": 0.7725,
+      "step": 17994
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8673027724463288e-06,
+      "loss": 0.8408,
+      "step": 17995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8659884800024119e-06,
+      "loss": 0.7207,
+      "step": 17996
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8646746276176197e-06,
+      "loss": 0.7598,
+      "step": 17997
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8633612153238468e-06,
+      "loss": 0.8262,
+      "step": 17998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8620482431529563e-06,
+      "loss": 0.7607,
+      "step": 17999
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8607357111368118e-06,
+      "loss": 0.833,
+      "step": 18000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.859423619307268e-06,
+      "loss": 0.7656,
+      "step": 18001
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8581119676961613e-06,
+      "loss": 0.9385,
+      "step": 18002
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8568007563353263e-06,
+      "loss": 0.7646,
+      "step": 18003
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8554899852565799e-06,
+      "loss": 0.7451,
+      "step": 18004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.85417965449173e-06,
+      "loss": 0.8418,
+      "step": 18005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8528697640725713e-06,
+      "loss": 0.7817,
+      "step": 18006
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8515603140308979e-06,
+      "loss": 0.7676,
+      "step": 18007
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8502513043984848e-06,
+      "loss": 0.8516,
+      "step": 18008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.848942735207091e-06,
+      "loss": 0.8418,
+      "step": 18009
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.847634606488482e-06,
+      "loss": 0.8965,
+      "step": 18010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8463269182743949e-06,
+      "loss": 0.8115,
+      "step": 18011
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8450196705965663e-06,
+      "loss": 0.7866,
+      "step": 18012
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8437128634867153e-06,
+      "loss": 0.9189,
+      "step": 18013
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.842406496976561e-06,
+      "loss": 0.9404,
+      "step": 18014
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8411005710978002e-06,
+      "loss": 0.9062,
+      "step": 18015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8397950858821255e-06,
+      "loss": 0.8213,
+      "step": 18016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8384900413612138e-06,
+      "loss": 0.7617,
+      "step": 18017
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8371854375667419e-06,
+      "loss": 0.918,
+      "step": 18018
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8358812745303645e-06,
+      "loss": 0.8359,
+      "step": 18019
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8345775522837273e-06,
+      "loss": 0.959,
+      "step": 18020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8332742708584695e-06,
+      "loss": 0.8252,
+      "step": 18021
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8319714302862212e-06,
+      "loss": 0.6921,
+      "step": 18022
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8306690305985952e-06,
+      "loss": 0.7471,
+      "step": 18023
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8293670718271993e-06,
+      "loss": 0.7979,
+      "step": 18024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8280655540036218e-06,
+      "loss": 0.7852,
+      "step": 18025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8267644771594574e-06,
+      "loss": 0.748,
+      "step": 18026
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8254638413262715e-06,
+      "loss": 0.8574,
+      "step": 18027
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8241636465356284e-06,
+      "loss": 0.854,
+      "step": 18028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8228638928190777e-06,
+      "loss": 0.8218,
+      "step": 18029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8215645802081682e-06,
+      "loss": 0.8408,
+      "step": 18030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8202657087344234e-06,
+      "loss": 0.7549,
+      "step": 18031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8189672784293667e-06,
+      "loss": 0.7158,
+      "step": 18032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8176692893244997e-06,
+      "loss": 0.8262,
+      "step": 18033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8163717414513326e-06,
+      "loss": 0.8447,
+      "step": 18034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8150746348413473e-06,
+      "loss": 0.8809,
+      "step": 18035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8137779695260182e-06,
+      "loss": 0.7344,
+      "step": 18036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8124817455368139e-06,
+      "loss": 0.7163,
+      "step": 18037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8111859629051909e-06,
+      "loss": 0.9043,
+      "step": 18038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8098906216625934e-06,
+      "loss": 0.7764,
+      "step": 18039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.808595721840456e-06,
+      "loss": 0.5996,
+      "step": 18040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.807301263470198e-06,
+      "loss": 0.9639,
+      "step": 18041
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.806007246583239e-06,
+      "loss": 0.7549,
+      "step": 18042
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8047136712109781e-06,
+      "loss": 0.833,
+      "step": 18043
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8034205373848056e-06,
+      "loss": 0.8848,
+      "step": 18044
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.802127845136099e-06,
+      "loss": 0.7363,
+      "step": 18045
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8008355944962352e-06,
+      "loss": 0.915,
+      "step": 18046
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7995437854965693e-06,
+      "loss": 0.6816,
+      "step": 18047
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.798252418168449e-06,
+      "loss": 0.875,
+      "step": 18048
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7969614925432122e-06,
+      "loss": 0.8926,
+      "step": 18049
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.795671008652189e-06,
+      "loss": 0.9082,
+      "step": 18050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.794380966526692e-06,
+      "loss": 0.9512,
+      "step": 18051
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7930913661980253e-06,
+      "loss": 0.7207,
+      "step": 18052
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7918022076974906e-06,
+      "loss": 0.918,
+      "step": 18053
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7905134910563694e-06,
+      "loss": 0.8027,
+      "step": 18054
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.789225216305932e-06,
+      "loss": 0.8418,
+      "step": 18055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7879373834774382e-06,
+      "loss": 0.7656,
+      "step": 18056
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7866499926021496e-06,
+      "loss": 0.9072,
+      "step": 18057
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7853630437113012e-06,
+      "loss": 0.7666,
+      "step": 18058
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7840765368361257e-06,
+      "loss": 0.8252,
+      "step": 18059
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.782790472007836e-06,
+      "loss": 0.9795,
+      "step": 18060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7815048492576515e-06,
+      "loss": 0.6284,
+      "step": 18061
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.780219668616765e-06,
+      "loss": 0.9434,
+      "step": 18062
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.778934930116365e-06,
+      "loss": 0.9268,
+      "step": 18063
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7776506337876243e-06,
+      "loss": 0.7197,
+      "step": 18064
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.776366779661718e-06,
+      "loss": 0.9561,
+      "step": 18065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7750833677697942e-06,
+      "loss": 0.7109,
+      "step": 18066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7738003981429997e-06,
+      "loss": 0.8037,
+      "step": 18067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7725178708124647e-06,
+      "loss": 0.9414,
+      "step": 18068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7712357858093198e-06,
+      "loss": 0.8447,
+      "step": 18069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7699541431646738e-06,
+      "loss": 0.7168,
+      "step": 18070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7686729429096283e-06,
+      "loss": 0.9258,
+      "step": 18071
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7673921850752696e-06,
+      "loss": 0.7686,
+      "step": 18072
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.766111869692686e-06,
+      "loss": 0.7549,
+      "step": 18073
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.764831996792944e-06,
+      "loss": 0.8086,
+      "step": 18074
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7635525664071008e-06,
+      "loss": 0.9658,
+      "step": 18075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7622735785662027e-06,
+      "loss": 0.7822,
+      "step": 18076
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7609950333012916e-06,
+      "loss": 0.9043,
+      "step": 18077
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7597169306433936e-06,
+      "loss": 0.8984,
+      "step": 18078
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.758439270623522e-06,
+      "loss": 0.9268,
+      "step": 18079
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7571620532726808e-06,
+      "loss": 0.7461,
+      "step": 18080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7558852786218695e-06,
+      "loss": 0.6733,
+      "step": 18081
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7546089467020677e-06,
+      "loss": 0.7764,
+      "step": 18082
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7533330575442464e-06,
+      "loss": 0.7363,
+      "step": 18083
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.752057611179372e-06,
+      "loss": 1.0,
+      "step": 18084
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7507826076383928e-06,
+      "loss": 0.957,
+      "step": 18085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.749508046952253e-06,
+      "loss": 0.8545,
+      "step": 18086
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7482339291518791e-06,
+      "loss": 0.7285,
+      "step": 18087
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7469602542681929e-06,
+      "loss": 0.9004,
+      "step": 18088
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7456870223320964e-06,
+      "loss": 0.7617,
+      "step": 18089
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7444142333744962e-06,
+      "loss": 0.9541,
+      "step": 18090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.743141887426274e-06,
+      "loss": 0.7959,
+      "step": 18091
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7418699845183051e-06,
+      "loss": 0.8516,
+      "step": 18092
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7405985246814584e-06,
+      "loss": 0.9004,
+      "step": 18093
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7393275079465866e-06,
+      "loss": 0.8398,
+      "step": 18094
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7380569343445342e-06,
+      "loss": 0.7852,
+      "step": 18095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7367868039061297e-06,
+      "loss": 0.7793,
+      "step": 18096
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.735517116662202e-06,
+      "loss": 0.8154,
+      "step": 18097
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7342478726435596e-06,
+      "loss": 0.7256,
+      "step": 18098
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7329790718810046e-06,
+      "loss": 0.7075,
+      "step": 18099
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7317107144053237e-06,
+      "loss": 0.9512,
+      "step": 18100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7304428002473006e-06,
+      "loss": 0.9873,
+      "step": 18101
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7291753294377022e-06,
+      "loss": 0.9248,
+      "step": 18102
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7279083020072862e-06,
+      "loss": 0.7891,
+      "step": 18103
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7266417179867967e-06,
+      "loss": 0.7734,
+      "step": 18104
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7253755774069759e-06,
+      "loss": 0.7656,
+      "step": 18105
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7241098802985456e-06,
+      "loss": 0.793,
+      "step": 18106
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7228446266922217e-06,
+      "loss": 0.6255,
+      "step": 18107
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7215798166187015e-06,
+      "loss": 0.9004,
+      "step": 18108
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7203154501086917e-06,
+      "loss": 0.8281,
+      "step": 18109
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7190515271928633e-06,
+      "loss": 0.749,
+      "step": 18110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.717788047901896e-06,
+      "loss": 0.7715,
+      "step": 18111
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7165250122664413e-06,
+      "loss": 0.8057,
+      "step": 18112
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7152624203171564e-06,
+      "loss": 0.8174,
+      "step": 18113
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7140002720846816e-06,
+      "loss": 0.8525,
+      "step": 18114
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7127385675996434e-06,
+      "loss": 0.7422,
+      "step": 18115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7114773068926528e-06,
+      "loss": 0.7949,
+      "step": 18116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7102164899943296e-06,
+      "loss": 0.8052,
+      "step": 18117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.708956116935263e-06,
+      "loss": 0.8154,
+      "step": 18118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7076961877460395e-06,
+      "loss": 0.7988,
+      "step": 18119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7064367024572303e-06,
+      "loss": 0.9141,
+      "step": 18120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7051776610994065e-06,
+      "loss": 0.9004,
+      "step": 18121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7039190637031167e-06,
+      "loss": 0.9014,
+      "step": 18122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7026609102989055e-06,
+      "loss": 0.7188,
+      "step": 18123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7014032009172997e-06,
+      "loss": 0.6406,
+      "step": 18124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.700145935588826e-06,
+      "loss": 0.6904,
+      "step": 18125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6988891143439956e-06,
+      "loss": 0.9834,
+      "step": 18126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6976327372133017e-06,
+      "loss": 0.9443,
+      "step": 18127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6963768042272355e-06,
+      "loss": 0.8857,
+      "step": 18128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.695121315416277e-06,
+      "loss": 0.793,
+      "step": 18129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6938662708108911e-06,
+      "loss": 0.8994,
+      "step": 18130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6926116704415353e-06,
+      "loss": 0.8789,
+      "step": 18131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6913575143386496e-06,
+      "loss": 0.918,
+      "step": 18132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6901038025326765e-06,
+      "loss": 0.7158,
+      "step": 18133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6888505350540384e-06,
+      "loss": 0.7881,
+      "step": 18134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6875977119331466e-06,
+      "loss": 0.8457,
+      "step": 18135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6863453332003988e-06,
+      "loss": 0.8398,
+      "step": 18136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.685093398886195e-06,
+      "loss": 0.8711,
+      "step": 18137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6838419090209113e-06,
+      "loss": 0.9131,
+      "step": 18138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6825908636349164e-06,
+      "loss": 0.873,
+      "step": 18139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.681340262758575e-06,
+      "loss": 0.9111,
+      "step": 18140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6800901064222319e-06,
+      "loss": 0.7773,
+      "step": 18141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.678840394656225e-06,
+      "loss": 0.8721,
+      "step": 18142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6775911274908784e-06,
+      "loss": 0.793,
+      "step": 18143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.676342304956513e-06,
+      "loss": 0.731,
+      "step": 18144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6750939270834331e-06,
+      "loss": 0.7568,
+      "step": 18145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6738459939019303e-06,
+      "loss": 0.9463,
+      "step": 18146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.672598505442289e-06,
+      "loss": 0.7783,
+      "step": 18147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6713514617347848e-06,
+      "loss": 0.7617,
+      "step": 18148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6701048628096783e-06,
+      "loss": 0.6299,
+      "step": 18149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6688587086972207e-06,
+      "loss": 1.1094,
+      "step": 18150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667612999427648e-06,
+      "loss": 0.834,
+      "step": 18151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6663677350312002e-06,
+      "loss": 0.8398,
+      "step": 18152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6651229155380865e-06,
+      "loss": 0.7778,
+      "step": 18153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6638785409785208e-06,
+      "loss": 0.918,
+      "step": 18154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6626346113826942e-06,
+      "loss": 0.877,
+      "step": 18155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6613911267808025e-06,
+      "loss": 0.9258,
+      "step": 18156
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6601480872030152e-06,
+      "loss": 0.8535,
+      "step": 18157
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6589054926794967e-06,
+      "loss": 0.8291,
+      "step": 18158
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.657663343240401e-06,
+      "loss": 0.7881,
+      "step": 18159
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.656421638915877e-06,
+      "loss": 0.8242,
+      "step": 18160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.655180379736052e-06,
+      "loss": 0.8594,
+      "step": 18161
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6539395657310487e-06,
+      "loss": 0.6401,
+      "step": 18162
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6526991969309757e-06,
+      "loss": 0.9902,
+      "step": 18163
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6514592733659385e-06,
+      "loss": 0.9473,
+      "step": 18164
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.650219795066026e-06,
+      "loss": 0.9277,
+      "step": 18165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6489807620613075e-06,
+      "loss": 0.8672,
+      "step": 18166
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6477421743818634e-06,
+      "loss": 0.8623,
+      "step": 18167
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6465040320577408e-06,
+      "loss": 0.8135,
+      "step": 18168
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6452663351189913e-06,
+      "loss": 0.8428,
+      "step": 18169
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6440290835956507e-06,
+      "loss": 0.8633,
+      "step": 18170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6427922775177396e-06,
+      "loss": 0.8721,
+      "step": 18171
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6415559169152717e-06,
+      "loss": 0.9482,
+      "step": 18172
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6403200018182518e-06,
+      "loss": 0.8896,
+      "step": 18173
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.639084532256674e-06,
+      "loss": 1.0068,
+      "step": 18174
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6378495082605184e-06,
+      "loss": 0.8408,
+      "step": 18175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6366149298597477e-06,
+      "loss": 0.9795,
+      "step": 18176
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.635380797084334e-06,
+      "loss": 0.8584,
+      "step": 18177
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.634147109964217e-06,
+      "loss": 0.7842,
+      "step": 18178
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6329138685293354e-06,
+      "loss": 0.8359,
+      "step": 18179
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6316810728096232e-06,
+      "loss": 0.8535,
+      "step": 18180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.630448722834992e-06,
+      "loss": 0.9346,
+      "step": 18181
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6292168186353463e-06,
+      "loss": 0.7705,
+      "step": 18182
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6279853602405783e-06,
+      "loss": 0.7812,
+      "step": 18183
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6267543476805792e-06,
+      "loss": 0.9248,
+      "step": 18184
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6255237809852188e-06,
+      "loss": 0.8047,
+      "step": 18185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6242936601843573e-06,
+      "loss": 0.5688,
+      "step": 18186
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6230639853078445e-06,
+      "loss": 0.8623,
+      "step": 18187
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6218347563855298e-06,
+      "loss": 0.9326,
+      "step": 18188
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6206059734472357e-06,
+      "loss": 0.8369,
+      "step": 18189
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6193776365227831e-06,
+      "loss": 0.8906,
+      "step": 18190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.618149745641977e-06,
+      "loss": 0.8633,
+      "step": 18191
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6169223008346202e-06,
+      "loss": 0.7656,
+      "step": 18192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6156953021304955e-06,
+      "loss": 0.8604,
+      "step": 18193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6144687495593814e-06,
+      "loss": 0.9023,
+      "step": 18194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6132426431510362e-06,
+      "loss": 0.8311,
+      "step": 18195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6120169829352228e-06,
+      "loss": 0.7803,
+      "step": 18196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6107917689416797e-06,
+      "loss": 0.8359,
+      "step": 18197
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6095670012001385e-06,
+      "loss": 0.9717,
+      "step": 18198
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6083426797403202e-06,
+      "loss": 0.9199,
+      "step": 18199
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6071188045919385e-06,
+      "loss": 0.7334,
+      "step": 18200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.605895375784694e-06,
+      "loss": 0.7598,
+      "step": 18201
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6046723933482722e-06,
+      "loss": 0.9238,
+      "step": 18202
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6034498573123513e-06,
+      "loss": 0.7646,
+      "step": 18203
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6022277677066012e-06,
+      "loss": 0.9424,
+      "step": 18204
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6010061245606778e-06,
+      "loss": 0.7812,
+      "step": 18205
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5997849279042265e-06,
+      "loss": 0.8164,
+      "step": 18206
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5985641777668793e-06,
+      "loss": 0.7598,
+      "step": 18207
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5973438741782653e-06,
+      "loss": 0.7773,
+      "step": 18208
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.596124017167997e-06,
+      "loss": 0.9541,
+      "step": 18209
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5949046067656726e-06,
+      "loss": 0.71,
+      "step": 18210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.593685643000884e-06,
+      "loss": 0.792,
+      "step": 18211
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5924671259032165e-06,
+      "loss": 0.8145,
+      "step": 18212
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5912490555022375e-06,
+      "loss": 0.8105,
+      "step": 18213
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5900314318275057e-06,
+      "loss": 0.9648,
+      "step": 18214
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.588814254908566e-06,
+      "loss": 0.7363,
+      "step": 18215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.587597524774962e-06,
+      "loss": 0.9805,
+      "step": 18216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5863812414562164e-06,
+      "loss": 0.7832,
+      "step": 18217
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5851654049818477e-06,
+      "loss": 0.8262,
+      "step": 18218
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5839500153813525e-06,
+      "loss": 0.9551,
+      "step": 18219
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.582735072684236e-06,
+      "loss": 0.668,
+      "step": 18220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5815205769199749e-06,
+      "loss": 0.6641,
+      "step": 18221
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5803065281180452e-06,
+      "loss": 0.873,
+      "step": 18222
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.579092926307899e-06,
+      "loss": 0.9883,
+      "step": 18223
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5778797715189997e-06,
+      "loss": 0.7998,
+      "step": 18224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.576667063780779e-06,
+      "loss": 0.874,
+      "step": 18225
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5754548031226646e-06,
+      "loss": 0.9668,
+      "step": 18226
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5742429895740796e-06,
+      "loss": 0.9609,
+      "step": 18227
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5730316231644316e-06,
+      "loss": 0.7148,
+      "step": 18228
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5718207039231126e-06,
+      "loss": 0.7861,
+      "step": 18229
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.570610231879508e-06,
+      "loss": 0.8203,
+      "step": 18230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5694002070629966e-06,
+      "loss": 0.8242,
+      "step": 18231
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5681906295029392e-06,
+      "loss": 0.749,
+      "step": 18232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5669814992286903e-06,
+      "loss": 0.8125,
+      "step": 18233
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5657728162695862e-06,
+      "loss": 1.0098,
+      "step": 18234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5645645806549658e-06,
+      "loss": 0.8086,
+      "step": 18235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.563356792414148e-06,
+      "loss": 0.8896,
+      "step": 18236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5621494515764402e-06,
+      "loss": 0.8633,
+      "step": 18237
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5609425581711368e-06,
+      "loss": 0.9004,
+      "step": 18238
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5597361122275344e-06,
+      "loss": 0.708,
+      "step": 18239
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5585301137749075e-06,
+      "loss": 0.8096,
+      "step": 18240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5573245628425194e-06,
+      "loss": 0.8145,
+      "step": 18241
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5561194594596219e-06,
+      "loss": 0.9443,
+      "step": 18242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5549148036554673e-06,
+      "loss": 0.8516,
+      "step": 18243
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5537105954592835e-06,
+      "loss": 0.7607,
+      "step": 18244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5525068349002982e-06,
+      "loss": 0.7559,
+      "step": 18245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5513035220077143e-06,
+      "loss": 0.7422,
+      "step": 18246
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5501006568107425e-06,
+      "loss": 0.8613,
+      "step": 18247
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5488982393385677e-06,
+      "loss": 0.6626,
+      "step": 18248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5476962696203713e-06,
+      "loss": 0.7603,
+      "step": 18249
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5464947476853164e-06,
+      "loss": 0.6826,
+      "step": 18250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5452936735625646e-06,
+      "loss": 0.8525,
+      "step": 18251
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5440930472812633e-06,
+      "loss": 0.9453,
+      "step": 18252
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5428928688705492e-06,
+      "loss": 0.749,
+      "step": 18253
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5416931383595436e-06,
+      "loss": 0.7212,
+      "step": 18254
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5404938557773585e-06,
+      "loss": 0.6143,
+      "step": 18255
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5392950211531045e-06,
+      "loss": 0.6001,
+      "step": 18256
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5380966345158666e-06,
+      "loss": 0.8936,
+      "step": 18257
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5368986958947307e-06,
+      "loss": 0.7588,
+      "step": 18258
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5357012053187626e-06,
+      "loss": 0.8408,
+      "step": 18259
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5345041628170276e-06,
+      "loss": 0.7949,
+      "step": 18260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5333075684185716e-06,
+      "loss": 0.8867,
+      "step": 18261
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5321114221524315e-06,
+      "loss": 0.5273,
+      "step": 18262
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5309157240476324e-06,
+      "loss": 0.7959,
+      "step": 18263
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5297204741331962e-06,
+      "loss": 0.7334,
+      "step": 18264
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5285256724381236e-06,
+      "loss": 0.8896,
+      "step": 18265
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5273313189914096e-06,
+      "loss": 0.9463,
+      "step": 18266
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5261374138220397e-06,
+      "loss": 0.7754,
+      "step": 18267
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5249439569589863e-06,
+      "loss": 0.7461,
+      "step": 18268
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5237509484312085e-06,
+      "loss": 0.9893,
+      "step": 18269
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5225583882676563e-06,
+      "loss": 0.792,
+      "step": 18270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5213662764972736e-06,
+      "loss": 0.8799,
+      "step": 18271
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5201746131489903e-06,
+      "loss": 0.9131,
+      "step": 18272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.518983398251721e-06,
+      "loss": 0.9775,
+      "step": 18273
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5177926318343693e-06,
+      "loss": 0.9248,
+      "step": 18274
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5166023139258411e-06,
+      "loss": 0.9023,
+      "step": 18275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5154124445550155e-06,
+      "loss": 0.6304,
+      "step": 18276
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5142230237507694e-06,
+      "loss": 0.7354,
+      "step": 18277
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5130340515419617e-06,
+      "loss": 0.9258,
+      "step": 18278
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.511845527957454e-06,
+      "loss": 0.8262,
+      "step": 18279
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.510657453026083e-06,
+      "loss": 0.7959,
+      "step": 18280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5094698267766817e-06,
+      "loss": 0.7651,
+      "step": 18281
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5082826492380642e-06,
+      "loss": 0.7998,
+      "step": 18282
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5070959204390478e-06,
+      "loss": 0.9053,
+      "step": 18283
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5059096404084273e-06,
+      "loss": 0.9307,
+      "step": 18284
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5047238091749928e-06,
+      "loss": 0.9121,
+      "step": 18285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5035384267675124e-06,
+      "loss": 1.0,
+      "step": 18286
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5023534932147654e-06,
+      "loss": 0.9932,
+      "step": 18287
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5011690085454955e-06,
+      "loss": 0.918,
+      "step": 18288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.499984972788453e-06,
+      "loss": 0.8154,
+      "step": 18289
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.498801385972366e-06,
+      "loss": 0.7725,
+      "step": 18290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4976182481259606e-06,
+      "loss": 0.8379,
+      "step": 18291
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4964355592779489e-06,
+      "loss": 0.8574,
+      "step": 18292
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4952533194570262e-06,
+      "loss": 0.8555,
+      "step": 18293
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.494071528691885e-06,
+      "loss": 0.6421,
+      "step": 18294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4928901870112046e-06,
+      "loss": 0.8574,
+      "step": 18295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4917092944436529e-06,
+      "loss": 0.877,
+      "step": 18296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.490528851017885e-06,
+      "loss": 0.874,
+      "step": 18297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4893488567625447e-06,
+      "loss": 1.0234,
+      "step": 18298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4881693117062734e-06,
+      "loss": 0.793,
+      "step": 18299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4869902158776927e-06,
+      "loss": 0.9512,
+      "step": 18300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4858115693054131e-06,
+      "loss": 0.8672,
+      "step": 18301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4846333720180338e-06,
+      "loss": 0.6929,
+      "step": 18302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4834556240441567e-06,
+      "loss": 0.9482,
+      "step": 18303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4822783254123562e-06,
+      "loss": 0.8135,
+      "step": 18304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.481101476151201e-06,
+      "loss": 0.9492,
+      "step": 18305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4799250762892481e-06,
+      "loss": 0.7061,
+      "step": 18306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4787491258550503e-06,
+      "loss": 0.9072,
+      "step": 18307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.477573624877142e-06,
+      "loss": 0.8223,
+      "step": 18308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.476398573384048e-06,
+      "loss": 0.9209,
+      "step": 18309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.475223971404287e-06,
+      "loss": 0.5859,
+      "step": 18310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4740498189663611e-06,
+      "loss": 0.9004,
+      "step": 18311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4728761160987626e-06,
+      "loss": 0.8809,
+      "step": 18312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4717028628299712e-06,
+      "loss": 0.8389,
+      "step": 18313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4705300591884642e-06,
+      "loss": 0.8799,
+      "step": 18314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4693577052027008e-06,
+      "loss": 0.9512,
+      "step": 18315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4681858009011274e-06,
+      "loss": 0.8008,
+      "step": 18316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4670143463121834e-06,
+      "loss": 0.9082,
+      "step": 18317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4658433414642992e-06,
+      "loss": 0.9639,
+      "step": 18318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4646727863858922e-06,
+      "loss": 0.7744,
+      "step": 18319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4635026811053643e-06,
+      "loss": 0.8809,
+      "step": 18320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4623330256511104e-06,
+      "loss": 0.7236,
+      "step": 18321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4611638200515188e-06,
+      "loss": 0.8018,
+      "step": 18322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4599950643349624e-06,
+      "loss": 0.9844,
+      "step": 18323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4588267585297988e-06,
+      "loss": 0.749,
+      "step": 18324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4576589026643805e-06,
+      "loss": 0.8506,
+      "step": 18325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.456491496767054e-06,
+      "loss": 0.8311,
+      "step": 18326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4553245408661432e-06,
+      "loss": 0.9795,
+      "step": 18327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4541580349899676e-06,
+      "loss": 0.8896,
+      "step": 18328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4529919791668313e-06,
+      "loss": 0.8193,
+      "step": 18329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4518263734250404e-06,
+      "loss": 0.9346,
+      "step": 18330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4506612177928726e-06,
+      "loss": 1.0361,
+      "step": 18331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4494965122986072e-06,
+      "loss": 0.8066,
+      "step": 18332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4483322569705016e-06,
+      "loss": 0.916,
+      "step": 18333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4471684518368157e-06,
+      "loss": 0.9287,
+      "step": 18334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.446005096925791e-06,
+      "loss": 0.791,
+      "step": 18335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4448421922656563e-06,
+      "loss": 0.9033,
+      "step": 18336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4436797378846356e-06,
+      "loss": 0.8193,
+      "step": 18337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.442517733810931e-06,
+      "loss": 0.7646,
+      "step": 18338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4413561800727482e-06,
+      "loss": 0.9365,
+      "step": 18339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.440195076698272e-06,
+      "loss": 0.8057,
+      "step": 18340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4390344237156795e-06,
+      "loss": 0.8496,
+      "step": 18341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4378742211531328e-06,
+      "loss": 0.7852,
+      "step": 18342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4367144690387914e-06,
+      "loss": 0.8164,
+      "step": 18343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4355551674007973e-06,
+      "loss": 0.9922,
+      "step": 18344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4343963162672859e-06,
+      "loss": 0.7178,
+      "step": 18345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.43323791566637e-06,
+      "loss": 0.8965,
+      "step": 18346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.432079965626172e-06,
+      "loss": 0.8633,
+      "step": 18347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4309224661747866e-06,
+      "loss": 0.8184,
+      "step": 18348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4297654173402985e-06,
+      "loss": 0.7812,
+      "step": 18349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4286088191507963e-06,
+      "loss": 0.9893,
+      "step": 18350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4274526716343396e-06,
+      "loss": 0.6201,
+      "step": 18351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4262969748189882e-06,
+      "loss": 1.0215,
+      "step": 18352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4251417287327818e-06,
+      "loss": 0.7891,
+      "step": 18353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4239869334037605e-06,
+      "loss": 0.7676,
+      "step": 18354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.422832588859948e-06,
+      "loss": 0.835,
+      "step": 18355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4216786951293559e-06,
+      "loss": 0.8174,
+      "step": 18356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4205252522399815e-06,
+      "loss": 0.748,
+      "step": 18357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4193722602198202e-06,
+      "loss": 0.7998,
+      "step": 18358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4182197190968515e-06,
+      "loss": 0.8525,
+      "step": 18359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4170676288990448e-06,
+      "loss": 0.7207,
+      "step": 18360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4159159896543528e-06,
+      "loss": 0.8711,
+      "step": 18361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4147648013907267e-06,
+      "loss": 0.9795,
+      "step": 18362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.413614064136104e-06,
+      "loss": 0.79,
+      "step": 18363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.412463777918407e-06,
+      "loss": 0.8838,
+      "step": 18364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4113139427655464e-06,
+      "loss": 0.9082,
+      "step": 18365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4101645587054313e-06,
+      "loss": 0.8516,
+      "step": 18366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4090156257659526e-06,
+      "loss": 0.8271,
+      "step": 18367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4078671439749902e-06,
+      "loss": 0.8457,
+      "step": 18368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.406719113360413e-06,
+      "loss": 1.002,
+      "step": 18369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.405571533950083e-06,
+      "loss": 0.7734,
+      "step": 18370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4044244057718492e-06,
+      "loss": 0.6489,
+      "step": 18371
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4032777288535471e-06,
+      "loss": 0.9111,
+      "step": 18372
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.402131503223001e-06,
+      "loss": 0.9678,
+      "step": 18373
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4009857289080309e-06,
+      "loss": 0.7505,
+      "step": 18374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.399840405936439e-06,
+      "loss": 0.8965,
+      "step": 18375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3986955343360209e-06,
+      "loss": 0.9141,
+      "step": 18376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3975511141345544e-06,
+      "loss": 0.8975,
+      "step": 18377
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3964071453598171e-06,
+      "loss": 0.8271,
+      "step": 18378
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.395263628039567e-06,
+      "loss": 0.7476,
+      "step": 18379
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.394120562201553e-06,
+      "loss": 0.9434,
+      "step": 18380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.392977947873515e-06,
+      "loss": 0.877,
+      "step": 18381
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.391835785083182e-06,
+      "loss": 0.6953,
+      "step": 18382
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3906940738582698e-06,
+      "loss": 0.8057,
+      "step": 18383
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3895528142264847e-06,
+      "loss": 0.8711,
+      "step": 18384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3884120062155183e-06,
+      "loss": 0.9004,
+      "step": 18385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3872716498530615e-06,
+      "loss": 0.9023,
+      "step": 18386
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3861317451667833e-06,
+      "loss": 0.9863,
+      "step": 18387
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3849922921843462e-06,
+      "loss": 0.7773,
+      "step": 18388
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3838532909333968e-06,
+      "loss": 0.9111,
+      "step": 18389
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3827147414415842e-06,
+      "loss": 0.7246,
+      "step": 18390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3815766437365353e-06,
+      "loss": 0.6982,
+      "step": 18391
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3804389978458633e-06,
+      "loss": 0.8887,
+      "step": 18392
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3793018037971772e-06,
+      "loss": 0.9189,
+      "step": 18393
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3781650616180776e-06,
+      "loss": 0.8604,
+      "step": 18394
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3770287713361463e-06,
+      "loss": 0.8896,
+      "step": 18395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3758929329789573e-06,
+      "loss": 0.8164,
+      "step": 18396
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3747575465740771e-06,
+      "loss": 0.8887,
+      "step": 18397
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.373622612149057e-06,
+      "loss": 0.9541,
+      "step": 18398
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3724881297314373e-06,
+      "loss": 0.9688,
+      "step": 18399
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3713540993487473e-06,
+      "loss": 0.9502,
+      "step": 18400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.370220521028509e-06,
+      "loss": 0.9404,
+      "step": 18401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.369087394798232e-06,
+      "loss": 0.874,
+      "step": 18402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3679547206854137e-06,
+      "loss": 0.7998,
+      "step": 18403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.366822498717535e-06,
+      "loss": 0.7881,
+      "step": 18404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3656907289220777e-06,
+      "loss": 0.8584,
+      "step": 18405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3645594113265071e-06,
+      "loss": 0.8018,
+      "step": 18406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.363428545958274e-06,
+      "loss": 0.6851,
+      "step": 18407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3622981328448193e-06,
+      "loss": 0.8994,
+      "step": 18408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3611681720135806e-06,
+      "loss": 0.8496,
+      "step": 18409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3600386634919737e-06,
+      "loss": 0.8359,
+      "step": 18410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3589096073074126e-06,
+      "loss": 0.9844,
+      "step": 18411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3577810034872885e-06,
+      "loss": 0.8037,
+      "step": 18412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3566528520589994e-06,
+      "loss": 0.832,
+      "step": 18413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.355525153049917e-06,
+      "loss": 0.9473,
+      "step": 18414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3543979064874079e-06,
+      "loss": 0.8896,
+      "step": 18415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.353271112398824e-06,
+      "loss": 0.7471,
+      "step": 18416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3521447708115121e-06,
+      "loss": 0.9248,
+      "step": 18417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3510188817528059e-06,
+      "loss": 1.0508,
+      "step": 18418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3498934452500302e-06,
+      "loss": 0.9346,
+      "step": 18419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3487684613304919e-06,
+      "loss": 0.5181,
+      "step": 18420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3476439300214872e-06,
+      "loss": 0.8369,
+      "step": 18421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3465198513503143e-06,
+      "loss": 0.834,
+      "step": 18422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3453962253442444e-06,
+      "loss": 0.8262,
+      "step": 18423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3442730520305492e-06,
+      "loss": 0.8594,
+      "step": 18424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.343150331436478e-06,
+      "loss": 0.7344,
+      "step": 18425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3420280635892847e-06,
+      "loss": 0.731,
+      "step": 18426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3409062485161983e-06,
+      "loss": 0.7676,
+      "step": 18427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.339784886244444e-06,
+      "loss": 0.8662,
+      "step": 18428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3386639768012288e-06,
+      "loss": 0.7666,
+      "step": 18429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3375435202137598e-06,
+      "loss": 0.8057,
+      "step": 18430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3364235165092265e-06,
+      "loss": 0.8213,
+      "step": 18431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3353039657148071e-06,
+      "loss": 0.9629,
+      "step": 18432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3341848678576663e-06,
+      "loss": 0.7988,
+      "step": 18433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3330662229649671e-06,
+      "loss": 0.8467,
+      "step": 18434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3319480310638544e-06,
+      "loss": 0.8486,
+      "step": 18435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3308302921814576e-06,
+      "loss": 0.9385,
+      "step": 18436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3297130063449105e-06,
+      "loss": 0.8086,
+      "step": 18437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3285961735813203e-06,
+      "loss": 0.835,
+      "step": 18438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3274797939177897e-06,
+      "loss": 0.8047,
+      "step": 18439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3263638673814084e-06,
+      "loss": 0.9336,
+      "step": 18440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.325248393999261e-06,
+      "loss": 0.6875,
+      "step": 18441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3241333737984152e-06,
+      "loss": 0.9131,
+      "step": 18442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.323018806805929e-06,
+      "loss": 0.9033,
+      "step": 18443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3219046930488477e-06,
+      "loss": 0.8076,
+      "step": 18444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3207910325542094e-06,
+      "loss": 0.7646,
+      "step": 18445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3196778253490417e-06,
+      "loss": 0.5898,
+      "step": 18446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3185650714603537e-06,
+      "loss": 0.9023,
+      "step": 18447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3174527709151508e-06,
+      "loss": 0.9561,
+      "step": 18448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3163409237404268e-06,
+      "loss": 0.959,
+      "step": 18449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3152295299631624e-06,
+      "loss": 0.8555,
+      "step": 18450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.314118589610327e-06,
+      "loss": 0.9619,
+      "step": 18451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3130081027088793e-06,
+      "loss": 0.8926,
+      "step": 18452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3118980692857686e-06,
+      "loss": 0.8867,
+      "step": 18453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3107884893679334e-06,
+      "loss": 1.0059,
+      "step": 18454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.309679362982299e-06,
+      "loss": 0.7124,
+      "step": 18455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.308570690155777e-06,
+      "loss": 0.8711,
+      "step": 18456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3074624709152773e-06,
+      "loss": 0.7461,
+      "step": 18457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3063547052876912e-06,
+      "loss": 0.6045,
+      "step": 18458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3052473932999e-06,
+      "loss": 0.8398,
+      "step": 18459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3041405349787728e-06,
+      "loss": 0.9297,
+      "step": 18460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3030341303511752e-06,
+      "loss": 0.9443,
+      "step": 18461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.301928179443952e-06,
+      "loss": 0.8145,
+      "step": 18462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3008226822839443e-06,
+      "loss": 0.8203,
+      "step": 18463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2997176388979748e-06,
+      "loss": 0.9375,
+      "step": 18464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2986130493128668e-06,
+      "loss": 0.8506,
+      "step": 18465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2975089135554187e-06,
+      "loss": 0.8672,
+      "step": 18466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2964052316524288e-06,
+      "loss": 0.7605,
+      "step": 18467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2953020036306741e-06,
+      "loss": 0.8154,
+      "step": 18468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2941992295169369e-06,
+      "loss": 1.0059,
+      "step": 18469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2930969093379698e-06,
+      "loss": 0.8682,
+      "step": 18470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2919950431205264e-06,
+      "loss": 0.7988,
+      "step": 18471
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2908936308913433e-06,
+      "loss": 0.8037,
+      "step": 18472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2897926726771527e-06,
+      "loss": 0.8779,
+      "step": 18473
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2886921685046682e-06,
+      "loss": 0.8486,
+      "step": 18474
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.287592118400598e-06,
+      "loss": 0.7295,
+      "step": 18475
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2864925223916336e-06,
+      "loss": 0.8223,
+      "step": 18476
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2853933805044627e-06,
+      "loss": 0.6149,
+      "step": 18477
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2842946927657573e-06,
+      "loss": 0.8096,
+      "step": 18478
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2831964592021783e-06,
+      "loss": 0.9619,
+      "step": 18479
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2820986798403756e-06,
+      "loss": 0.8945,
+      "step": 18480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.281001354706992e-06,
+      "loss": 0.6377,
+      "step": 18481
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2799044838286556e-06,
+      "loss": 0.7168,
+      "step": 18482
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2788080672319802e-06,
+      "loss": 0.8916,
+      "step": 18483
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2777121049435781e-06,
+      "loss": 0.6934,
+      "step": 18484
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2766165969900435e-06,
+      "loss": 0.7158,
+      "step": 18485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2755215433979617e-06,
+      "loss": 0.9639,
+      "step": 18486
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2744269441939027e-06,
+      "loss": 0.9111,
+      "step": 18487
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2733327994044341e-06,
+      "loss": 0.8193,
+      "step": 18488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2722391090561036e-06,
+      "loss": 0.8037,
+      "step": 18489
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2711458731754566e-06,
+      "loss": 0.7329,
+      "step": 18490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.270053091789014e-06,
+      "loss": 0.9492,
+      "step": 18491
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2689607649233038e-06,
+      "loss": 0.7812,
+      "step": 18492
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.267868892604831e-06,
+      "loss": 0.8965,
+      "step": 18493
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2667774748600903e-06,
+      "loss": 0.9268,
+      "step": 18494
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2656865117155649e-06,
+      "loss": 0.8262,
+      "step": 18495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2645960031977334e-06,
+      "loss": 0.7251,
+      "step": 18496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2635059493330615e-06,
+      "loss": 0.8623,
+      "step": 18497
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2624163501479946e-06,
+      "loss": 0.7637,
+      "step": 18498
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2613272056689763e-06,
+      "loss": 0.7031,
+      "step": 18499
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2602385159224384e-06,
+      "loss": 0.9033,
+      "step": 18500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2591502809348044e-06,
+      "loss": 0.9678,
+      "step": 18501
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2580625007324753e-06,
+      "loss": 0.9111,
+      "step": 18502
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2569751753418547e-06,
+      "loss": 0.9873,
+      "step": 18503
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.255888304789319e-06,
+      "loss": 0.8994,
+      "step": 18504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2548018891012537e-06,
+      "loss": 0.8271,
+      "step": 18505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2537159283040202e-06,
+      "loss": 0.6396,
+      "step": 18506
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2526304224239683e-06,
+      "loss": 1.0195,
+      "step": 18507
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2515453714874416e-06,
+      "loss": 0.7646,
+      "step": 18508
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.250460775520772e-06,
+      "loss": 0.8584,
+      "step": 18509
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.249376634550281e-06,
+      "loss": 0.877,
+      "step": 18510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2482929486022743e-06,
+      "loss": 0.7158,
+      "step": 18511
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2472097177030485e-06,
+      "loss": 0.8345,
+      "step": 18512
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2461269418788958e-06,
+      "loss": 1.0,
+      "step": 18513
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2450446211560884e-06,
+      "loss": 0.9238,
+      "step": 18514
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2439627555608901e-06,
+      "loss": 0.7334,
+      "step": 18515
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2428813451195553e-06,
+      "loss": 0.8154,
+      "step": 18516
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2418003898583319e-06,
+      "loss": 0.6992,
+      "step": 18517
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2407198898034456e-06,
+      "loss": 0.9951,
+      "step": 18518
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2396398449811175e-06,
+      "loss": 1.0566,
+      "step": 18519
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2385602554175557e-06,
+      "loss": 0.7891,
+      "step": 18520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.237481121138966e-06,
+      "loss": 0.7163,
+      "step": 18521
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2364024421715294e-06,
+      "loss": 0.7549,
+      "step": 18522
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2353242185414205e-06,
+      "loss": 0.7173,
+      "step": 18523
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.234246450274812e-06,
+      "loss": 0.7021,
+      "step": 18524
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2331691373978538e-06,
+      "loss": 0.8174,
+      "step": 18525
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2320922799366874e-06,
+      "loss": 0.8994,
+      "step": 18526
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2310158779174454e-06,
+      "loss": 0.6992,
+      "step": 18527
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.229939931366253e-06,
+      "loss": 0.748,
+      "step": 18528
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2288644403092187e-06,
+      "loss": 0.8867,
+      "step": 18529
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2277894047724416e-06,
+      "loss": 0.8066,
+      "step": 18530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2267148247820027e-06,
+      "loss": 0.9229,
+      "step": 18531
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.225640700363988e-06,
+      "loss": 0.9365,
+      "step": 18532
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2245670315444614e-06,
+      "loss": 0.9043,
+      "step": 18533
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2234938183494772e-06,
+      "loss": 0.7612,
+      "step": 18534
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2224210608050746e-06,
+      "loss": 0.8271,
+      "step": 18535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2213487589372908e-06,
+      "loss": 0.8047,
+      "step": 18536
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2202769127721492e-06,
+      "loss": 0.832,
+      "step": 18537
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.219205522335658e-06,
+      "loss": 0.8369,
+      "step": 18538
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.218134587653812e-06,
+      "loss": 0.957,
+      "step": 18539
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.217064108752606e-06,
+      "loss": 0.8701,
+      "step": 18540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2159940856580166e-06,
+      "loss": 0.8457,
+      "step": 18541
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2149245183960101e-06,
+      "loss": 0.8652,
+      "step": 18542
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2138554069925369e-06,
+      "loss": 0.7886,
+      "step": 18543
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2127867514735469e-06,
+      "loss": 0.7344,
+      "step": 18544
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2117185518649731e-06,
+      "loss": 0.8496,
+      "step": 18545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.210650808192735e-06,
+      "loss": 0.7969,
+      "step": 18546
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.20958352048274e-06,
+      "loss": 0.8027,
+      "step": 18547
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2085166887608968e-06,
+      "loss": 0.874,
+      "step": 18548
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.207450313053089e-06,
+      "loss": 0.9287,
+      "step": 18549
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2063843933851959e-06,
+      "loss": 0.8105,
+      "step": 18550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2053189297830793e-06,
+      "loss": 0.8672,
+      "step": 18551
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2042539222726046e-06,
+      "loss": 0.6904,
+      "step": 18552
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2031893708796094e-06,
+      "loss": 0.9121,
+      "step": 18553
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2021252756299284e-06,
+      "loss": 0.8662,
+      "step": 18554
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2010616365493811e-06,
+      "loss": 0.9141,
+      "step": 18555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1999984536637866e-06,
+      "loss": 0.792,
+      "step": 18556
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1989357269989376e-06,
+      "loss": 0.8877,
+      "step": 18557
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1978734565806293e-06,
+      "loss": 0.811,
+      "step": 18558
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1968116424346344e-06,
+      "loss": 0.8398,
+      "step": 18559
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1957502845867252e-06,
+      "loss": 0.7041,
+      "step": 18560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.194689383062655e-06,
+      "loss": 0.8682,
+      "step": 18561
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1936289378881695e-06,
+      "loss": 0.8633,
+      "step": 18562
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1925689490889992e-06,
+      "loss": 0.7183,
+      "step": 18563
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1915094166908725e-06,
+      "loss": 0.6284,
+      "step": 18564
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1904503407194979e-06,
+      "loss": 0.8779,
+      "step": 18565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1893917212005767e-06,
+      "loss": 0.8496,
+      "step": 18566
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1883335581597954e-06,
+      "loss": 0.8252,
+      "step": 18567
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.18727585162284e-06,
+      "loss": 0.8848,
+      "step": 18568
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.186218601615372e-06,
+      "loss": 0.9092,
+      "step": 18569
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1851618081630467e-06,
+      "loss": 0.751,
+      "step": 18570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1841054712915145e-06,
+      "loss": 0.9072,
+      "step": 18571
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1830495910264083e-06,
+      "loss": 0.7109,
+      "step": 18572
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1819941673933477e-06,
+      "loss": 0.6807,
+      "step": 18573
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1809392004179454e-06,
+      "loss": 0.7617,
+      "step": 18574
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1798846901258054e-06,
+      "loss": 0.8643,
+      "step": 18575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1788306365425185e-06,
+      "loss": 0.9629,
+      "step": 18576
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1777770396936594e-06,
+      "loss": 0.9707,
+      "step": 18577
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1767238996047947e-06,
+      "loss": 0.7881,
+      "step": 18578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1756712163014861e-06,
+      "loss": 0.8906,
+      "step": 18579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1746189898092774e-06,
+      "loss": 0.873,
+      "step": 18580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.173567220153702e-06,
+      "loss": 0.8945,
+      "step": 18581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.172515907360281e-06,
+      "loss": 0.8037,
+      "step": 18582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.171465051454528e-06,
+      "loss": 0.9316,
+      "step": 18583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1704146524619509e-06,
+      "loss": 0.9688,
+      "step": 18584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.169364710408032e-06,
+      "loss": 0.8652,
+      "step": 18585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.168315225318253e-06,
+      "loss": 1.0059,
+      "step": 18586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.16726619721808e-06,
+      "loss": 0.8843,
+      "step": 18587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1662176261329727e-06,
+      "loss": 0.8027,
+      "step": 18588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1651695120883778e-06,
+      "loss": 0.7402,
+      "step": 18589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1641218551097255e-06,
+      "loss": 0.9209,
+      "step": 18590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.163074655222438e-06,
+      "loss": 0.6538,
+      "step": 18591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1620279124519351e-06,
+      "loss": 0.8115,
+      "step": 18592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.160981626823614e-06,
+      "loss": 0.8086,
+      "step": 18593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1599357983628657e-06,
+      "loss": 0.8477,
+      "step": 18594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1588904270950652e-06,
+      "loss": 0.9736,
+      "step": 18595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.157845513045588e-06,
+      "loss": 0.8838,
+      "step": 18596
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1568010562397892e-06,
+      "loss": 1.0088,
+      "step": 18597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1557570567030108e-06,
+      "loss": 0.7773,
+      "step": 18598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1547135144605858e-06,
+      "loss": 0.7505,
+      "step": 18599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1536704295378454e-06,
+      "loss": 0.8203,
+      "step": 18600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1526278019601001e-06,
+      "loss": 0.7939,
+      "step": 18601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1515856317526497e-06,
+      "loss": 0.8779,
+      "step": 18602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1505439189407808e-06,
+      "loss": 0.9082,
+      "step": 18603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1495026635497796e-06,
+      "loss": 0.9766,
+      "step": 18604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.148461865604913e-06,
+      "loss": 0.7471,
+      "step": 18605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1474215251314314e-06,
+      "loss": 0.8428,
+      "step": 18606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1463816421545903e-06,
+      "loss": 0.958,
+      "step": 18607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1453422166996186e-06,
+      "loss": 0.7715,
+      "step": 18608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.144303248791745e-06,
+      "loss": 0.8633,
+      "step": 18609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1432647384561735e-06,
+      "loss": 0.8057,
+      "step": 18610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.142226685718113e-06,
+      "loss": 0.7261,
+      "step": 18611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1411890906027546e-06,
+      "loss": 0.7622,
+      "step": 18612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1401519531352756e-06,
+      "loss": 0.8169,
+      "step": 18613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1391152733408383e-06,
+      "loss": 0.8779,
+      "step": 18614
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1380790512446094e-06,
+      "loss": 0.9258,
+      "step": 18615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1370432868717328e-06,
+      "loss": 0.9297,
+      "step": 18616
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1360079802473401e-06,
+      "loss": 0.8652,
+      "step": 18617
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1349731313965529e-06,
+      "loss": 0.7861,
+      "step": 18618
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1339387403444912e-06,
+      "loss": 0.9795,
+      "step": 18619
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1329048071162552e-06,
+      "loss": 0.8018,
+      "step": 18620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1318713317369312e-06,
+      "loss": 0.8057,
+      "step": 18621
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1308383142315992e-06,
+      "loss": 0.79,
+      "step": 18622
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1298057546253304e-06,
+      "loss": 0.7607,
+      "step": 18623
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1287736529431824e-06,
+      "loss": 0.6465,
+      "step": 18624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1277420092101976e-06,
+      "loss": 0.9121,
+      "step": 18625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1267108234514113e-06,
+      "loss": 0.8955,
+      "step": 18626
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1256800956918523e-06,
+      "loss": 0.7812,
+      "step": 18627
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1246498259565297e-06,
+      "loss": 0.9521,
+      "step": 18628
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1236200142704457e-06,
+      "loss": 0.8652,
+      "step": 18629
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1225906606585868e-06,
+      "loss": 0.9717,
+      "step": 18630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1215617651459398e-06,
+      "loss": 0.793,
+      "step": 18631
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1205333277574692e-06,
+      "loss": 0.9219,
+      "step": 18632
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1195053485181306e-06,
+      "loss": 0.957,
+      "step": 18633
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1184778274528707e-06,
+      "loss": 0.8496,
+      "step": 18634
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1174507645866273e-06,
+      "loss": 0.916,
+      "step": 18635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1164241599443226e-06,
+      "loss": 0.6399,
+      "step": 18636
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.115398013550868e-06,
+      "loss": 0.9961,
+      "step": 18637
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1143723254311656e-06,
+      "loss": 0.9131,
+      "step": 18638
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1133470956101067e-06,
+      "loss": 0.873,
+      "step": 18639
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1123223241125713e-06,
+      "loss": 0.9512,
+      "step": 18640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1112980109634263e-06,
+      "loss": 0.916,
+      "step": 18641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1102741561875253e-06,
+      "loss": 0.8564,
+      "step": 18642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1092507598097213e-06,
+      "loss": 0.7266,
+      "step": 18643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1082278218548458e-06,
+      "loss": 0.9023,
+      "step": 18644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1072053423477213e-06,
+      "loss": 0.5645,
+      "step": 18645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.106183321313159e-06,
+      "loss": 0.8447,
+      "step": 18646
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1051617587759677e-06,
+      "loss": 0.9199,
+      "step": 18647
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1041406547609302e-06,
+      "loss": 0.9238,
+      "step": 18648
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1031200092928285e-06,
+      "loss": 0.9082,
+      "step": 18649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1020998223964296e-06,
+      "loss": 0.8037,
+      "step": 18650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1010800940964916e-06,
+      "loss": 0.8584,
+      "step": 18651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1000608244177614e-06,
+      "loss": 0.6807,
+      "step": 18652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.09904201338497e-06,
+      "loss": 0.8916,
+      "step": 18653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0980236610228446e-06,
+      "loss": 0.8525,
+      "step": 18654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0970057673560964e-06,
+      "loss": 0.8643,
+      "step": 18655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0959883324094278e-06,
+      "loss": 0.7778,
+      "step": 18656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0949713562075237e-06,
+      "loss": 0.9336,
+      "step": 18657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0939548387750708e-06,
+      "loss": 0.7861,
+      "step": 18658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0929387801367319e-06,
+      "loss": 0.9375,
+      "step": 18659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0919231803171671e-06,
+      "loss": 0.8525,
+      "step": 18660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0909080393410166e-06,
+      "loss": 0.8237,
+      "step": 18661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0898933572329206e-06,
+      "loss": 0.7998,
+      "step": 18662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0888791340174999e-06,
+      "loss": 0.8633,
+      "step": 18663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0878653697193676e-06,
+      "loss": 0.9131,
+      "step": 18664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.086852064363122e-06,
+      "loss": 0.7568,
+      "step": 18665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0858392179733524e-06,
+      "loss": 0.9619,
+      "step": 18666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0848268305746457e-06,
+      "loss": 0.8525,
+      "step": 18667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0838149021915645e-06,
+      "loss": 0.7617,
+      "step": 18668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0828034328486647e-06,
+      "loss": 0.8555,
+      "step": 18669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.081792422570489e-06,
+      "loss": 0.8086,
+      "step": 18670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0807818713815776e-06,
+      "loss": 0.9209,
+      "step": 18671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.079771779306451e-06,
+      "loss": 0.9014,
+      "step": 18672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0787621463696207e-06,
+      "loss": 0.9795,
+      "step": 18673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0777529725955848e-06,
+      "loss": 0.8525,
+      "step": 18674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0767442580088395e-06,
+      "loss": 0.7422,
+      "step": 18675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0757360026338581e-06,
+      "loss": 0.6484,
+      "step": 18676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0747282064951103e-06,
+      "loss": 0.9922,
+      "step": 18677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0737208696170497e-06,
+      "loss": 0.9502,
+      "step": 18678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0727139920241259e-06,
+      "loss": 0.9443,
+      "step": 18679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.07170757374077e-06,
+      "loss": 0.7891,
+      "step": 18680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0707016147914073e-06,
+      "loss": 0.6992,
+      "step": 18681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0696961152004426e-06,
+      "loss": 0.9463,
+      "step": 18682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.068691074992283e-06,
+      "loss": 0.8252,
+      "step": 18683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.067686494191318e-06,
+      "loss": 0.7153,
+      "step": 18684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0666823728219234e-06,
+      "loss": 0.959,
+      "step": 18685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0656787109084622e-06,
+      "loss": 0.9512,
+      "step": 18686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.064675508475299e-06,
+      "loss": 0.9512,
+      "step": 18687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0636727655467749e-06,
+      "loss": 0.8125,
+      "step": 18688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0626704821472233e-06,
+      "loss": 0.7246,
+      "step": 18689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0616686583009628e-06,
+      "loss": 0.7783,
+      "step": 18690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0606672940323116e-06,
+      "loss": 0.873,
+      "step": 18691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0596663893655656e-06,
+      "loss": 0.8418,
+      "step": 18692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0586659443250125e-06,
+      "loss": 0.9189,
+      "step": 18693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0576659589349347e-06,
+      "loss": 0.9141,
+      "step": 18694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0566664332195974e-06,
+      "loss": 0.9131,
+      "step": 18695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0556673672032547e-06,
+      "loss": 0.7666,
+      "step": 18696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0546687609101513e-06,
+      "loss": 0.9385,
+      "step": 18697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0536706143645214e-06,
+      "loss": 0.9248,
+      "step": 18698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0526729275905878e-06,
+      "loss": 0.8857,
+      "step": 18699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0516757006125577e-06,
+      "loss": 0.7861,
+      "step": 18700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0506789334546342e-06,
+      "loss": 0.8643,
+      "step": 18701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0496826261410043e-06,
+      "loss": 0.8164,
+      "step": 18702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0486867786958487e-06,
+      "loss": 0.8838,
+      "step": 18703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0476913911433306e-06,
+      "loss": 0.793,
+      "step": 18704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0466964635076015e-06,
+      "loss": 0.6992,
+      "step": 18705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0457019958128134e-06,
+      "loss": 0.8311,
+      "step": 18706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0447079880830957e-06,
+      "loss": 0.918,
+      "step": 18707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0437144403425693e-06,
+      "loss": 0.8262,
+      "step": 18708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0427213526153412e-06,
+      "loss": 0.959,
+      "step": 18709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0417287249255192e-06,
+      "loss": 0.7549,
+      "step": 18710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0407365572971862e-06,
+      "loss": 0.9014,
+      "step": 18711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0397448497544183e-06,
+      "loss": 0.8398,
+      "step": 18712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0387536023212808e-06,
+      "loss": 0.7441,
+      "step": 18713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0377628150218343e-06,
+      "loss": 0.917,
+      "step": 18714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0367724878801178e-06,
+      "loss": 0.8252,
+      "step": 18715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0357826209201628e-06,
+      "loss": 0.8994,
+      "step": 18716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.03479321416599e-06,
+      "loss": 0.8008,
+      "step": 18717
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.033804267641616e-06,
+      "loss": 0.8125,
+      "step": 18718
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0328157813710327e-06,
+      "loss": 0.9111,
+      "step": 18719
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0318277553782296e-06,
+      "loss": 0.874,
+      "step": 18720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0308401896871812e-06,
+      "loss": 0.7109,
+      "step": 18721
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0298530843218568e-06,
+      "loss": 0.6621,
+      "step": 18722
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0288664393062108e-06,
+      "loss": 0.8438,
+      "step": 18723
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0278802546641842e-06,
+      "loss": 0.7725,
+      "step": 18724
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0268945304197041e-06,
+      "loss": 0.7393,
+      "step": 18725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0259092665967008e-06,
+      "loss": 0.8848,
+      "step": 18726
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0249244632190769e-06,
+      "loss": 0.8066,
+      "step": 18727
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0239401203107335e-06,
+      "loss": 0.9805,
+      "step": 18728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0229562378955537e-06,
+      "loss": 0.7959,
+      "step": 18729
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0219728159974208e-06,
+      "loss": 0.8418,
+      "step": 18730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.020989854640193e-06,
+      "loss": 0.9297,
+      "step": 18731
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0200073538477274e-06,
+      "loss": 0.918,
+      "step": 18732
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.019025313643862e-06,
+      "loss": 0.874,
+      "step": 18733
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0180437340524342e-06,
+      "loss": 0.9189,
+      "step": 18734
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0170626150972618e-06,
+      "loss": 0.9697,
+      "step": 18735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.016081956802153e-06,
+      "loss": 0.9541,
+      "step": 18736
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.015101759190902e-06,
+      "loss": 0.959,
+      "step": 18737
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0141220222873005e-06,
+      "loss": 0.8604,
+      "step": 18738
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0131427461151234e-06,
+      "loss": 0.8467,
+      "step": 18739
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.012163930698129e-06,
+      "loss": 0.7471,
+      "step": 18740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0111855760600787e-06,
+      "loss": 0.8105,
+      "step": 18741
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0102076822247109e-06,
+      "loss": 0.8623,
+      "step": 18742
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0092302492157558e-06,
+      "loss": 0.8008,
+      "step": 18743
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0082532770569297e-06,
+      "loss": 0.9336,
+      "step": 18744
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0072767657719451e-06,
+      "loss": 0.8936,
+      "step": 18745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0063007153844983e-06,
+      "loss": 1.0,
+      "step": 18746
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0053251259182772e-06,
+      "loss": 0.833,
+      "step": 18747
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0043499973969496e-06,
+      "loss": 0.9141,
+      "step": 18748
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.003375329844185e-06,
+      "loss": 0.9453,
+      "step": 18749
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0024011232836316e-06,
+      "loss": 0.6626,
+      "step": 18750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.001427377738937e-06,
+      "loss": 0.9521,
+      "step": 18751
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.000454093233727e-06,
+      "loss": 0.7432,
+      "step": 18752
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.994812697916157e-07,
+      "loss": 0.8125,
+      "step": 18753
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.985089074362197e-07,
+      "loss": 0.7539,
+      "step": 18754
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.975370061911317e-07,
+      "loss": 0.8701,
+      "step": 18755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.965655660799368e-07,
+      "loss": 0.7534,
+      "step": 18756
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.955945871262052e-07,
+      "loss": 0.8965,
+      "step": 18757
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.946240693535047e-07,
+      "loss": 1.0078,
+      "step": 18758
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.936540127853856e-07,
+      "loss": 0.8115,
+      "step": 18759
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.926844174453887e-07,
+      "loss": 0.8438,
+      "step": 18760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.917152833570399e-07,
+      "loss": 0.7988,
+      "step": 18761
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.907466105438623e-07,
+      "loss": 0.9355,
+      "step": 18762
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.897783990293598e-07,
+      "loss": 0.8447,
+      "step": 18763
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.888106488370287e-07,
+      "loss": 0.6187,
+      "step": 18764
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.878433599903502e-07,
+      "loss": 0.9121,
+      "step": 18765
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.868765325128037e-07,
+      "loss": 0.918,
+      "step": 18766
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.859101664278482e-07,
+      "loss": 0.752,
+      "step": 18767
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.849442617589333e-07,
+      "loss": 0.7559,
+      "step": 18768
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.839788185294986e-07,
+      "loss": 0.8145,
+      "step": 18769
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.83013836762976e-07,
+      "loss": 0.916,
+      "step": 18770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.820493164827804e-07,
+      "loss": 0.8496,
+      "step": 18771
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.810852577123198e-07,
+      "loss": 0.8652,
+      "step": 18772
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.80121660474982e-07,
+      "loss": 0.9707,
+      "step": 18773
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.791585247941616e-07,
+      "loss": 0.6797,
+      "step": 18774
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.781958506932242e-07,
+      "loss": 0.7168,
+      "step": 18775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.772336381955316e-07,
+      "loss": 0.8145,
+      "step": 18776
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.762718873244336e-07,
+      "loss": 0.8301,
+      "step": 18777
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.753105981032719e-07,
+      "loss": 0.8076,
+      "step": 18778
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.743497705553718e-07,
+      "loss": 0.835,
+      "step": 18779
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.733894047040482e-07,
+      "loss": 0.7539,
+      "step": 18780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.724295005726115e-07,
+      "loss": 0.707,
+      "step": 18781
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.714700581843518e-07,
+      "loss": 0.9424,
+      "step": 18782
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.705110775625547e-07,
+      "loss": 0.7803,
+      "step": 18783
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.695525587304866e-07,
+      "loss": 0.8701,
+      "step": 18784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.685945017114128e-07,
+      "loss": 0.9414,
+      "step": 18785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.676369065285817e-07,
+      "loss": 0.7998,
+      "step": 18786
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.666797732052325e-07,
+      "loss": 1.0566,
+      "step": 18787
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.65723101764584e-07,
+      "loss": 0.8477,
+      "step": 18788
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.647668922298625e-07,
+      "loss": 0.9863,
+      "step": 18789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.638111446242692e-07,
+      "loss": 0.8242,
+      "step": 18790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.628558589709946e-07,
+      "loss": 0.7979,
+      "step": 18791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.6190103529322e-07,
+      "loss": 0.8545,
+      "step": 18792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.609466736141205e-07,
+      "loss": 0.8271,
+      "step": 18793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.599927739568549e-07,
+      "loss": 0.6528,
+      "step": 18794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.59039336344567e-07,
+      "loss": 0.8398,
+      "step": 18795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.580863608003964e-07,
+      "loss": 0.8545,
+      "step": 18796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.57133847347471e-07,
+      "loss": 0.8447,
+      "step": 18797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.56181796008906e-07,
+      "loss": 0.6592,
+      "step": 18798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.552302068078023e-07,
+      "loss": 0.7422,
+      "step": 18799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.542790797672485e-07,
+      "loss": 0.5923,
+      "step": 18800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.53328414910335e-07,
+      "loss": 0.8276,
+      "step": 18801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.523782122601255e-07,
+      "loss": 0.8135,
+      "step": 18802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.514284718396794e-07,
+      "loss": 0.9014,
+      "step": 18803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.504791936720426e-07,
+      "loss": 0.9199,
+      "step": 18804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.495303777802545e-07,
+      "loss": 0.8203,
+      "step": 18805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.485820241873389e-07,
+      "loss": 0.7715,
+      "step": 18806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.476341329163108e-07,
+      "loss": 0.5215,
+      "step": 18807
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.466867039901672e-07,
+      "loss": 0.8525,
+      "step": 18808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.457397374319078e-07,
+      "loss": 0.8906,
+      "step": 18809
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.447932332645072e-07,
+      "loss": 0.8105,
+      "step": 18810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.438471915109359e-07,
+      "loss": 0.7021,
+      "step": 18811
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.42901612194147e-07,
+      "loss": 0.8271,
+      "step": 18812
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.419564953370952e-07,
+      "loss": 0.8428,
+      "step": 18813
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.410118409627111e-07,
+      "loss": 0.833,
+      "step": 18814
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.400676490939209e-07,
+      "loss": 0.8936,
+      "step": 18815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.391239197536306e-07,
+      "loss": 0.7754,
+      "step": 18816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.381806529647508e-07,
+      "loss": 0.8398,
+      "step": 18817
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.372378487501676e-07,
+      "loss": 0.999,
+      "step": 18818
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.362955071327584e-07,
+      "loss": 0.7822,
+      "step": 18819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.353536281353914e-07,
+      "loss": 0.8037,
+      "step": 18820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.344122117809284e-07,
+      "loss": 0.7441,
+      "step": 18821
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.334712580922111e-07,
+      "loss": 0.9707,
+      "step": 18822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.325307670920725e-07,
+      "loss": 0.7158,
+      "step": 18823
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.315907388033341e-07,
+      "loss": 0.835,
+      "step": 18824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.306511732488132e-07,
+      "loss": 0.8174,
+      "step": 18825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.297120704513097e-07,
+      "loss": 0.9355,
+      "step": 18826
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.287734304336049e-07,
+      "loss": 0.9551,
+      "step": 18827
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.278352532184853e-07,
+      "loss": 0.6797,
+      "step": 18828
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.268975388287172e-07,
+      "loss": 0.8169,
+      "step": 18829
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.25960287287051e-07,
+      "loss": 0.8467,
+      "step": 18830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.250234986162332e-07,
+      "loss": 0.8701,
+      "step": 18831
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.240871728389989e-07,
+      "loss": 0.9512,
+      "step": 18832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.231513099780676e-07,
+      "loss": 0.8564,
+      "step": 18833
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.222159100561523e-07,
+      "loss": 0.9873,
+      "step": 18834
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.212809730959526e-07,
+      "loss": 0.8135,
+      "step": 18835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.203464991201505e-07,
+      "loss": 0.9795,
+      "step": 18836
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.194124881514322e-07,
+      "loss": 0.7305,
+      "step": 18837
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.184789402124572e-07,
+      "loss": 0.7471,
+      "step": 18838
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.175458553258831e-07,
+      "loss": 0.9043,
+      "step": 18839
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.166132335143474e-07,
+      "loss": 0.96,
+      "step": 18840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.156810748004918e-07,
+      "loss": 0.8496,
+      "step": 18841
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.147493792069295e-07,
+      "loss": 0.8643,
+      "step": 18842
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.138181467562713e-07,
+      "loss": 0.6768,
+      "step": 18843
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.128873774711145e-07,
+      "loss": 0.687,
+      "step": 18844
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.1195707137405e-07,
+      "loss": 0.8994,
+      "step": 18845
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.110272284876531e-07,
+      "loss": 0.8926,
+      "step": 18846
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.100978488344836e-07,
+      "loss": 0.8037,
+      "step": 18847
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.091689324370989e-07,
+      "loss": 0.9287,
+      "step": 18848
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.08240479318041e-07,
+      "loss": 0.9268,
+      "step": 18849
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.073124894998387e-07,
+      "loss": 0.748,
+      "step": 18850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.063849630050137e-07,
+      "loss": 0.8604,
+      "step": 18851
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.054578998560704e-07,
+      "loss": 0.8418,
+      "step": 18852
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.045313000755129e-07,
+      "loss": 0.7842,
+      "step": 18853
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.036051636858212e-07,
+      "loss": 0.9766,
+      "step": 18854
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.026794907094727e-07,
+      "loss": 0.8818,
+      "step": 18855
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.017542811689272e-07,
+      "loss": 0.9023,
+      "step": 18856
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.008295350866425e-07,
+      "loss": 0.7422,
+      "step": 18857
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.99905252485056e-07,
+      "loss": 0.8564,
+      "step": 18858
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.989814333865965e-07,
+      "loss": 0.8623,
+      "step": 18859
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.980580778136815e-07,
+      "loss": 0.917,
+      "step": 18860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.971351857887222e-07,
+      "loss": 0.7744,
+      "step": 18861
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.962127573341139e-07,
+      "loss": 0.6846,
+      "step": 18862
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.952907924722387e-07,
+      "loss": 0.8037,
+      "step": 18863
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.943692912254697e-07,
+      "loss": 0.8135,
+      "step": 18864
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.934482536161715e-07,
+      "loss": 0.9111,
+      "step": 18865
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.925276796666948e-07,
+      "loss": 0.8076,
+      "step": 18866
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.916075693993753e-07,
+      "loss": 0.9609,
+      "step": 18867
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.906879228365462e-07,
+      "loss": 0.8701,
+      "step": 18868
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.897687400005229e-07,
+      "loss": 0.9473,
+      "step": 18869
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.888500209136119e-07,
+      "loss": 0.8447,
+      "step": 18870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.879317655981046e-07,
+      "loss": 0.8965,
+      "step": 18871
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.870139740762895e-07,
+      "loss": 0.9404,
+      "step": 18872
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.860966463704357e-07,
+      "loss": 0.752,
+      "step": 18873
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.85179782502803e-07,
+      "loss": 0.7715,
+      "step": 18874
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.842633824956404e-07,
+      "loss": 0.8945,
+      "step": 18875
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.833474463711922e-07,
+      "loss": 0.791,
+      "step": 18876
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.824319741516785e-07,
+      "loss": 1.0332,
+      "step": 18877
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.815169658593215e-07,
+      "loss": 0.9717,
+      "step": 18878
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.806024215163167e-07,
+      "loss": 0.8936,
+      "step": 18879
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.796883411448687e-07,
+      "loss": 0.9561,
+      "step": 18880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.787747247671507e-07,
+      "loss": 0.7744,
+      "step": 18881
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.778615724053385e-07,
+      "loss": 0.7881,
+      "step": 18882
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.769488840815854e-07,
+      "loss": 0.9062,
+      "step": 18883
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.760366598180492e-07,
+      "loss": 0.8955,
+      "step": 18884
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.751248996368611e-07,
+      "loss": 0.9629,
+      "step": 18885
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.742136035601456e-07,
+      "loss": 0.8525,
+      "step": 18886
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.733027716100184e-07,
+      "loss": 0.8789,
+      "step": 18887
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.723924038085841e-07,
+      "loss": 0.9238,
+      "step": 18888
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.714825001779359e-07,
+      "loss": 0.9307,
+      "step": 18889
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.705730607401496e-07,
+      "loss": 0.6641,
+      "step": 18890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.696640855172989e-07,
+      "loss": 0.873,
+      "step": 18891
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.687555745314391e-07,
+      "loss": 0.877,
+      "step": 18892
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.678475278046217e-07,
+      "loss": 0.7246,
+      "step": 18893
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.669399453588779e-07,
+      "loss": 0.8965,
+      "step": 18894
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.6603282721623e-07,
+      "loss": 0.9873,
+      "step": 18895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.651261733986981e-07,
+      "loss": 0.6987,
+      "step": 18896
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.642199839282805e-07,
+      "loss": 0.7451,
+      "step": 18897
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.633142588269661e-07,
+      "loss": 0.6045,
+      "step": 18898
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.624089981167349e-07,
+      "loss": 0.7871,
+      "step": 18899
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.615042018195563e-07,
+      "loss": 0.9629,
+      "step": 18900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.605998699573881e-07,
+      "loss": 0.8906,
+      "step": 18901
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.596960025521727e-07,
+      "loss": 0.9355,
+      "step": 18902
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.587925996258439e-07,
+      "loss": 0.8955,
+      "step": 18903
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.578896612003307e-07,
+      "loss": 0.8838,
+      "step": 18904
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.569871872975377e-07,
+      "loss": 0.7324,
+      "step": 18905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.560851779393698e-07,
+      "loss": 0.668,
+      "step": 18906
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.551836331477114e-07,
+      "loss": 0.8066,
+      "step": 18907
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.542825529444454e-07,
+      "loss": 0.7031,
+      "step": 18908
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.533819373514363e-07,
+      "loss": 0.7725,
+      "step": 18909
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.524817863905377e-07,
+      "loss": 0.9229,
+      "step": 18910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.515821000835989e-07,
+      "loss": 0.8789,
+      "step": 18911
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.506828784524468e-07,
+      "loss": 0.793,
+      "step": 18912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.497841215189063e-07,
+      "loss": 0.9424,
+      "step": 18913
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.488858293047819e-07,
+      "loss": 0.6748,
+      "step": 18914
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.479880018318831e-07,
+      "loss": 0.834,
+      "step": 18915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.470906391219857e-07,
+      "loss": 1.0391,
+      "step": 18916
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.461937411968768e-07,
+      "loss": 0.749,
+      "step": 18917
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.452973080783144e-07,
+      "loss": 0.7578,
+      "step": 18918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.444013397880524e-07,
+      "loss": 0.8682,
+      "step": 18919
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.435058363478376e-07,
+      "loss": 0.8809,
+      "step": 18920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.426107977793996e-07,
+      "loss": 0.8984,
+      "step": 18921
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.417162241044585e-07,
+      "loss": 0.9131,
+      "step": 18922
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.408221153447194e-07,
+      "loss": 0.8242,
+      "step": 18923
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.399284715218825e-07,
+      "loss": 0.8691,
+      "step": 18924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.390352926576373e-07,
+      "loss": 0.7349,
+      "step": 18925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.38142578773653e-07,
+      "loss": 0.8213,
+      "step": 18926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.372503298915924e-07,
+      "loss": 0.7939,
+      "step": 18927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.363585460331136e-07,
+      "loss": 0.6895,
+      "step": 18928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.354672272198571e-07,
+      "loss": 0.7705,
+      "step": 18929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.34576373473448e-07,
+      "loss": 0.9375,
+      "step": 18930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.336859848155043e-07,
+      "loss": 0.5664,
+      "step": 18931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.327960612676377e-07,
+      "loss": 0.8857,
+      "step": 18932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.319066028514422e-07,
+      "loss": 0.5913,
+      "step": 18933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.310176095885025e-07,
+      "loss": 0.8486,
+      "step": 18934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.301290815003904e-07,
+      "loss": 0.7266,
+      "step": 18935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.292410186086685e-07,
+      "loss": 0.8994,
+      "step": 18936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.283534209348909e-07,
+      "loss": 0.9062,
+      "step": 18937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.274662885005913e-07,
+      "loss": 0.8291,
+      "step": 18938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.265796213272992e-07,
+      "loss": 0.6133,
+      "step": 18939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.256934194365351e-07,
+      "loss": 0.9307,
+      "step": 18940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.248076828498019e-07,
+      "loss": 0.7959,
+      "step": 18941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.239224115885957e-07,
+      "loss": 0.833,
+      "step": 18942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.230376056743927e-07,
+      "loss": 0.873,
+      "step": 18943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.221532651286734e-07,
+      "loss": 0.7734,
+      "step": 18944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.212693899728941e-07,
+      "loss": 0.8857,
+      "step": 18945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.203859802285042e-07,
+      "loss": 0.6533,
+      "step": 18946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.195030359169398e-07,
+      "loss": 0.8359,
+      "step": 18947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.186205570596306e-07,
+      "loss": 0.8652,
+      "step": 18948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.177385436779883e-07,
+      "loss": 0.9209,
+      "step": 18949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.16856995793418e-07,
+      "loss": 0.8223,
+      "step": 18950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.159759134273159e-07,
+      "loss": 0.7148,
+      "step": 18951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.150952966010583e-07,
+      "loss": 0.7969,
+      "step": 18952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.142151453360192e-07,
+      "loss": 0.8174,
+      "step": 18953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.133354596535503e-07,
+      "loss": 0.8164,
+      "step": 18954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.124562395750079e-07,
+      "loss": 0.8184,
+      "step": 18955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.11577485121724e-07,
+      "loss": 0.7295,
+      "step": 18956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.106991963150235e-07,
+      "loss": 0.8818,
+      "step": 18957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.098213731762161e-07,
+      "loss": 0.8242,
+      "step": 18958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.089440157266115e-07,
+      "loss": 0.8184,
+      "step": 18959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.080671239874948e-07,
+      "loss": 0.8223,
+      "step": 18960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.071906979801492e-07,
+      "loss": 0.9385,
+      "step": 18961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.063147377258396e-07,
+      "loss": 0.7583,
+      "step": 18962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.054392432458246e-07,
+      "loss": 0.8848,
+      "step": 18963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.045642145613519e-07,
+      "loss": 0.9736,
+      "step": 18964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.036896516936533e-07,
+      "loss": 0.96,
+      "step": 18965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.028155546639516e-07,
+      "loss": 0.791,
+      "step": 18966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.019419234934611e-07,
+      "loss": 0.8516,
+      "step": 18967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.010687582033805e-07,
+      "loss": 0.9141,
+      "step": 18968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.001960588148993e-07,
+      "loss": 0.877,
+      "step": 18969
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.993238253491919e-07,
+      "loss": 0.7402,
+      "step": 18970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.984520578274302e-07,
+      "loss": 0.9229,
+      "step": 18971
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.975807562707682e-07,
+      "loss": 0.6343,
+      "step": 18972
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.967099207003492e-07,
+      "loss": 0.7549,
+      "step": 18973
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.95839551137303e-07,
+      "loss": 0.9385,
+      "step": 18974
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.949696476027546e-07,
+      "loss": 0.8701,
+      "step": 18975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.941002101178141e-07,
+      "loss": 0.7607,
+      "step": 18976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.932312387035801e-07,
+      "loss": 0.8926,
+      "step": 18977
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.923627333811357e-07,
+      "loss": 0.7432,
+      "step": 18978
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.914946941715617e-07,
+      "loss": 0.7222,
+      "step": 18979
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.906271210959215e-07,
+      "loss": 0.8135,
+      "step": 18980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.897600141752693e-07,
+      "loss": 0.8311,
+      "step": 18981
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.888933734306437e-07,
+      "loss": 0.9102,
+      "step": 18982
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.88027198883079e-07,
+      "loss": 0.9141,
+      "step": 18983
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.871614905535941e-07,
+      "loss": 0.8828,
+      "step": 18984
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.862962484631986e-07,
+      "loss": 0.9053,
+      "step": 18985
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.854314726328826e-07,
+      "loss": 0.8574,
+      "step": 18986
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.845671630836427e-07,
+      "loss": 0.9834,
+      "step": 18987
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.837033198364441e-07,
+      "loss": 0.9629,
+      "step": 18988
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.828399429122546e-07,
+      "loss": 0.7451,
+      "step": 18989
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.819770323320198e-07,
+      "loss": 0.6719,
+      "step": 18990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.811145881166893e-07,
+      "loss": 0.8477,
+      "step": 18991
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.802526102871843e-07,
+      "loss": 0.9316,
+      "step": 18992
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.793910988644283e-07,
+      "loss": 0.79,
+      "step": 18993
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.785300538693197e-07,
+      "loss": 0.9414,
+      "step": 18994
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.77669475322762e-07,
+      "loss": 0.7549,
+      "step": 18995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.768093632456342e-07,
+      "loss": 0.8535,
+      "step": 18996
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.759497176588082e-07,
+      "loss": 0.6743,
+      "step": 18997
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.750905385831475e-07,
+      "loss": 0.7617,
+      "step": 18998
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.742318260395021e-07,
+      "loss": 0.8086,
+      "step": 18999
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.733735800487086e-07,
+      "loss": 0.8359,
+      "step": 19000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.725158006315947e-07,
+      "loss": 0.8594,
+      "step": 19001
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.71658487808975e-07,
+      "loss": 0.7764,
+      "step": 19002
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.708016416016573e-07,
+      "loss": 0.8369,
+      "step": 19003
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.699452620304338e-07,
+      "loss": 0.9131,
+      "step": 19004
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.690893491160833e-07,
+      "loss": 0.748,
+      "step": 19005
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.682339028793783e-07,
+      "loss": 0.8975,
+      "step": 19006
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.673789233410778e-07,
+      "loss": 0.7754,
+      "step": 19007
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.665244105219316e-07,
+      "loss": 0.6973,
+      "step": 19008
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.656703644426743e-07,
+      "loss": 0.8945,
+      "step": 19009
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.648167851240273e-07,
+      "loss": 0.7246,
+      "step": 19010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.639636725867117e-07,
+      "loss": 0.9443,
+      "step": 19011
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.631110268514263e-07,
+      "loss": 0.9385,
+      "step": 19012
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.622588479388637e-07,
+      "loss": 0.9336,
+      "step": 19013
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.614071358696984e-07,
+      "loss": 0.8682,
+      "step": 19014
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.605558906646071e-07,
+      "loss": 0.7969,
+      "step": 19015
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.597051123442423e-07,
+      "loss": 0.7842,
+      "step": 19016
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.588548009292518e-07,
+      "loss": 0.7969,
+      "step": 19017
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.580049564402659e-07,
+      "loss": 0.709,
+      "step": 19018
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.571555788979123e-07,
+      "loss": 0.9531,
+      "step": 19019
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.563066683228037e-07,
+      "loss": 0.708,
+      "step": 19020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.554582247355369e-07,
+      "loss": 0.9434,
+      "step": 19021
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.546102481567019e-07,
+      "loss": 0.8652,
+      "step": 19022
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.537627386068802e-07,
+      "loss": 0.8291,
+      "step": 19023
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.529156961066331e-07,
+      "loss": 0.7393,
+      "step": 19024
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.520691206765218e-07,
+      "loss": 0.792,
+      "step": 19025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.512230123370811e-07,
+      "loss": 0.8086,
+      "step": 19026
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.503773711088547e-07,
+      "loss": 0.9082,
+      "step": 19027
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.49532197012357e-07,
+      "loss": 0.8838,
+      "step": 19028
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.486874900680984e-07,
+      "loss": 0.7529,
+      "step": 19029
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.47843250296576e-07,
+      "loss": 0.7354,
+      "step": 19030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.469994777182843e-07,
+      "loss": 0.8594,
+      "step": 19031
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.461561723536914e-07,
+      "loss": 0.8208,
+      "step": 19032
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.453133342232655e-07,
+      "loss": 0.9473,
+      "step": 19033
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.444709633474567e-07,
+      "loss": 0.918,
+      "step": 19034
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.436290597467132e-07,
+      "loss": 0.7871,
+      "step": 19035
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.427876234414589e-07,
+      "loss": 0.8301,
+      "step": 19036
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.419466544521148e-07,
+      "loss": 0.9248,
+      "step": 19037
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.411061527990915e-07,
+      "loss": 0.9609,
+      "step": 19038
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.402661185027838e-07,
+      "loss": 0.6445,
+      "step": 19039
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.394265515835752e-07,
+      "loss": 0.9014,
+      "step": 19040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.385874520618386e-07,
+      "loss": 0.7607,
+      "step": 19041
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.377488199579419e-07,
+      "loss": 0.8262,
+      "step": 19042
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.369106552922333e-07,
+      "loss": 0.8672,
+      "step": 19043
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.360729580850523e-07,
+      "loss": 0.8145,
+      "step": 19044
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.352357283567246e-07,
+      "loss": 0.8857,
+      "step": 19045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.343989661275741e-07,
+      "loss": 0.8564,
+      "step": 19046
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.335626714179001e-07,
+      "loss": 0.7637,
+      "step": 19047
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.327268442480018e-07,
+      "loss": 0.9521,
+      "step": 19048
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.318914846381564e-07,
+      "loss": 0.7256,
+      "step": 19049
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.31056592608641e-07,
+      "loss": 0.916,
+      "step": 19050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.30222168179715e-07,
+      "loss": 0.8066,
+      "step": 19051
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.293882113716267e-07,
+      "loss": 0.8408,
+      "step": 19052
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.285547222046107e-07,
+      "loss": 0.6758,
+      "step": 19053
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.277217006989001e-07,
+      "loss": 0.9932,
+      "step": 19054
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.268891468747053e-07,
+      "loss": 0.8809,
+      "step": 19055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.2605706075223e-07,
+      "loss": 0.7949,
+      "step": 19056
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.252254423516669e-07,
+      "loss": 0.7939,
+      "step": 19057
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.243942916931978e-07,
+      "loss": 0.9756,
+      "step": 19058
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.23563608796991e-07,
+      "loss": 0.7588,
+      "step": 19059
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.22733393683206e-07,
+      "loss": 0.8838,
+      "step": 19060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.219036463719886e-07,
+      "loss": 0.665,
+      "step": 19061
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.21074366883474e-07,
+      "loss": 0.9502,
+      "step": 19062
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.202455552377907e-07,
+      "loss": 0.8916,
+      "step": 19063
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.194172114550469e-07,
+      "loss": 0.8936,
+      "step": 19064
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.18589335555342e-07,
+      "loss": 0.9043,
+      "step": 19065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.17761927558771e-07,
+      "loss": 0.9199,
+      "step": 19066
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.169349874854137e-07,
+      "loss": 0.8477,
+      "step": 19067
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.161085153553316e-07,
+      "loss": 0.9346,
+      "step": 19068
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.152825111885842e-07,
+      "loss": 0.8223,
+      "step": 19069
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.144569750052177e-07,
+      "loss": 0.6326,
+      "step": 19070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.136319068252629e-07,
+      "loss": 0.8174,
+      "step": 19071
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.128073066687413e-07,
+      "loss": 0.957,
+      "step": 19072
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.11983174555666e-07,
+      "loss": 1.0518,
+      "step": 19073
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.111595105060343e-07,
+      "loss": 0.8506,
+      "step": 19074
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.103363145398368e-07,
+      "loss": 0.6885,
+      "step": 19075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.095135866770464e-07,
+      "loss": 0.8184,
+      "step": 19076
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.086913269376294e-07,
+      "loss": 0.7656,
+      "step": 19077
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.07869535341541e-07,
+      "loss": 0.8896,
+      "step": 19078
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.070482119087229e-07,
+      "loss": 0.8574,
+      "step": 19079
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.06227356659106e-07,
+      "loss": 0.915,
+      "step": 19080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.054069696126098e-07,
+      "loss": 0.6924,
+      "step": 19081
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.04587050789145e-07,
+      "loss": 0.9717,
+      "step": 19082
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.037676002086024e-07,
+      "loss": 1.043,
+      "step": 19083
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.029486178908751e-07,
+      "loss": 0.7505,
+      "step": 19084
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.021301038558337e-07,
+      "loss": 0.9053,
+      "step": 19085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.013120581233424e-07,
+      "loss": 0.7686,
+      "step": 19086
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.004944807132519e-07,
+      "loss": 0.6914,
+      "step": 19087
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.99677371645402e-07,
+      "loss": 0.8711,
+      "step": 19088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.988607309396211e-07,
+      "loss": 0.7168,
+      "step": 19089
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.980445586157292e-07,
+      "loss": 0.9512,
+      "step": 19090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.972288546935324e-07,
+      "loss": 0.8428,
+      "step": 19091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.964136191928217e-07,
+      "loss": 0.9648,
+      "step": 19092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.955988521333811e-07,
+      "loss": 0.8311,
+      "step": 19093
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.947845535349884e-07,
+      "loss": 0.8047,
+      "step": 19094
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.939707234173965e-07,
+      "loss": 0.7251,
+      "step": 19095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.931573618003606e-07,
+      "loss": 0.915,
+      "step": 19096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.92344468703614e-07,
+      "loss": 0.8398,
+      "step": 19097
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.915320441468853e-07,
+      "loss": 0.8994,
+      "step": 19098
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.907200881498922e-07,
+      "loss": 0.9043,
+      "step": 19099
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.899086007323341e-07,
+      "loss": 0.7783,
+      "step": 19100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.890975819139045e-07,
+      "loss": 0.7881,
+      "step": 19101
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.882870317142875e-07,
+      "loss": 0.7578,
+      "step": 19102
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.874769501531497e-07,
+      "loss": 0.9277,
+      "step": 19103
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.866673372501509e-07,
+      "loss": 0.7866,
+      "step": 19104
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.858581930249353e-07,
+      "loss": 0.876,
+      "step": 19105
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.850495174971428e-07,
+      "loss": 0.9785,
+      "step": 19106
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.842413106863932e-07,
+      "loss": 0.7578,
+      "step": 19107
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.834335726123042e-07,
+      "loss": 0.9062,
+      "step": 19108
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.826263032944714e-07,
+      "loss": 0.8516,
+      "step": 19109
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.818195027524877e-07,
+      "loss": 0.8281,
+      "step": 19110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.810131710059353e-07,
+      "loss": 0.874,
+      "step": 19111
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.802073080743765e-07,
+      "loss": 0.9688,
+      "step": 19112
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.794019139773666e-07,
+      "loss": 0.8193,
+      "step": 19113
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.785969887344546e-07,
+      "loss": 0.7471,
+      "step": 19114
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.777925323651713e-07,
+      "loss": 0.8135,
+      "step": 19115
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.76988544889039e-07,
+      "loss": 0.8848,
+      "step": 19116
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.761850263255665e-07,
+      "loss": 0.8174,
+      "step": 19117
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.75381976694256e-07,
+      "loss": 0.8496,
+      "step": 19118
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.745793960145919e-07,
+      "loss": 0.6685,
+      "step": 19119
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.737772843060542e-07,
+      "loss": 0.8213,
+      "step": 19120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.729756415881006e-07,
+      "loss": 0.9756,
+      "step": 19121
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.721744678801934e-07,
+      "loss": 0.96,
+      "step": 19122
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.713737632017703e-07,
+      "loss": 0.9658,
+      "step": 19123
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.705735275722602e-07,
+      "loss": 0.709,
+      "step": 19124
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.697737610110855e-07,
+      "loss": 0.8828,
+      "step": 19125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.68974463537655e-07,
+      "loss": 0.8809,
+      "step": 19126
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.681756351713642e-07,
+      "loss": 0.6074,
+      "step": 19127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.673772759315955e-07,
+      "loss": 1.085,
+      "step": 19128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.665793858377268e-07,
+      "loss": 0.8945,
+      "step": 19129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.657819649091202e-07,
+      "loss": 0.6641,
+      "step": 19130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.649850131651248e-07,
+      "loss": 0.9512,
+      "step": 19131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.641885306250762e-07,
+      "loss": 0.832,
+      "step": 19132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.633925173083122e-07,
+      "loss": 0.8408,
+      "step": 19133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.625969732341442e-07,
+      "loss": 0.7812,
+      "step": 19134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.618018984218788e-07,
+      "loss": 0.749,
+      "step": 19135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.610072928908074e-07,
+      "loss": 0.8809,
+      "step": 19136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.602131566602166e-07,
+      "loss": 0.7036,
+      "step": 19137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.594194897493777e-07,
+      "loss": 0.8359,
+      "step": 19138
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.586262921775466e-07,
+      "loss": 0.8359,
+      "step": 19139
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.578335639639744e-07,
+      "loss": 0.8838,
+      "step": 19140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.570413051279012e-07,
+      "loss": 0.8496,
+      "step": 19141
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.562495156885474e-07,
+      "loss": 0.8984,
+      "step": 19142
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.554581956651307e-07,
+      "loss": 0.9336,
+      "step": 19143
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.546673450768515e-07,
+      "loss": 0.8496,
+      "step": 19144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.538769639429076e-07,
+      "loss": 0.7773,
+      "step": 19145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.530870522824729e-07,
+      "loss": 0.9531,
+      "step": 19146
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.522976101147182e-07,
+      "loss": 0.7715,
+      "step": 19147
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.515086374587975e-07,
+      "loss": 0.835,
+      "step": 19148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.507201343338642e-07,
+      "loss": 0.8574,
+      "step": 19149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.499321007590476e-07,
+      "loss": 0.9248,
+      "step": 19150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.491445367534743e-07,
+      "loss": 0.8604,
+      "step": 19151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.483574423362515e-07,
+      "loss": 0.7871,
+      "step": 19152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.475708175264839e-07,
+      "loss": 0.7871,
+      "step": 19153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.467846623432606e-07,
+      "loss": 0.916,
+      "step": 19154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.459989768056551e-07,
+      "loss": 0.8613,
+      "step": 19155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.452137609327369e-07,
+      "loss": 0.9141,
+      "step": 19156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.444290147435617e-07,
+      "loss": 0.7988,
+      "step": 19157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.43644738257172e-07,
+      "loss": 0.6313,
+      "step": 19158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.428609314925971e-07,
+      "loss": 0.8799,
+      "step": 19159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.420775944688595e-07,
+      "loss": 0.7129,
+      "step": 19160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.41294727204973e-07,
+      "loss": 0.8867,
+      "step": 19161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.40512329719929e-07,
+      "loss": 0.8467,
+      "step": 19162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.397304020327188e-07,
+      "loss": 0.9219,
+      "step": 19163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.38948944162312e-07,
+      "loss": 0.5645,
+      "step": 19164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.381679561276755e-07,
+      "loss": 0.8203,
+      "step": 19165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.373874379477629e-07,
+      "loss": 1.0137,
+      "step": 19166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.366073896415148e-07,
+      "loss": 0.9346,
+      "step": 19167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.35827811227856e-07,
+      "loss": 0.9941,
+      "step": 19168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.350487027257113e-07,
+      "loss": 0.9805,
+      "step": 19169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.342700641539857e-07,
+      "loss": 0.8184,
+      "step": 19170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.334918955315728e-07,
+      "loss": 0.7632,
+      "step": 19171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.327141968773532e-07,
+      "loss": 1.0342,
+      "step": 19172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.31936968210205e-07,
+      "loss": 0.7295,
+      "step": 19173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.311602095489888e-07,
+      "loss": 0.874,
+      "step": 19174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.303839209125517e-07,
+      "loss": 0.7412,
+      "step": 19175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.296081023197298e-07,
+      "loss": 0.9238,
+      "step": 19176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.288327537893568e-07,
+      "loss": 0.8242,
+      "step": 19177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.280578753402422e-07,
+      "loss": 0.8135,
+      "step": 19178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.272834669911932e-07,
+      "loss": 0.751,
+      "step": 19179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.265095287609991e-07,
+      "loss": 0.8145,
+      "step": 19180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.257360606684471e-07,
+      "loss": 0.8135,
+      "step": 19181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.249630627323022e-07,
+      "loss": 0.8398,
+      "step": 19182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.241905349713229e-07,
+      "loss": 0.9189,
+      "step": 19183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.234184774042563e-07,
+      "loss": 0.8418,
+      "step": 19184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.226468900498384e-07,
+      "loss": 0.918,
+      "step": 19185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.218757729267966e-07,
+      "loss": 0.752,
+      "step": 19186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.211051260538403e-07,
+      "loss": 0.7617,
+      "step": 19187
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.20334949449668e-07,
+      "loss": 0.8174,
+      "step": 19188
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.195652431329757e-07,
+      "loss": 0.8477,
+      "step": 19189
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.187960071224397e-07,
+      "loss": 0.8623,
+      "step": 19190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.18027241436725e-07,
+      "loss": 0.9824,
+      "step": 19191
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.172589460944877e-07,
+      "loss": 0.8721,
+      "step": 19192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.164911211143754e-07,
+      "loss": 0.7324,
+      "step": 19193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.157237665150196e-07,
+      "loss": 0.7676,
+      "step": 19194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.149568823150387e-07,
+      "loss": 0.8857,
+      "step": 19195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.141904685330446e-07,
+      "loss": 0.7637,
+      "step": 19196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.134245251876381e-07,
+      "loss": 0.9043,
+      "step": 19197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.12659052297404e-07,
+      "loss": 0.7246,
+      "step": 19198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.11894049880919e-07,
+      "loss": 0.8164,
+      "step": 19199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.111295179567434e-07,
+      "loss": 0.7578,
+      "step": 19200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.103654565434381e-07,
+      "loss": 0.8027,
+      "step": 19201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.096018656595393e-07,
+      "loss": 0.8623,
+      "step": 19202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.088387453235767e-07,
+      "loss": 0.8486,
+      "step": 19203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.08076095554071e-07,
+      "loss": 0.7744,
+      "step": 19204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.073139163695296e-07,
+      "loss": 0.8408,
+      "step": 19205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.065522077884467e-07,
+      "loss": 0.6211,
+      "step": 19206
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.057909698293075e-07,
+      "loss": 0.7529,
+      "step": 19207
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.050302025105859e-07,
+      "loss": 0.8271,
+      "step": 19208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.042699058507428e-07,
+      "loss": 0.603,
+      "step": 19209
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.035100798682281e-07,
+      "loss": 0.7129,
+      "step": 19210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.027507245814801e-07,
+      "loss": 0.9414,
+      "step": 19211
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.019918400089264e-07,
+      "loss": 0.8066,
+      "step": 19212
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.012334261689856e-07,
+      "loss": 0.875,
+      "step": 19213
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.004754830800585e-07,
+      "loss": 0.874,
+      "step": 19214
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.997180107605394e-07,
+      "loss": 0.8672,
+      "step": 19215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.989610092288089e-07,
+      "loss": 0.9102,
+      "step": 19216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.982044785032415e-07,
+      "loss": 0.7407,
+      "step": 19217
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.974484186021912e-07,
+      "loss": 0.8887,
+      "step": 19218
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.966928295440033e-07,
+      "loss": 0.8916,
+      "step": 19219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.95937711347021e-07,
+      "loss": 0.6963,
+      "step": 19220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.951830640295653e-07,
+      "loss": 0.6611,
+      "step": 19221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.944288876099502e-07,
+      "loss": 0.5869,
+      "step": 19222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.936751821064724e-07,
+      "loss": 0.8828,
+      "step": 19223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.929219475374282e-07,
+      "loss": 0.6299,
+      "step": 19224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.921691839210942e-07,
+      "loss": 0.7217,
+      "step": 19225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.914168912757378e-07,
+      "loss": 0.8105,
+      "step": 19226
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.906650696196136e-07,
+      "loss": 0.8564,
+      "step": 19227
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.899137189709692e-07,
+      "loss": 0.875,
+      "step": 19228
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.891628393480364e-07,
+      "loss": 0.8691,
+      "step": 19229
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.884124307690342e-07,
+      "loss": 0.7881,
+      "step": 19230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.876624932521746e-07,
+      "loss": 0.9619,
+      "step": 19231
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.869130268156586e-07,
+      "loss": 0.7617,
+      "step": 19232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.861640314776696e-07,
+      "loss": 0.9395,
+      "step": 19233
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.854155072563883e-07,
+      "loss": 0.8652,
+      "step": 19234
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.846674541699715e-07,
+      "loss": 0.9219,
+      "step": 19235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.839198722365825e-07,
+      "loss": 0.8066,
+      "step": 19236
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.831727614743555e-07,
+      "loss": 0.8057,
+      "step": 19237
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.824261219014226e-07,
+      "loss": 1.0264,
+      "step": 19238
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.816799535359008e-07,
+      "loss": 0.7734,
+      "step": 19239
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.809342563959019e-07,
+      "loss": 0.874,
+      "step": 19240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.801890304995184e-07,
+      "loss": 0.8506,
+      "step": 19241
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.794442758648378e-07,
+      "loss": 0.8447,
+      "step": 19242
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.786999925099257e-07,
+      "loss": 0.9287,
+      "step": 19243
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.779561804528522e-07,
+      "loss": 0.9023,
+      "step": 19244
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.77212839711665e-07,
+      "loss": 0.7061,
+      "step": 19245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.764699703044008e-07,
+      "loss": 0.7441,
+      "step": 19246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.757275722490852e-07,
+      "loss": 0.8594,
+      "step": 19247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.749856455637393e-07,
+      "loss": 0.876,
+      "step": 19248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.742441902663665e-07,
+      "loss": 1.0117,
+      "step": 19249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.735032063749546e-07,
+      "loss": 0.7666,
+      "step": 19250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.727626939074871e-07,
+      "loss": 0.8486,
+      "step": 19251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.720226528819383e-07,
+      "loss": 0.8887,
+      "step": 19252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.712830833162652e-07,
+      "loss": 1.0186,
+      "step": 19253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.705439852284133e-07,
+      "loss": 0.6772,
+      "step": 19254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.698053586363173e-07,
+      "loss": 0.9414,
+      "step": 19255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.690672035579048e-07,
+      "loss": 0.8379,
+      "step": 19256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.683295200110884e-07,
+      "loss": 0.8672,
+      "step": 19257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.675923080137668e-07,
+      "loss": 0.9072,
+      "step": 19258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.668555675838284e-07,
+      "loss": 0.8691,
+      "step": 19259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.661192987391584e-07,
+      "loss": 0.6914,
+      "step": 19260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.653835014976183e-07,
+      "loss": 0.9229,
+      "step": 19261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.64648175877065e-07,
+      "loss": 0.9717,
+      "step": 19262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.639133218953441e-07,
+      "loss": 0.7773,
+      "step": 19263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.631789395702858e-07,
+      "loss": 0.8018,
+      "step": 19264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.624450289197159e-07,
+      "loss": 0.8379,
+      "step": 19265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.617115899614401e-07,
+      "loss": 0.7598,
+      "step": 19266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.609786227132552e-07,
+      "loss": 0.9619,
+      "step": 19267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.602461271929538e-07,
+      "loss": 0.8604,
+      "step": 19268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.595141034183105e-07,
+      "loss": 0.8936,
+      "step": 19269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.587825514070843e-07,
+      "loss": 0.8789,
+      "step": 19270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.580514711770324e-07,
+      "loss": 0.9268,
+      "step": 19271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.573208627458937e-07,
+      "loss": 0.8135,
+      "step": 19272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.565907261314008e-07,
+      "loss": 0.5928,
+      "step": 19273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.558610613512683e-07,
+      "loss": 0.9219,
+      "step": 19274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.551318684232021e-07,
+      "loss": 0.8896,
+      "step": 19275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.544031473649036e-07,
+      "loss": 0.793,
+      "step": 19276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.536748981940543e-07,
+      "loss": 0.9229,
+      "step": 19277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.529471209283244e-07,
+      "loss": 0.7998,
+      "step": 19278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.522198155853731e-07,
+      "loss": 0.6692,
+      "step": 19279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.514929821828574e-07,
+      "loss": 0.9404,
+      "step": 19280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.507666207384099e-07,
+      "loss": 1.001,
+      "step": 19281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.500407312696565e-07,
+      "loss": 1.0576,
+      "step": 19282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.493153137942142e-07,
+      "loss": 0.873,
+      "step": 19283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.48590368329689e-07,
+      "loss": 0.9141,
+      "step": 19284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.47865894893671e-07,
+      "loss": 0.5449,
+      "step": 19285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.471418935037398e-07,
+      "loss": 0.7744,
+      "step": 19286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.464183641774657e-07,
+      "loss": 0.7998,
+      "step": 19287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.456953069324078e-07,
+      "loss": 0.8604,
+      "step": 19288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.449727217861122e-07,
+      "loss": 0.6023,
+      "step": 19289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.442506087561139e-07,
+      "loss": 0.9346,
+      "step": 19290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.43528967859932e-07,
+      "loss": 0.9404,
+      "step": 19291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.428077991150882e-07,
+      "loss": 0.7959,
+      "step": 19292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.42087102539075e-07,
+      "loss": 0.9346,
+      "step": 19293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.413668781493831e-07,
+      "loss": 0.6948,
+      "step": 19294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.406471259634915e-07,
+      "loss": 1.0342,
+      "step": 19295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.399278459988688e-07,
+      "loss": 0.8076,
+      "step": 19296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.392090382729653e-07,
+      "loss": 0.8926,
+      "step": 19297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.384907028032249e-07,
+      "loss": 0.9141,
+      "step": 19298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.377728396070825e-07,
+      "loss": 0.7241,
+      "step": 19299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.370554487019575e-07,
+      "loss": 0.8623,
+      "step": 19300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.363385301052582e-07,
+      "loss": 0.834,
+      "step": 19301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.356220838343817e-07,
+      "loss": 0.9121,
+      "step": 19302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.349061099067143e-07,
+      "loss": 0.7207,
+      "step": 19303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.341906083396331e-07,
+      "loss": 0.8975,
+      "step": 19304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.334755791504975e-07,
+      "loss": 0.8906,
+      "step": 19305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.327610223566581e-07,
+      "loss": 0.9092,
+      "step": 19306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.320469379754589e-07,
+      "loss": 0.876,
+      "step": 19307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.313333260242304e-07,
+      "loss": 0.9258,
+      "step": 19308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.306201865202831e-07,
+      "loss": 0.7891,
+      "step": 19309
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.299075194809278e-07,
+      "loss": 0.6792,
+      "step": 19310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.291953249234572e-07,
+      "loss": 0.8428,
+      "step": 19311
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.284836028651552e-07,
+      "loss": 0.8369,
+      "step": 19312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.277723533232926e-07,
+      "loss": 0.9238,
+      "step": 19313
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.270615763151266e-07,
+      "loss": 0.8721,
+      "step": 19314
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.263512718579122e-07,
+      "loss": 0.7842,
+      "step": 19315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.256414399688803e-07,
+      "loss": 0.6787,
+      "step": 19316
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.249320806652591e-07,
+      "loss": 0.79,
+      "step": 19317
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.242231939642617e-07,
+      "loss": 0.9316,
+      "step": 19318
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.23514779883092e-07,
+      "loss": 0.957,
+      "step": 19319
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.228068384389407e-07,
+      "loss": 0.8115,
+      "step": 19320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.220993696489873e-07,
+      "loss": 0.9023,
+      "step": 19321
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.213923735303961e-07,
+      "loss": 0.7441,
+      "step": 19322
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.206858501003309e-07,
+      "loss": 0.8672,
+      "step": 19323
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.199797993759336e-07,
+      "loss": 0.8398,
+      "step": 19324
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.192742213743373e-07,
+      "loss": 0.667,
+      "step": 19325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.185691161126638e-07,
+      "loss": 0.9863,
+      "step": 19326
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.178644836080259e-07,
+      "loss": 0.834,
+      "step": 19327
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.171603238775213e-07,
+      "loss": 0.7935,
+      "step": 19328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.164566369382407e-07,
+      "loss": 0.8164,
+      "step": 19329
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.157534228072569e-07,
+      "loss": 0.6377,
+      "step": 19330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.150506815016365e-07,
+      "loss": 0.8643,
+      "step": 19331
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.143484130384324e-07,
+      "loss": 0.875,
+      "step": 19332
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.136466174346866e-07,
+      "loss": 0.8818,
+      "step": 19333
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.129452947074298e-07,
+      "loss": 0.9727,
+      "step": 19334
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.122444448736841e-07,
+      "loss": 0.7803,
+      "step": 19335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.115440679504535e-07,
+      "loss": 0.7197,
+      "step": 19336
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.108441639547335e-07,
+      "loss": 0.9316,
+      "step": 19337
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.101447329035103e-07,
+      "loss": 0.8301,
+      "step": 19338
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.094457748137571e-07,
+      "loss": 0.707,
+      "step": 19339
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.087472897024382e-07,
+      "loss": 0.8232,
+      "step": 19340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.080492775864975e-07,
+      "loss": 0.8418,
+      "step": 19341
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.073517384828774e-07,
+      "loss": 0.9268,
+      "step": 19342
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.066546724085041e-07,
+      "loss": 0.8184,
+      "step": 19343
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.059580793802977e-07,
+      "loss": 0.8916,
+      "step": 19344
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.052619594151553e-07,
+      "loss": 0.9004,
+      "step": 19345
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.045663125299727e-07,
+      "loss": 0.8447,
+      "step": 19346
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.03871138741634e-07,
+      "loss": 0.7915,
+      "step": 19347
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.031764380670057e-07,
+      "loss": 0.8105,
+      "step": 19348
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.024822105229477e-07,
+      "loss": 0.7954,
+      "step": 19349
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.017884561263042e-07,
+      "loss": 0.8252,
+      "step": 19350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.010951748939152e-07,
+      "loss": 0.8652,
+      "step": 19351
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.004023668426028e-07,
+      "loss": 0.8447,
+      "step": 19352
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.997100319891779e-07,
+      "loss": 0.7344,
+      "step": 19353
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.990181703504405e-07,
+      "loss": 0.8896,
+      "step": 19354
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.983267819431836e-07,
+      "loss": 0.873,
+      "step": 19355
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.97635866784183e-07,
+      "loss": 0.73,
+      "step": 19356
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.96945424890205e-07,
+      "loss": 0.8027,
+      "step": 19357
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.962554562780031e-07,
+      "loss": 0.9512,
+      "step": 19358
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.95565960964326e-07,
+      "loss": 0.835,
+      "step": 19359
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.948769389659025e-07,
+      "loss": 0.9287,
+      "step": 19360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.941883902994526e-07,
+      "loss": 0.791,
+      "step": 19361
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.93500314981683e-07,
+      "loss": 0.9141,
+      "step": 19362
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.928127130292982e-07,
+      "loss": 0.7461,
+      "step": 19363
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.921255844589778e-07,
+      "loss": 0.7754,
+      "step": 19364
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.914389292874022e-07,
+      "loss": 0.7852,
+      "step": 19365
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.907527475312268e-07,
+      "loss": 0.8311,
+      "step": 19366
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.900670392071095e-07,
+      "loss": 0.7905,
+      "step": 19367
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8938180433169e-07,
+      "loss": 0.835,
+      "step": 19368
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.886970429215932e-07,
+      "loss": 0.6924,
+      "step": 19369
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.880127549934388e-07,
+      "loss": 1.0527,
+      "step": 19370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.873289405638338e-07,
+      "loss": 0.8623,
+      "step": 19371
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.866455996493691e-07,
+      "loss": 0.8682,
+      "step": 19372
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.859627322666294e-07,
+      "loss": 0.9189,
+      "step": 19373
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.852803384321858e-07,
+      "loss": 0.8389,
+      "step": 19374
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.845984181625962e-07,
+      "loss": 0.8799,
+      "step": 19375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.839169714744118e-07,
+      "loss": 0.6719,
+      "step": 19376
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.832359983841683e-07,
+      "loss": 0.9258,
+      "step": 19377
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.825554989083881e-07,
+      "loss": 0.9072,
+      "step": 19378
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.818754730635888e-07,
+      "loss": 0.874,
+      "step": 19379
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.811959208662731e-07,
+      "loss": 0.8818,
+      "step": 19380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.805168423329253e-07,
+      "loss": 0.7642,
+      "step": 19381
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.798382374800326e-07,
+      "loss": 0.6616,
+      "step": 19382
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.791601063240569e-07,
+      "loss": 1.0293,
+      "step": 19383
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.784824488814588e-07,
+      "loss": 0.9434,
+      "step": 19384
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.778052651686782e-07,
+      "loss": 0.9355,
+      "step": 19385
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.771285552021532e-07,
+      "loss": 0.7725,
+      "step": 19386
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.764523189983017e-07,
+      "loss": 0.8945,
+      "step": 19387
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.757765565735373e-07,
+      "loss": 0.7588,
+      "step": 19388
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.751012679442557e-07,
+      "loss": 0.7236,
+      "step": 19389
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.744264531268483e-07,
+      "loss": 0.7241,
+      "step": 19390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.737521121376865e-07,
+      "loss": 0.8076,
+      "step": 19391
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.7307824499313706e-07,
+      "loss": 0.7812,
+      "step": 19392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.724048517095492e-07,
+      "loss": 0.833,
+      "step": 19393
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.7173193230326985e-07,
+      "loss": 0.8633,
+      "step": 19394
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.71059486790626e-07,
+      "loss": 0.8896,
+      "step": 19395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.703875151879356e-07,
+      "loss": 0.958,
+      "step": 19396
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.697160175115034e-07,
+      "loss": 0.749,
+      "step": 19397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.690449937776298e-07,
+      "loss": 0.7637,
+      "step": 19398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.68374444002595e-07,
+      "loss": 0.668,
+      "step": 19399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6770436820267053e-07,
+      "loss": 0.9502,
+      "step": 19400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.670347663941188e-07,
+      "loss": 0.917,
+      "step": 19401
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.663656385931914e-07,
+      "loss": 0.7617,
+      "step": 19402
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.656969848161219e-07,
+      "loss": 0.8115,
+      "step": 19403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.650288050791396e-07,
+      "loss": 0.9141,
+      "step": 19404
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6436109939845375e-07,
+      "loss": 0.8428,
+      "step": 19405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.636938677902758e-07,
+      "loss": 1.0234,
+      "step": 19406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.630271102707928e-07,
+      "loss": 0.6562,
+      "step": 19407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6236082685618524e-07,
+      "loss": 0.7705,
+      "step": 19408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6169501756262225e-07,
+      "loss": 0.6367,
+      "step": 19409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.610296824062621e-07,
+      "loss": 0.8193,
+      "step": 19410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6036482140324744e-07,
+      "loss": 0.8301,
+      "step": 19411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.597004345697165e-07,
+      "loss": 0.7207,
+      "step": 19412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5903652192178736e-07,
+      "loss": 0.7939,
+      "step": 19413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.583730834755762e-07,
+      "loss": 0.8281,
+      "step": 19414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.577101192471811e-07,
+      "loss": 0.9727,
+      "step": 19415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.570476292526871e-07,
+      "loss": 0.7852,
+      "step": 19416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.563856135081768e-07,
+      "loss": 0.8809,
+      "step": 19417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5572407202970847e-07,
+      "loss": 0.8164,
+      "step": 19418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5506300483334044e-07,
+      "loss": 0.7773,
+      "step": 19419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.544024119351154e-07,
+      "loss": 0.9248,
+      "step": 19420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5374229335105825e-07,
+      "loss": 0.8262,
+      "step": 19421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5308264909719624e-07,
+      "loss": 0.8145,
+      "step": 19422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5242347918953214e-07,
+      "loss": 0.9541,
+      "step": 19423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5176478364406194e-07,
+      "loss": 0.8408,
+      "step": 19424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5110656247677075e-07,
+      "loss": 0.9268,
+      "step": 19425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.504488157036324e-07,
+      "loss": 0.9004,
+      "step": 19426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4979154334060973e-07,
+      "loss": 0.8828,
+      "step": 19427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4913474540364985e-07,
+      "loss": 0.7207,
+      "step": 19428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.484784219086913e-07,
+      "loss": 0.7148,
+      "step": 19429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.478225728716634e-07,
+      "loss": 0.7715,
+      "step": 19430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.471671983084824e-07,
+      "loss": 0.8584,
+      "step": 19431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4651229823505113e-07,
+      "loss": 0.8164,
+      "step": 19432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.458578726672569e-07,
+      "loss": 0.8096,
+      "step": 19433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4520392162098915e-07,
+      "loss": 0.8691,
+      "step": 19434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.445504451121152e-07,
+      "loss": 0.8359,
+      "step": 19435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4389744315648906e-07,
+      "loss": 0.9023,
+      "step": 19436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.432449157699581e-07,
+      "loss": 0.7549,
+      "step": 19437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.425928629683607e-07,
+      "loss": 0.7832,
+      "step": 19438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4194128476751974e-07,
+      "loss": 0.8154,
+      "step": 19439
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4129018118324486e-07,
+      "loss": 0.8955,
+      "step": 19440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4063955223133446e-07,
+      "loss": 0.8311,
+      "step": 19441
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3998939792758265e-07,
+      "loss": 0.8457,
+      "step": 19442
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3933971828776345e-07,
+      "loss": 0.6748,
+      "step": 19443
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3869051332764423e-07,
+      "loss": 0.9014,
+      "step": 19444
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.380417830629746e-07,
+      "loss": 0.9062,
+      "step": 19445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.373935275095065e-07,
+      "loss": 0.9219,
+      "step": 19446
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.367457466829628e-07,
+      "loss": 0.8828,
+      "step": 19447
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.360984405990687e-07,
+      "loss": 0.7725,
+      "step": 19448
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.354516092735273e-07,
+      "loss": 0.7324,
+      "step": 19449
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.348052527220392e-07,
+      "loss": 0.7969,
+      "step": 19450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3415937096028984e-07,
+      "loss": 0.915,
+      "step": 19451
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3351396400395093e-07,
+      "loss": 0.8467,
+      "step": 19452
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3286903186868344e-07,
+      "loss": 0.7119,
+      "step": 19453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3222457457014143e-07,
+      "loss": 0.7812,
+      "step": 19454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.315805921239613e-07,
+      "loss": 0.9043,
+      "step": 19455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.309370845457728e-07,
+      "loss": 0.8867,
+      "step": 19456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3029405185118775e-07,
+      "loss": 0.9551,
+      "step": 19457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.296514940558161e-07,
+      "loss": 0.7803,
+      "step": 19458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.290094111752496e-07,
+      "loss": 0.9453,
+      "step": 19459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.283678032250671e-07,
+      "loss": 0.9521,
+      "step": 19460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2772667022083824e-07,
+      "loss": 0.873,
+      "step": 19461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2708601217812397e-07,
+      "loss": 0.7422,
+      "step": 19462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.264458291124718e-07,
+      "loss": 0.9043,
+      "step": 19463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.25806121039416e-07,
+      "loss": 0.7207,
+      "step": 19464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.251668879744775e-07,
+      "loss": 0.6738,
+      "step": 19465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.245281299331727e-07,
+      "loss": 0.7861,
+      "step": 19466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.238898469310004e-07,
+      "loss": 0.9307,
+      "step": 19467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2325203898344825e-07,
+      "loss": 0.8076,
+      "step": 19468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.226147061059993e-07,
+      "loss": 0.9492,
+      "step": 19469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.219778483141146e-07,
+      "loss": 0.8496,
+      "step": 19470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2134146562325283e-07,
+      "loss": 0.7969,
+      "step": 19471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.207055580488506e-07,
+      "loss": 0.9082,
+      "step": 19472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2007012560634664e-07,
+      "loss": 0.9219,
+      "step": 19473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.194351683111597e-07,
+      "loss": 0.7876,
+      "step": 19474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.188006861786953e-07,
+      "loss": 0.9209,
+      "step": 19475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.181666792243499e-07,
+      "loss": 0.7358,
+      "step": 19476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1753314746351357e-07,
+      "loss": 0.7285,
+      "step": 19477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1690009091155837e-07,
+      "loss": 0.7695,
+      "step": 19478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1626750958384533e-07,
+      "loss": 0.8662,
+      "step": 19479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1563540349572437e-07,
+      "loss": 0.915,
+      "step": 19480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1500377266253887e-07,
+      "loss": 0.8379,
+      "step": 19481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.143726170996143e-07,
+      "loss": 0.7764,
+      "step": 19482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1374193682226727e-07,
+      "loss": 0.8076,
+      "step": 19483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.131117318457989e-07,
+      "loss": 0.9189,
+      "step": 19484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.124820021855103e-07,
+      "loss": 0.7417,
+      "step": 19485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.118527478566758e-07,
+      "loss": 0.79,
+      "step": 19486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1122396887457003e-07,
+      "loss": 0.7822,
+      "step": 19487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.105956652544474e-07,
+      "loss": 0.8281,
+      "step": 19488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0996783701156004e-07,
+      "loss": 0.6846,
+      "step": 19489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0934048416114256e-07,
+      "loss": 0.8291,
+      "step": 19490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.087136067184161e-07,
+      "loss": 0.8213,
+      "step": 19491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.080872046985929e-07,
+      "loss": 0.8818,
+      "step": 19492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.074612781168763e-07,
+      "loss": 0.9541,
+      "step": 19493
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0683582698845426e-07,
+      "loss": 0.8008,
+      "step": 19494
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.062108513285079e-07,
+      "loss": 0.8389,
+      "step": 19495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0558635115219845e-07,
+      "loss": 0.7969,
+      "step": 19496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.049623264746827e-07,
+      "loss": 0.8594,
+      "step": 19497
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0433877731110624e-07,
+      "loss": 0.749,
+      "step": 19498
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0371570367659927e-07,
+      "loss": 0.7983,
+      "step": 19499
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0309310558627855e-07,
+      "loss": 0.8447,
+      "step": 19500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0247098305525645e-07,
+      "loss": 0.8604,
+      "step": 19501
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0184933609863195e-07,
+      "loss": 0.79,
+      "step": 19502
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.012281647314864e-07,
+      "loss": 0.8975,
+      "step": 19503
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0060746896889437e-07,
+      "loss": 0.9414,
+      "step": 19504
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9998724882592154e-07,
+      "loss": 0.8008,
+      "step": 19505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.99367504317616e-07,
+      "loss": 0.9121,
+      "step": 19506
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9874823545901663e-07,
+      "loss": 0.8242,
+      "step": 19507
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.981294422651516e-07,
+      "loss": 0.6523,
+      "step": 19508
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9751112475103994e-07,
+      "loss": 0.875,
+      "step": 19509
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9689328293168515e-07,
+      "loss": 0.7539,
+      "step": 19510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.962759168220798e-07,
+      "loss": 0.8271,
+      "step": 19511
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9565902643720066e-07,
+      "loss": 0.8916,
+      "step": 19512
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9504261179202696e-07,
+      "loss": 0.9414,
+      "step": 19513
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.944266729015134e-07,
+      "loss": 0.7432,
+      "step": 19514
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.938112097806057e-07,
+      "loss": 0.8818,
+      "step": 19515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.931962224442387e-07,
+      "loss": 0.9307,
+      "step": 19516
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.925817109073382e-07,
+      "loss": 0.8008,
+      "step": 19517
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9196767518481895e-07,
+      "loss": 0.7031,
+      "step": 19518
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.913541152915779e-07,
+      "loss": 0.9219,
+      "step": 19519
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9074103124250304e-07,
+      "loss": 0.8965,
+      "step": 19520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.90128423052476e-07,
+      "loss": 0.9561,
+      "step": 19521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8951629073636253e-07,
+      "loss": 0.7002,
+      "step": 19522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.889046343090175e-07,
+      "loss": 0.8643,
+      "step": 19523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.882934537852778e-07,
+      "loss": 0.8594,
+      "step": 19524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8768274917998503e-07,
+      "loss": 0.7637,
+      "step": 19525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8707252050795176e-07,
+      "loss": 0.752,
+      "step": 19526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.864627677839905e-07,
+      "loss": 0.7773,
+      "step": 19527
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.858534910228939e-07,
+      "loss": 0.9512,
+      "step": 19528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8524469023945245e-07,
+      "loss": 0.8906,
+      "step": 19529
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.846363654484364e-07,
+      "loss": 0.7471,
+      "step": 19530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8402851666460963e-07,
+      "loss": 0.8867,
+      "step": 19531
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.834211439027202e-07,
+      "loss": 0.8027,
+      "step": 19532
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8281424717751203e-07,
+      "loss": 0.835,
+      "step": 19533
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.822078265037088e-07,
+      "loss": 0.96,
+      "step": 19534
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8160188189602986e-07,
+      "loss": 0.8164,
+      "step": 19535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.809964133691746e-07,
+      "loss": 0.9658,
+      "step": 19536
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.803914209378401e-07,
+      "loss": 0.8467,
+      "step": 19537
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7978690461670577e-07,
+      "loss": 0.8516,
+      "step": 19538
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.791828644204443e-07,
+      "loss": 0.8486,
+      "step": 19539
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.785793003637106e-07,
+      "loss": 0.8896,
+      "step": 19540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7797621246115303e-07,
+      "loss": 0.8027,
+      "step": 19541
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7737360072740646e-07,
+      "loss": 0.7378,
+      "step": 19542
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7677146517709487e-07,
+      "loss": 0.9932,
+      "step": 19543
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7616980582482866e-07,
+      "loss": 0.8438,
+      "step": 19544
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.755686226852095e-07,
+      "loss": 0.8809,
+      "step": 19545
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.749679157728281e-07,
+      "loss": 0.9746,
+      "step": 19546
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.743676851022615e-07,
+      "loss": 0.9277,
+      "step": 19547
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.737679306880715e-07,
+      "loss": 0.9766,
+      "step": 19548
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.731686525448153e-07,
+      "loss": 0.8569,
+      "step": 19549
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.725698506870368e-07,
+      "loss": 0.8975,
+      "step": 19550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7197152512926436e-07,
+      "loss": 0.791,
+      "step": 19551
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.713736758860198e-07,
+      "loss": 0.7764,
+      "step": 19552
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7077630297181146e-07,
+      "loss": 0.9355,
+      "step": 19553
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.701794064011344e-07,
+      "loss": 0.9424,
+      "step": 19554
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.6958298618847254e-07,
+      "loss": 0.8525,
+      "step": 19555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.6898704234830105e-07,
+      "loss": 0.5894,
+      "step": 19556
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.683915748950817e-07,
+      "loss": 0.918,
+      "step": 19557
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.677965838432651e-07,
+      "loss": 0.7705,
+      "step": 19558
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.6720206920728865e-07,
+      "loss": 0.9727,
+      "step": 19559
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.666080310015785e-07,
+      "loss": 0.8828,
+      "step": 19560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.660144692405543e-07,
+      "loss": 0.8857,
+      "step": 19561
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.654213839386178e-07,
+      "loss": 0.959,
+      "step": 19562
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.6482877511015756e-07,
+      "loss": 0.8984,
+      "step": 19563
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.6423664276955985e-07,
+      "loss": 0.8428,
+      "step": 19564
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.636449869311931e-07,
+      "loss": 0.6333,
+      "step": 19565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.630538076094126e-07,
+      "loss": 0.7842,
+      "step": 19566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.624631048185667e-07,
+      "loss": 0.7832,
+      "step": 19567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.618728785729886e-07,
+      "loss": 0.8252,
+      "step": 19568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.6128312888700447e-07,
+      "loss": 0.9658,
+      "step": 19569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.606938557749207e-07,
+      "loss": 0.7666,
+      "step": 19570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.601050592510391e-07,
+      "loss": 0.9287,
+      "step": 19571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.595167393296528e-07,
+      "loss": 0.9277,
+      "step": 19572
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5892889602503254e-07,
+      "loss": 0.8369,
+      "step": 19573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.583415293514447e-07,
+      "loss": 0.7852,
+      "step": 19574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.577546393231446e-07,
+      "loss": 0.8887,
+      "step": 19575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.571682259543741e-07,
+      "loss": 1.0391,
+      "step": 19576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.56582289259364e-07,
+      "loss": 0.7354,
+      "step": 19577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.559968292523319e-07,
+      "loss": 0.8477,
+      "step": 19578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.554118459474842e-07,
+      "loss": 0.877,
+      "step": 19579
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5482733935901845e-07,
+      "loss": 0.8018,
+      "step": 19580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5424330950112106e-07,
+      "loss": 0.9053,
+      "step": 19581
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.536597563879607e-07,
+      "loss": 0.915,
+      "step": 19582
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5307668003370154e-07,
+      "loss": 0.9521,
+      "step": 19583
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.524940804524879e-07,
+      "loss": 0.9316,
+      "step": 19584
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.519119576584662e-07,
+      "loss": 0.7969,
+      "step": 19585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.513303116657585e-07,
+      "loss": 0.7949,
+      "step": 19586
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.507491424884779e-07,
+      "loss": 0.9082,
+      "step": 19587
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5016845014072876e-07,
+      "loss": 0.8206,
+      "step": 19588
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.495882346366042e-07,
+      "loss": 0.7402,
+      "step": 19589
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4900849599018185e-07,
+      "loss": 0.9365,
+      "step": 19590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4842923421553265e-07,
+      "loss": 0.8428,
+      "step": 19591
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.478504493267143e-07,
+      "loss": 0.8457,
+      "step": 19592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.472721413377689e-07,
+      "loss": 0.8076,
+      "step": 19593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.46694310262734e-07,
+      "loss": 0.8848,
+      "step": 19594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4611695611562746e-07,
+      "loss": 0.7861,
+      "step": 19595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4554007891046235e-07,
+      "loss": 0.7666,
+      "step": 19596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.449636786612409e-07,
+      "loss": 0.9805,
+      "step": 19597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4438775538194525e-07,
+      "loss": 0.7549,
+      "step": 19598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4381230908655307e-07,
+      "loss": 0.8438,
+      "step": 19599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4323733978903097e-07,
+      "loss": 0.9473,
+      "step": 19600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4266284750333004e-07,
+      "loss": 0.8867,
+      "step": 19601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4208883224339017e-07,
+      "loss": 0.8711,
+      "step": 19602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4151529402314256e-07,
+      "loss": 0.8242,
+      "step": 19603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.409422328565071e-07,
+      "loss": 0.6689,
+      "step": 19604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.403696487573882e-07,
+      "loss": 0.8994,
+      "step": 19605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.397975417396815e-07,
+      "loss": 0.8203,
+      "step": 19606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3922591181726693e-07,
+      "loss": 0.7607,
+      "step": 19607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3865475900402233e-07,
+      "loss": 0.8311,
+      "step": 19608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3808408331380327e-07,
+      "loss": 0.7305,
+      "step": 19609
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3751388476046085e-07,
+      "loss": 0.7976,
+      "step": 19610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3694416335782855e-07,
+      "loss": 0.9082,
+      "step": 19611
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.363749191197374e-07,
+      "loss": 0.7676,
+      "step": 19612
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3580615205999646e-07,
+      "loss": 0.7773,
+      "step": 19613
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.352378621924124e-07,
+      "loss": 0.8154,
+      "step": 19614
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.346700495307698e-07,
+      "loss": 0.8042,
+      "step": 19615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3410271408885307e-07,
+      "loss": 0.7539,
+      "step": 19616
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3353585588042913e-07,
+      "loss": 0.7275,
+      "step": 19617
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3296947491925356e-07,
+      "loss": 0.8838,
+      "step": 19618
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3240357121906874e-07,
+      "loss": 1.0254,
+      "step": 19619
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.318381447936081e-07,
+      "loss": 0.9375,
+      "step": 19620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.312731956565962e-07,
+      "loss": 0.8584,
+      "step": 19621
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.307087238217377e-07,
+      "loss": 1.0029,
+      "step": 19622
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3014472930273266e-07,
+      "loss": 0.8701,
+      "step": 19623
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.295812121132702e-07,
+      "loss": 1.0322,
+      "step": 19624
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.290181722670238e-07,
+      "loss": 0.6792,
+      "step": 19625
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.284556097776559e-07,
+      "loss": 0.7251,
+      "step": 19626
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2789352465881553e-07,
+      "loss": 0.8232,
+      "step": 19627
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.273319169241473e-07,
+      "loss": 0.8838,
+      "step": 19628
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.267707865872782e-07,
+      "loss": 0.9141,
+      "step": 19629
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.262101336618262e-07,
+      "loss": 0.8506,
+      "step": 19630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2564995816139367e-07,
+      "loss": 0.666,
+      "step": 19631
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.250902600995787e-07,
+      "loss": 0.7422,
+      "step": 19632
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2453103948996146e-07,
+      "loss": 0.7173,
+      "step": 19633
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2397229634611113e-07,
+      "loss": 0.8936,
+      "step": 19634
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2341403068158583e-07,
+      "loss": 0.7915,
+      "step": 19635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.228562425099391e-07,
+      "loss": 0.7217,
+      "step": 19636
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.222989318447023e-07,
+      "loss": 0.8838,
+      "step": 19637
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2174209869939797e-07,
+      "loss": 0.7314,
+      "step": 19638
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2118574308754424e-07,
+      "loss": 0.9238,
+      "step": 19639
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.206298650226369e-07,
+      "loss": 0.8623,
+      "step": 19640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.200744645181697e-07,
+      "loss": 0.874,
+      "step": 19641
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.195195415876162e-07,
+      "loss": 0.8115,
+      "step": 19642
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1896509624444794e-07,
+      "loss": 0.9336,
+      "step": 19643
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.184111285021141e-07,
+      "loss": 0.8682,
+      "step": 19644
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.178576383740639e-07,
+      "loss": 0.8027,
+      "step": 19645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.173046258737222e-07,
+      "loss": 0.9668,
+      "step": 19646
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1675209101451607e-07,
+      "loss": 0.7676,
+      "step": 19647
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.162000338098481e-07,
+      "loss": 0.8359,
+      "step": 19648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1564845427311863e-07,
+      "loss": 0.7725,
+      "step": 19649
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.150973524177103e-07,
+      "loss": 0.8203,
+      "step": 19650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.145467282569992e-07,
+      "loss": 0.9668,
+      "step": 19651
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.139965818043456e-07,
+      "loss": 0.833,
+      "step": 19652
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.13446913073101e-07,
+      "loss": 0.9736,
+      "step": 19653
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1289772207660163e-07,
+      "loss": 0.9082,
+      "step": 19654
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1234900882817886e-07,
+      "loss": 0.9502,
+      "step": 19655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1180077334114657e-07,
+      "loss": 0.7246,
+      "step": 19656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1125301562880737e-07,
+      "loss": 0.791,
+      "step": 19657
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1070573570445517e-07,
+      "loss": 0.7383,
+      "step": 19658
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.101589335813704e-07,
+      "loss": 0.8467,
+      "step": 19659
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0961260927282467e-07,
+      "loss": 0.8301,
+      "step": 19660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.090667627920718e-07,
+      "loss": 0.9404,
+      "step": 19661
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0852139415235903e-07,
+      "loss": 0.8926,
+      "step": 19662
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.079765033669224e-07,
+      "loss": 0.9688,
+      "step": 19663
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.074320904489847e-07,
+      "loss": 0.9072,
+      "step": 19664
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.068881554117553e-07,
+      "loss": 0.8291,
+      "step": 19665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.063446982684348e-07,
+      "loss": 0.957,
+      "step": 19666
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0580171903221043e-07,
+      "loss": 0.8721,
+      "step": 19667
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.052592177162605e-07,
+      "loss": 0.9912,
+      "step": 19668
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.047171943337501e-07,
+      "loss": 0.8408,
+      "step": 19669
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.041756488978331e-07,
+      "loss": 0.916,
+      "step": 19670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0363458142164794e-07,
+      "loss": 0.7656,
+      "step": 19671
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0309399191832625e-07,
+      "loss": 0.8398,
+      "step": 19672
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0255388040098864e-07,
+      "loss": 0.7275,
+      "step": 19673
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0201424688273807e-07,
+      "loss": 0.9014,
+      "step": 19674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0147509137667286e-07,
+      "loss": 0.7988,
+      "step": 19675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.009364138958759e-07,
+      "loss": 0.7666,
+      "step": 19676
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.00398214453419e-07,
+      "loss": 0.8887,
+      "step": 19677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.998604930623628e-07,
+      "loss": 0.9609,
+      "step": 19678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.993232497357568e-07,
+      "loss": 0.8633,
+      "step": 19679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.987864844866395e-07,
+      "loss": 0.9229,
+      "step": 19680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.982501973280338e-07,
+      "loss": 1.0254,
+      "step": 19681
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9771438827295385e-07,
+      "loss": 0.9893,
+      "step": 19682
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.971790573344047e-07,
+      "loss": 0.9902,
+      "step": 19683
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.966442045253759e-07,
+      "loss": 0.7441,
+      "step": 19684
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.961098298588483e-07,
+      "loss": 0.998,
+      "step": 19685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9557593334778476e-07,
+      "loss": 0.8789,
+      "step": 19686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9504251500514614e-07,
+      "loss": 0.7354,
+      "step": 19687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.945095748438753e-07,
+      "loss": 0.8506,
+      "step": 19688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9397711287690647e-07,
+      "loss": 0.6562,
+      "step": 19689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9344512911715803e-07,
+      "loss": 0.8135,
+      "step": 19690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9291362357754206e-07,
+      "loss": 0.8145,
+      "step": 19691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.92382596270957e-07,
+      "loss": 0.7861,
+      "step": 19692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9185204721028813e-07,
+      "loss": 0.834,
+      "step": 19693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9132197640840965e-07,
+      "loss": 0.8066,
+      "step": 19694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9079238387818675e-07,
+      "loss": 0.9521,
+      "step": 19695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.902632696324692e-07,
+      "loss": 0.9199,
+      "step": 19696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.897346336840978e-07,
+      "loss": 0.8184,
+      "step": 19697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8920647604590014e-07,
+      "loss": 0.8418,
+      "step": 19698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8867879673069476e-07,
+      "loss": 0.8955,
+      "step": 19699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8815159575128705e-07,
+      "loss": 0.8701,
+      "step": 19700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.876248731204689e-07,
+      "loss": 0.8877,
+      "step": 19701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8709862885102133e-07,
+      "loss": 0.7373,
+      "step": 19702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.865728629557185e-07,
+      "loss": 0.835,
+      "step": 19703
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8604757544731687e-07,
+      "loss": 0.7539,
+      "step": 19704
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.85522766338564e-07,
+      "loss": 0.8818,
+      "step": 19705
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.849984356421942e-07,
+      "loss": 0.8496,
+      "step": 19706
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.84474583370935e-07,
+      "loss": 0.9336,
+      "step": 19707
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.839512095374941e-07,
+      "loss": 0.9668,
+      "step": 19708
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8342831415457686e-07,
+      "loss": 0.8369,
+      "step": 19709
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.829058972348664e-07,
+      "loss": 0.8125,
+      "step": 19710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.823839587910482e-07,
+      "loss": 0.7461,
+      "step": 19711
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.81862498835781e-07,
+      "loss": 0.7803,
+      "step": 19712
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8134151738172355e-07,
+      "loss": 0.7661,
+      "step": 19713
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.808210144415147e-07,
+      "loss": 0.8154,
+      "step": 19714
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8030099002779087e-07,
+      "loss": 0.9287,
+      "step": 19715
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7978144415316656e-07,
+      "loss": 0.959,
+      "step": 19716
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.792623768302538e-07,
+      "loss": 0.7593,
+      "step": 19717
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7874378807164257e-07,
+      "loss": 0.8652,
+      "step": 19718
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7822567788992507e-07,
+      "loss": 0.8291,
+      "step": 19719
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.777080462976689e-07,
+      "loss": 0.8066,
+      "step": 19720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.771908933074374e-07,
+      "loss": 0.8115,
+      "step": 19721
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.766742189317806e-07,
+      "loss": 0.8213,
+      "step": 19722
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7615802318323724e-07,
+      "loss": 0.7388,
+      "step": 19723
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7564230607433074e-07,
+      "loss": 0.8335,
+      "step": 19724
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.751270676175799e-07,
+      "loss": 0.9814,
+      "step": 19725
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7461230782548586e-07,
+      "loss": 0.8184,
+      "step": 19726
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.740980267105409e-07,
+      "loss": 0.9219,
+      "step": 19727
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.735842242852238e-07,
+      "loss": 0.7358,
+      "step": 19728
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.730709005620047e-07,
+      "loss": 0.9648,
+      "step": 19729
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7255805555334247e-07,
+      "loss": 0.8047,
+      "step": 19730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.720456892716783e-07,
+      "loss": 0.7373,
+      "step": 19731
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7153380172944664e-07,
+      "loss": 0.8174,
+      "step": 19732
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7102239293907096e-07,
+      "loss": 0.832,
+      "step": 19733
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7051146291296124e-07,
+      "loss": 0.75,
+      "step": 19734
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7000101166351656e-07,
+      "loss": 0.9053,
+      "step": 19735
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.694910392031225e-07,
+      "loss": 0.9014,
+      "step": 19736
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6898154554415356e-07,
+      "loss": 0.957,
+      "step": 19737
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.684725306989777e-07,
+      "loss": 0.7812,
+      "step": 19738
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.679639946799473e-07,
+      "loss": 0.7803,
+      "step": 19739
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.67455937499399e-07,
+      "loss": 0.8926,
+      "step": 19740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.669483591696631e-07,
+      "loss": 0.9463,
+      "step": 19741
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.664412597030586e-07,
+      "loss": 0.6343,
+      "step": 19742
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6593463911188886e-07,
+      "loss": 0.8125,
+      "step": 19743
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.654284974084509e-07,
+      "loss": 0.8213,
+      "step": 19744
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6492283460502587e-07,
+      "loss": 0.9355,
+      "step": 19745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6441765071388623e-07,
+      "loss": 0.918,
+      "step": 19746
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6391294574728886e-07,
+      "loss": 0.707,
+      "step": 19747
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.634087197174817e-07,
+      "loss": 0.875,
+      "step": 19748
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.629049726367017e-07,
+      "loss": 0.7256,
+      "step": 19749
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.624017045171723e-07,
+      "loss": 0.7969,
+      "step": 19750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6189891537111046e-07,
+      "loss": 0.8467,
+      "step": 19751
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6139660521071307e-07,
+      "loss": 0.8887,
+      "step": 19752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6089477404817045e-07,
+      "loss": 0.8467,
+      "step": 19753
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.603934218956594e-07,
+      "loss": 0.8594,
+      "step": 19754
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.598925487653503e-07,
+      "loss": 0.7959,
+      "step": 19755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.593921546693956e-07,
+      "loss": 0.731,
+      "step": 19756
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5889223961993895e-07,
+      "loss": 0.9258,
+      "step": 19757
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5839280362910834e-07,
+      "loss": 0.9336,
+      "step": 19758
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5789384670902753e-07,
+      "loss": 0.9268,
+      "step": 19759
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5739536887180447e-07,
+      "loss": 0.7842,
+      "step": 19760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.568973701295363e-07,
+      "loss": 0.689,
+      "step": 19761
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5639985049430215e-07,
+      "loss": 0.8252,
+      "step": 19762
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.559028099781835e-07,
+      "loss": 0.9365,
+      "step": 19763
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.554062485932374e-07,
+      "loss": 0.7256,
+      "step": 19764
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.549101663515141e-07,
+      "loss": 0.8369,
+      "step": 19765
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5441456326505296e-07,
+      "loss": 1.042,
+      "step": 19766
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5391943934588216e-07,
+      "loss": 0.7969,
+      "step": 19767
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.534247946060142e-07,
+      "loss": 1.0059,
+      "step": 19768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.529306290574529e-07,
+      "loss": 0.8711,
+      "step": 19769
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.524369427121931e-07,
+      "loss": 0.9668,
+      "step": 19770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5194373558221187e-07,
+      "loss": 0.8906,
+      "step": 19771
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.514510076794796e-07,
+      "loss": 0.7959,
+      "step": 19772
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5095875901595346e-07,
+      "loss": 0.8047,
+      "step": 19773
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.504669896035772e-07,
+      "loss": 0.9102,
+      "step": 19774
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.499756994542879e-07,
+      "loss": 0.7798,
+      "step": 19775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.49484888580005e-07,
+      "loss": 0.8311,
+      "step": 19776
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.489945569926389e-07,
+      "loss": 0.7227,
+      "step": 19777
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4850470470408894e-07,
+      "loss": 0.9141,
+      "step": 19778
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.480153317262435e-07,
+      "loss": 0.7725,
+      "step": 19779
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4752643807097743e-07,
+      "loss": 0.8174,
+      "step": 19780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.470380237501546e-07,
+      "loss": 0.8418,
+      "step": 19781
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.465500887756278e-07,
+      "loss": 0.8516,
+      "step": 19782
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4606263315923864e-07,
+      "loss": 0.8018,
+      "step": 19783
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.455756569128132e-07,
+      "loss": 0.9043,
+      "step": 19784
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4508916004817087e-07,
+      "loss": 0.7188,
+      "step": 19785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.446031425771178e-07,
+      "loss": 0.7344,
+      "step": 19786
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4411760451144906e-07,
+      "loss": 0.9268,
+      "step": 19787
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4363254586294625e-07,
+      "loss": 0.8711,
+      "step": 19788
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4314796664337783e-07,
+      "loss": 0.8477,
+      "step": 19789
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.426638668645076e-07,
+      "loss": 1.0723,
+      "step": 19790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.421802465380796e-07,
+      "loss": 0.8184,
+      "step": 19791
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.416971056758333e-07,
+      "loss": 0.8535,
+      "step": 19792
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.412144442894881e-07,
+      "loss": 0.9902,
+      "step": 19793
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.407322623907615e-07,
+      "loss": 0.873,
+      "step": 19794
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4025055999135515e-07,
+      "loss": 0.7676,
+      "step": 19795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.397693371029552e-07,
+      "loss": 0.8945,
+      "step": 19796
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.392885937372369e-07,
+      "loss": 0.9258,
+      "step": 19797
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.388083299058752e-07,
+      "loss": 0.9316,
+      "step": 19798
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3832854562051866e-07,
+      "loss": 0.8379,
+      "step": 19799
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3784924089280904e-07,
+      "loss": 0.7949,
+      "step": 19800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3737041573438147e-07,
+      "loss": 0.6885,
+      "step": 19801
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3689207015685334e-07,
+      "loss": 0.8496,
+      "step": 19802
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3641420417183536e-07,
+      "loss": 0.9004,
+      "step": 19803
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3593681779092048e-07,
+      "loss": 0.917,
+      "step": 19804
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.35459911025695e-07,
+      "loss": 0.7852,
+      "step": 19805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3498348388773183e-07,
+      "loss": 0.873,
+      "step": 19806
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3450753638859514e-07,
+      "loss": 0.8027,
+      "step": 19807
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3403206853982897e-07,
+      "loss": 0.8418,
+      "step": 19808
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3355708035297743e-07,
+      "loss": 0.7544,
+      "step": 19809
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.330825718395624e-07,
+      "loss": 0.7393,
+      "step": 19810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.326085430111036e-07,
+      "loss": 0.7529,
+      "step": 19811
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3213499387909845e-07,
+      "loss": 0.8301,
+      "step": 19812
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3166192445504443e-07,
+      "loss": 0.957,
+      "step": 19813
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3118933475041684e-07,
+      "loss": 0.8569,
+      "step": 19814
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3071722477668867e-07,
+      "loss": 0.7861,
+      "step": 19815
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.302455945453108e-07,
+      "loss": 0.9502,
+      "step": 19816
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2977444406773407e-07,
+      "loss": 0.8379,
+      "step": 19817
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2930377335538935e-07,
+      "loss": 0.7236,
+      "step": 19818
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2883358241969854e-07,
+      "loss": 0.7549,
+      "step": 19819
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2836387127207037e-07,
+      "loss": 0.5503,
+      "step": 19820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2789463992390458e-07,
+      "loss": 0.8955,
+      "step": 19821
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2742588838658987e-07,
+      "loss": 0.8223,
+      "step": 19822
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.269576166714993e-07,
+      "loss": 0.9209,
+      "step": 19823
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2648982478999492e-07,
+      "loss": 0.835,
+      "step": 19824
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2602251275343434e-07,
+      "loss": 0.6045,
+      "step": 19825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.255556805731529e-07,
+      "loss": 0.8125,
+      "step": 19826
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2508932826047934e-07,
+      "loss": 0.7783,
+      "step": 19827
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2462345582673128e-07,
+      "loss": 0.9883,
+      "step": 19828
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2415806328321743e-07,
+      "loss": 0.7812,
+      "step": 19829
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2369315064122655e-07,
+      "loss": 0.8916,
+      "step": 19830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2322871791204294e-07,
+      "loss": 0.7812,
+      "step": 19831
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2276476510693756e-07,
+      "loss": 0.7939,
+      "step": 19832
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2230129223716811e-07,
+      "loss": 0.8691,
+      "step": 19833
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2183829931398336e-07,
+      "loss": 0.8691,
+      "step": 19834
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2137578634861655e-07,
+      "loss": 0.8408,
+      "step": 19835
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2091375335229203e-07,
+      "loss": 0.7939,
+      "step": 19836
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2045220033622306e-07,
+      "loss": 0.9424,
+      "step": 19837
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1999112731160954e-07,
+      "loss": 0.8779,
+      "step": 19838
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1953053428964256e-07,
+      "loss": 0.9092,
+      "step": 19839
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1907042128149537e-07,
+      "loss": 0.8711,
+      "step": 19840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.186107882983346e-07,
+      "loss": 0.7725,
+      "step": 19841
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.181516353513158e-07,
+      "loss": 0.7275,
+      "step": 19842
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1769296245158112e-07,
+      "loss": 0.7046,
+      "step": 19843
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1723476961026168e-07,
+      "loss": 0.7393,
+      "step": 19844
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1677705683847082e-07,
+      "loss": 0.7607,
+      "step": 19845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1631982414732411e-07,
+      "loss": 0.998,
+      "step": 19846
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1586307154791485e-07,
+      "loss": 0.8232,
+      "step": 19847
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1540679905132088e-07,
+      "loss": 0.9619,
+      "step": 19848
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1495100666862446e-07,
+      "loss": 0.8379,
+      "step": 19849
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1449569441087892e-07,
+      "loss": 0.6904,
+      "step": 19850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1404086228913767e-07,
+      "loss": 1.0068,
+      "step": 19851
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.135865103144319e-07,
+      "loss": 0.6416,
+      "step": 19852
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1313263849779498e-07,
+      "loss": 0.8281,
+      "step": 19853
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.126792468502381e-07,
+      "loss": 0.9854,
+      "step": 19854
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.122263353827636e-07,
+      "loss": 0.7441,
+      "step": 19855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1177390410636044e-07,
+      "loss": 0.8945,
+      "step": 19856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1132195303201097e-07,
+      "loss": 0.7236,
+      "step": 19857
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1087048217068197e-07,
+      "loss": 0.8496,
+      "step": 19858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1041949153332686e-07,
+      "loss": 0.8779,
+      "step": 19859
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0996898113089248e-07,
+      "loss": 0.7373,
+      "step": 19860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0951895097431007e-07,
+      "loss": 0.9385,
+      "step": 19861
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0906940107450203e-07,
+      "loss": 0.8613,
+      "step": 19862
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0862033144237736e-07,
+      "loss": 0.9141,
+      "step": 19863
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0817174208882963e-07,
+      "loss": 0.9424,
+      "step": 19864
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0772363302475007e-07,
+      "loss": 0.7754,
+      "step": 19865
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0727600426101002e-07,
+      "loss": 0.9707,
+      "step": 19866
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.068288558084741e-07,
+      "loss": 0.9004,
+      "step": 19867
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0638218767799145e-07,
+      "loss": 0.8066,
+      "step": 19868
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0593599988040448e-07,
+      "loss": 0.7861,
+      "step": 19869
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.054902924265356e-07,
+      "loss": 0.8936,
+      "step": 19870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0504506532720514e-07,
+      "loss": 0.71,
+      "step": 19871
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.046003185932155e-07,
+      "loss": 0.8027,
+      "step": 19872
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0415605223536028e-07,
+      "loss": 0.8496,
+      "step": 19873
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0371226626441976e-07,
+      "loss": 0.8955,
+      "step": 19874
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.032689606911653e-07,
+      "loss": 0.8262,
+      "step": 19875
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0282613552635054e-07,
+      "loss": 0.7617,
+      "step": 19876
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.023837907807269e-07,
+      "loss": 0.8711,
+      "step": 19877
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0194192646502354e-07,
+      "loss": 0.9443,
+      "step": 19878
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0150054258996743e-07,
+      "loss": 0.8027,
+      "step": 19879
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.010596391662656e-07,
+      "loss": 0.9453,
+      "step": 19880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0061921620462277e-07,
+      "loss": 0.8662,
+      "step": 19881
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0017927371572155e-07,
+      "loss": 0.8135,
+      "step": 19882
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9973981171024003e-07,
+      "loss": 0.6768,
+      "step": 19883
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9930083019884306e-07,
+      "loss": 0.708,
+      "step": 19884
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9886232919218428e-07,
+      "loss": 0.8198,
+      "step": 19885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9842430870090413e-07,
+      "loss": 0.8438,
+      "step": 19886
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.979867687356296e-07,
+      "loss": 0.8877,
+      "step": 19887
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9754970930698115e-07,
+      "loss": 0.8965,
+      "step": 19888
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9711313042556579e-07,
+      "loss": 0.8594,
+      "step": 19889
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9667703210197508e-07,
+      "loss": 0.8135,
+      "step": 19890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9624141434679388e-07,
+      "loss": 0.8306,
+      "step": 19891
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9580627717059153e-07,
+      "loss": 0.8955,
+      "step": 19892
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9537162058393068e-07,
+      "loss": 0.668,
+      "step": 19893
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9493744459735619e-07,
+      "loss": 0.9834,
+      "step": 19894
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9450374922140414e-07,
+      "loss": 0.7871,
+      "step": 19895
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9407053446660384e-07,
+      "loss": 0.6543,
+      "step": 19896
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9363780034346246e-07,
+      "loss": 0.7852,
+      "step": 19897
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.932055468624827e-07,
+      "loss": 0.9229,
+      "step": 19898
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9277377403415399e-07,
+      "loss": 0.8564,
+      "step": 19899
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9234248186895677e-07,
+      "loss": 0.877,
+      "step": 19900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.919116703773538e-07,
+      "loss": 0.8799,
+      "step": 19901
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.914813395698034e-07,
+      "loss": 0.8701,
+      "step": 19902
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.910514894567439e-07,
+      "loss": 0.9873,
+      "step": 19903
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9062212004860914e-07,
+      "loss": 0.9404,
+      "step": 19904
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9019323135581748e-07,
+      "loss": 0.96,
+      "step": 19905
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8976482338878055e-07,
+      "loss": 0.8496,
+      "step": 19906
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8933689615788786e-07,
+      "loss": 0.7725,
+      "step": 19907
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8890944967352886e-07,
+      "loss": 0.958,
+      "step": 19908
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8848248394607528e-07,
+      "loss": 1.0312,
+      "step": 19909
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8805599898588988e-07,
+      "loss": 0.791,
+      "step": 19910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8762999480331779e-07,
+      "loss": 0.8643,
+      "step": 19911
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.872044714087018e-07,
+      "loss": 0.9111,
+      "step": 19912
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8677942881236476e-07,
+      "loss": 0.8623,
+      "step": 19913
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8635486702462512e-07,
+      "loss": 0.7979,
+      "step": 19914
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8593078605577907e-07,
+      "loss": 0.9805,
+      "step": 19915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8550718591612283e-07,
+      "loss": 0.8516,
+      "step": 19916
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.850840666159348e-07,
+      "loss": 0.7832,
+      "step": 19917
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.846614281654846e-07,
+      "loss": 0.6909,
+      "step": 19918
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8423927057502622e-07,
+      "loss": 0.8682,
+      "step": 19919
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8381759385480259e-07,
+      "loss": 0.7178,
+      "step": 19920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8339639801504995e-07,
+      "loss": 0.8789,
+      "step": 19921
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.82975683065989e-07,
+      "loss": 0.8486,
+      "step": 19922
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.825554490178294e-07,
+      "loss": 0.7969,
+      "step": 19923
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8213569588076518e-07,
+      "loss": 0.8262,
+      "step": 19924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.817164236649882e-07,
+      "loss": 0.7329,
+      "step": 19925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8129763238066812e-07,
+      "loss": 0.8398,
+      "step": 19926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.808793220379701e-07,
+      "loss": 0.751,
+      "step": 19927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.804614926470438e-07,
+      "loss": 0.8457,
+      "step": 19928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.800441442180323e-07,
+      "loss": 1.0605,
+      "step": 19929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7962727676106074e-07,
+      "loss": 0.7422,
+      "step": 19930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.792108902862455e-07,
+      "loss": 0.8809,
+      "step": 19931
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.787949848036896e-07,
+      "loss": 0.8916,
+      "step": 19932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7837956032348946e-07,
+      "loss": 0.8643,
+      "step": 19933
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7796461685572142e-07,
+      "loss": 0.793,
+      "step": 19934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7755015441045965e-07,
+      "loss": 0.8115,
+      "step": 19935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7713617299775832e-07,
+      "loss": 0.8711,
+      "step": 19936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7672267262766718e-07,
+      "loss": 0.9277,
+      "step": 19937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.763096533102182e-07,
+      "loss": 0.9629,
+      "step": 19938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7589711505543227e-07,
+      "loss": 0.7793,
+      "step": 19939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.754850578733258e-07,
+      "loss": 0.832,
+      "step": 19940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7507348177389305e-07,
+      "loss": 0.9482,
+      "step": 19941
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7466238676712598e-07,
+      "loss": 0.9199,
+      "step": 19942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7425177286299667e-07,
+      "loss": 0.7236,
+      "step": 19943
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7384164007147263e-07,
+      "loss": 0.8281,
+      "step": 19944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7343198840250375e-07,
+      "loss": 0.8438,
+      "step": 19945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7302281786603536e-07,
+      "loss": 0.792,
+      "step": 19946
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7261412847199066e-07,
+      "loss": 0.8037,
+      "step": 19947
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.722059202302928e-07,
+      "loss": 0.7969,
+      "step": 19948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.71798193150845e-07,
+      "loss": 0.7861,
+      "step": 19949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7139094724354378e-07,
+      "loss": 0.7793,
+      "step": 19950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.709841825182701e-07,
+      "loss": 0.8198,
+      "step": 19951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7057789898489384e-07,
+      "loss": 0.8086,
+      "step": 19952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7017209665327826e-07,
+      "loss": 0.8867,
+      "step": 19953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.697667755332666e-07,
+      "loss": 0.9043,
+      "step": 19954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6936193563469537e-07,
+      "loss": 0.918,
+      "step": 19955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6895757696739235e-07,
+      "loss": 0.9121,
+      "step": 19956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.685536995411674e-07,
+      "loss": 0.793,
+      "step": 19957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6815030336582384e-07,
+      "loss": 0.9609,
+      "step": 19958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6774738845114492e-07,
+      "loss": 0.9365,
+      "step": 19959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6734495480691616e-07,
+      "loss": 0.8057,
+      "step": 19960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.669430024428964e-07,
+      "loss": 0.8994,
+      "step": 19961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6654153136884677e-07,
+      "loss": 0.8955,
+      "step": 19962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6614054159450165e-07,
+      "loss": 0.7344,
+      "step": 19963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6574003312959997e-07,
+      "loss": 0.8516,
+      "step": 19964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6534000598385613e-07,
+      "loss": 0.8682,
+      "step": 19965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6494046016697796e-07,
+      "loss": 0.9473,
+      "step": 19966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6454139568866213e-07,
+      "loss": 0.8164,
+      "step": 19967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.64142812558592e-07,
+      "loss": 0.792,
+      "step": 19968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6374471078644206e-07,
+      "loss": 0.7212,
+      "step": 19969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6334709038187125e-07,
+      "loss": 0.8682,
+      "step": 19970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6294995135452963e-07,
+      "loss": 0.8535,
+      "step": 19971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6255329371405393e-07,
+      "loss": 0.9009,
+      "step": 19972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.621571174700698e-07,
+      "loss": 0.8379,
+      "step": 19973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6176142263219173e-07,
+      "loss": 0.8164,
+      "step": 19974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6136620921001878e-07,
+      "loss": 0.9863,
+      "step": 19975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6097147721314766e-07,
+      "loss": 0.8193,
+      "step": 19976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6057722665115517e-07,
+      "loss": 0.8984,
+      "step": 19977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6018345753360475e-07,
+      "loss": 0.7988,
+      "step": 19978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5979016987005547e-07,
+      "loss": 0.8096,
+      "step": 19979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5939736367005298e-07,
+      "loss": 0.7451,
+      "step": 19980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5900503894312525e-07,
+      "loss": 0.8169,
+      "step": 19981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5861319569879573e-07,
+      "loss": 0.832,
+      "step": 19982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5822183394657242e-07,
+      "loss": 0.7676,
+      "step": 19983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5783095369595437e-07,
+      "loss": 0.8545,
+      "step": 19984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5744055495642284e-07,
+      "loss": 0.9707,
+      "step": 19985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5705063773745476e-07,
+      "loss": 0.6875,
+      "step": 19986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5666120204851365e-07,
+      "loss": 0.9307,
+      "step": 19987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5627224789904528e-07,
+      "loss": 0.8779,
+      "step": 19988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.558837752984932e-07,
+      "loss": 0.6821,
+      "step": 19989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5549578425628098e-07,
+      "loss": 0.9014,
+      "step": 19990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5510827478182555e-07,
+      "loss": 0.8115,
+      "step": 19991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5472124688453273e-07,
+      "loss": 0.7788,
+      "step": 19992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5433470057379053e-07,
+      "loss": 0.8086,
+      "step": 19993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5394863585898034e-07,
+      "loss": 0.6387,
+      "step": 19994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5356305274947247e-07,
+      "loss": 0.8955,
+      "step": 19995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5317795125462388e-07,
+      "loss": 0.9297,
+      "step": 19996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5279333138377815e-07,
+      "loss": 0.7466,
+      "step": 19997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.524091931462701e-07,
+      "loss": 0.9209,
+      "step": 19998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5202553655142115e-07,
+      "loss": 0.8359,
+      "step": 19999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5164236160854162e-07,
+      "loss": 0.8398,
+      "step": 20000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5125966832692852e-07,
+      "loss": 0.9258,
+      "step": 20001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5087745671587217e-07,
+      "loss": 0.7847,
+      "step": 20002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.504957267846452e-07,
+      "loss": 0.7871,
+      "step": 20003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5011447854251127e-07,
+      "loss": 0.7271,
+      "step": 20004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.49733711998723e-07,
+      "loss": 0.7529,
+      "step": 20005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4935342716252187e-07,
+      "loss": 0.9688,
+      "step": 20006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4897362404313164e-07,
+      "loss": 0.6685,
+      "step": 20007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.485943026497716e-07,
+      "loss": 0.8291,
+      "step": 20008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4821546299164992e-07,
+      "loss": 0.9434,
+      "step": 20009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4783710507795478e-07,
+      "loss": 0.8682,
+      "step": 20010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4745922891787003e-07,
+      "loss": 0.8193,
+      "step": 20011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4708183452056601e-07,
+      "loss": 0.8447,
+      "step": 20012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4670492189519992e-07,
+      "loss": 0.9082,
+      "step": 20013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4632849105091996e-07,
+      "loss": 0.9424,
+      "step": 20014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4595254199685882e-07,
+      "loss": 0.8398,
+      "step": 20015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4557707474214257e-07,
+      "loss": 1.0762,
+      "step": 20016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4520208929587942e-07,
+      "loss": 0.8682,
+      "step": 20017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4482758566717104e-07,
+      "loss": 1.0039,
+      "step": 20018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4445356386510345e-07,
+      "loss": 0.9814,
+      "step": 20019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.440800238987583e-07,
+      "loss": 0.8672,
+      "step": 20020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.43706965777195e-07,
+      "loss": 0.7466,
+      "step": 20021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4333438950946632e-07,
+      "loss": 0.8145,
+      "step": 20022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4296229510461835e-07,
+      "loss": 0.7842,
+      "step": 20023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4259068257167942e-07,
+      "loss": 0.8271,
+      "step": 20024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.422195519196645e-07,
+      "loss": 0.9531,
+      "step": 20025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4184890315757982e-07,
+      "loss": 0.832,
+      "step": 20026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4147873629442255e-07,
+      "loss": 0.8965,
+      "step": 20027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.411090513391744e-07,
+      "loss": 1.0273,
+      "step": 20028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4073984830080822e-07,
+      "loss": 0.9883,
+      "step": 20029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4037112718827906e-07,
+      "loss": 0.8125,
+      "step": 20030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4000288801053974e-07,
+      "loss": 0.708,
+      "step": 20031
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3963513077652536e-07,
+      "loss": 0.9121,
+      "step": 20032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3926785549515765e-07,
+      "loss": 0.9453,
+      "step": 20033
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.389010621753495e-07,
+      "loss": 0.9521,
+      "step": 20034
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3853475082600486e-07,
+      "loss": 0.8926,
+      "step": 20035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3816892145601e-07,
+      "loss": 0.7197,
+      "step": 20036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.378035740742445e-07,
+      "loss": 0.8721,
+      "step": 20037
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3743870868957454e-07,
+      "loss": 0.999,
+      "step": 20038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3707432531085309e-07,
+      "loss": 0.8184,
+      "step": 20039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.367104239469219e-07,
+      "loss": 0.9121,
+      "step": 20040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3634700460661399e-07,
+      "loss": 0.7549,
+      "step": 20041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3598406729874668e-07,
+      "loss": 0.9014,
+      "step": 20042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3562161203212853e-07,
+      "loss": 0.6953,
+      "step": 20043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.352596388155547e-07,
+      "loss": 0.9395,
+      "step": 20044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3489814765780928e-07,
+      "loss": 0.7871,
+      "step": 20045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3453713856766304e-07,
+      "loss": 0.8027,
+      "step": 20046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3417661155388007e-07,
+      "loss": 0.9355,
+      "step": 20047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3381656662520892e-07,
+      "loss": 0.7095,
+      "step": 20048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3345700379038263e-07,
+      "loss": 0.7959,
+      "step": 20049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3309792305812973e-07,
+      "loss": 0.7988,
+      "step": 20050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3273932443716553e-07,
+      "loss": 0.8887,
+      "step": 20051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3238120793618969e-07,
+      "loss": 1.0439,
+      "step": 20052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3202357356389305e-07,
+      "loss": 0.8975,
+      "step": 20053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3166642132895313e-07,
+      "loss": 0.9453,
+      "step": 20054
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3130975124004075e-07,
+      "loss": 0.9385,
+      "step": 20055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.30953563305809e-07,
+      "loss": 0.9902,
+      "step": 20056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3059785753490207e-07,
+      "loss": 0.6152,
+      "step": 20057
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.302426339359486e-07,
+      "loss": 0.7168,
+      "step": 20058
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2988789251757282e-07,
+      "loss": 0.958,
+      "step": 20059
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2953363328838342e-07,
+      "loss": 0.833,
+      "step": 20060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.291798562569735e-07,
+      "loss": 0.8047,
+      "step": 20061
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2882656143193173e-07,
+      "loss": 0.7529,
+      "step": 20062
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2847374882182906e-07,
+      "loss": 0.8057,
+      "step": 20063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2812141843522974e-07,
+      "loss": 0.79,
+      "step": 20064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2776957028068249e-07,
+      "loss": 0.9561,
+      "step": 20065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.274182043667249e-07,
+      "loss": 1.0264,
+      "step": 20066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2706732070188576e-07,
+      "loss": 0.8223,
+      "step": 20067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2671691929467822e-07,
+      "loss": 0.7656,
+      "step": 20068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2636700015360438e-07,
+      "loss": 0.8906,
+      "step": 20069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2601756328715742e-07,
+      "loss": 0.8374,
+      "step": 20070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2566860870381947e-07,
+      "loss": 0.8496,
+      "step": 20071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2532013641205265e-07,
+      "loss": 0.9521,
+      "step": 20072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2497214642031907e-07,
+      "loss": 0.6807,
+      "step": 20073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.246246387370609e-07,
+      "loss": 0.8203,
+      "step": 20074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.242776133707091e-07,
+      "loss": 0.8828,
+      "step": 20075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.239310703296903e-07,
+      "loss": 0.7363,
+      "step": 20076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2358500962240893e-07,
+      "loss": 0.8037,
+      "step": 20077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.232394312572649e-07,
+      "loss": 0.9512,
+      "step": 20078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2289433524264705e-07,
+      "loss": 0.8564,
+      "step": 20079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2254972158692423e-07,
+      "loss": 0.8086,
+      "step": 20080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2220559029846312e-07,
+      "loss": 0.8447,
+      "step": 20081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.218619413856148e-07,
+      "loss": 1.0107,
+      "step": 20082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2151877485671482e-07,
+      "loss": 0.7822,
+      "step": 20083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2117609072009652e-07,
+      "loss": 0.8574,
+      "step": 20084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2083388898407322e-07,
+      "loss": 0.8584,
+      "step": 20085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.204921696569472e-07,
+      "loss": 0.8516,
+      "step": 20086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2015093274701407e-07,
+      "loss": 0.8252,
+      "step": 20087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.198101782625516e-07,
+      "loss": 0.8623,
+      "step": 20088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.194699062118332e-07,
+      "loss": 0.8008,
+      "step": 20089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1913011660311003e-07,
+      "loss": 0.6895,
+      "step": 20090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1879080944463329e-07,
+      "loss": 0.873,
+      "step": 20091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1845198474463637e-07,
+      "loss": 0.8096,
+      "step": 20092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1811364251133938e-07,
+      "loss": 0.8711,
+      "step": 20093
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1777578275295132e-07,
+      "loss": 0.7988,
+      "step": 20094
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1743840547767671e-07,
+      "loss": 0.9336,
+      "step": 20095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.171015106936979e-07,
+      "loss": 0.9072,
+      "step": 20096
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1676509840919281e-07,
+      "loss": 0.9697,
+      "step": 20097
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1642916863232378e-07,
+      "loss": 0.8413,
+      "step": 20098
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1609372137124431e-07,
+      "loss": 0.7891,
+      "step": 20099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.157587566340923e-07,
+      "loss": 0.9189,
+      "step": 20100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1542427442899906e-07,
+      "loss": 0.9746,
+      "step": 20101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1509027476407808e-07,
+      "loss": 0.8359,
+      "step": 20102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1475675764743843e-07,
+      "loss": 0.791,
+      "step": 20103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1442372308717142e-07,
+      "loss": 0.7588,
+      "step": 20104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1409117109135948e-07,
+      "loss": 0.8105,
+      "step": 20105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.137591016680739e-07,
+      "loss": 0.8154,
+      "step": 20106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1342751482537052e-07,
+      "loss": 0.8545,
+      "step": 20107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1309641057129617e-07,
+      "loss": 0.8223,
+      "step": 20108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1276578891388668e-07,
+      "loss": 0.8662,
+      "step": 20109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1243564986116451e-07,
+      "loss": 0.7798,
+      "step": 20110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1210599342114326e-07,
+      "loss": 0.9678,
+      "step": 20111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1177681960182318e-07,
+      "loss": 0.8579,
+      "step": 20112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1144812841118679e-07,
+      "loss": 0.8818,
+      "step": 20113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.111199198572166e-07,
+      "loss": 0.6924,
+      "step": 20114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1079219394787511e-07,
+      "loss": 0.9053,
+      "step": 20115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1046495069111374e-07,
+      "loss": 0.9082,
+      "step": 20116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1013819009487503e-07,
+      "loss": 0.8799,
+      "step": 20117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.098119121670882e-07,
+      "loss": 0.752,
+      "step": 20118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0948611691567356e-07,
+      "loss": 0.7637,
+      "step": 20119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0916080434853371e-07,
+      "loss": 0.7451,
+      "step": 20120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0883597447356231e-07,
+      "loss": 0.8701,
+      "step": 20121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.085116272986464e-07,
+      "loss": 0.9141,
+      "step": 20122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.08187762831653e-07,
+      "loss": 0.8818,
+      "step": 20123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0786438108044473e-07,
+      "loss": 0.8477,
+      "step": 20124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0754148205286418e-07,
+      "loss": 0.875,
+      "step": 20125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.07219065756754e-07,
+      "loss": 0.8916,
+      "step": 20126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0689713219993236e-07,
+      "loss": 0.8564,
+      "step": 20127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0657568139021524e-07,
+      "loss": 0.9707,
+      "step": 20128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.062547133353986e-07,
+      "loss": 0.7139,
+      "step": 20129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0593422804327847e-07,
+      "loss": 0.8701,
+      "step": 20130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0561422552162637e-07,
+      "loss": 0.6875,
+      "step": 20131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0529470577820944e-07,
+      "loss": 0.7959,
+      "step": 20132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0497566882078369e-07,
+      "loss": 0.8223,
+      "step": 20133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0465711465708738e-07,
+      "loss": 0.8975,
+      "step": 20134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0433904329485434e-07,
+      "loss": 0.8799,
+      "step": 20135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0402145474180059e-07,
+      "loss": 0.7734,
+      "step": 20136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0370434900563331e-07,
+      "loss": 0.812,
+      "step": 20137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0338772609405079e-07,
+      "loss": 0.8354,
+      "step": 20138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0307158601473355e-07,
+      "loss": 0.8203,
+      "step": 20139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0275592877535323e-07,
+      "loss": 0.8232,
+      "step": 20140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0244075438357038e-07,
+      "loss": 0.7646,
+      "step": 20141
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0212606284703441e-07,
+      "loss": 0.7598,
+      "step": 20142
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0181185417338369e-07,
+      "loss": 0.9912,
+      "step": 20143
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0149812837023876e-07,
+      "loss": 0.8701,
+      "step": 20144
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0118488544521355e-07,
+      "loss": 0.8574,
+      "step": 20145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0087212540591307e-07,
+      "loss": 0.9043,
+      "step": 20146
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0055984825992459e-07,
+      "loss": 0.9473,
+      "step": 20147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0024805401482429e-07,
+      "loss": 0.8467,
+      "step": 20148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.993674267818165e-08,
+      "loss": 0.9023,
+      "step": 20149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.962591425755286e-08,
+      "loss": 0.7861,
+      "step": 20150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.931556876047633e-08,
+      "loss": 0.7881,
+      "step": 20151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.900570619448379e-08,
+      "loss": 0.8926,
+      "step": 20152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.869632656709816e-08,
+      "loss": 0.7705,
+      "step": 20153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.83874298858245e-08,
+      "loss": 0.9287,
+      "step": 20154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.807901615815907e-08,
+      "loss": 0.7598,
+      "step": 20155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.777108539158475e-08,
+      "loss": 0.9023,
+      "step": 20156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.746363759357779e-08,
+      "loss": 0.6924,
+      "step": 20157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.715667277159891e-08,
+      "loss": 0.917,
+      "step": 20158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.685019093309323e-08,
+      "loss": 0.8174,
+      "step": 20159
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.654419208549926e-08,
+      "loss": 0.9355,
+      "step": 20160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.623867623624439e-08,
+      "loss": 0.8701,
+      "step": 20161
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.593364339274491e-08,
+      "loss": 0.9141,
+      "step": 20162
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.562909356239713e-08,
+      "loss": 0.8896,
+      "step": 20163
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.532502675259514e-08,
+      "loss": 0.8926,
+      "step": 20164
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.502144297071747e-08,
+      "loss": 0.7354,
+      "step": 20165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.471834222412935e-08,
+      "loss": 0.791,
+      "step": 20166
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.441572452018932e-08,
+      "loss": 0.9629,
+      "step": 20167
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.411358986624042e-08,
+      "loss": 0.792,
+      "step": 20168
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.381193826961233e-08,
+      "loss": 0.8066,
+      "step": 20169
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.351076973762585e-08,
+      "loss": 0.8525,
+      "step": 20170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.321008427759293e-08,
+      "loss": 0.96,
+      "step": 20171
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.29098818968055e-08,
+      "loss": 0.7549,
+      "step": 20172
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.261016260254885e-08,
+      "loss": 0.7168,
+      "step": 20173
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.23109264021016e-08,
+      "loss": 0.918,
+      "step": 20174
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.201217330272016e-08,
+      "loss": 0.9941,
+      "step": 20175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.17139033116543e-08,
+      "loss": 0.8359,
+      "step": 20176
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.14161164361449e-08,
+      "loss": 0.875,
+      "step": 20177
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.111881268341948e-08,
+      "loss": 0.8105,
+      "step": 20178
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.082199206068787e-08,
+      "loss": 0.9512,
+      "step": 20179
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.052565457515761e-08,
+      "loss": 0.8213,
+      "step": 20180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.022980023401629e-08,
+      "loss": 0.8672,
+      "step": 20181
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.993442904444705e-08,
+      "loss": 0.6807,
+      "step": 20182
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.963954101361528e-08,
+      "loss": 0.7676,
+      "step": 20183
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.934513614867746e-08,
+      "loss": 0.8066,
+      "step": 20184
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.905121445677677e-08,
+      "loss": 0.8916,
+      "step": 20185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.875777594504753e-08,
+      "loss": 0.8799,
+      "step": 20186
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.846482062061068e-08,
+      "loss": 1.0049,
+      "step": 20187
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.81723484905761e-08,
+      "loss": 0.6333,
+      "step": 20188
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.78803595620381e-08,
+      "loss": 0.7656,
+      "step": 20189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.758885384208437e-08,
+      "loss": 0.8223,
+      "step": 20190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.729783133779146e-08,
+      "loss": 0.7588,
+      "step": 20191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.700729205621594e-08,
+      "loss": 0.8916,
+      "step": 20192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.671723600441217e-08,
+      "loss": 0.8652,
+      "step": 20193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.642766318941898e-08,
+      "loss": 0.8867,
+      "step": 20194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.613857361826405e-08,
+      "loss": 0.9336,
+      "step": 20195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.584996729795958e-08,
+      "loss": 0.6377,
+      "step": 20196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.556184423551106e-08,
+      "loss": 0.8906,
+      "step": 20197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.527420443791069e-08,
+      "loss": 0.8955,
+      "step": 20198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.498704791213729e-08,
+      "loss": 0.9199,
+      "step": 20199
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.47003746651609e-08,
+      "loss": 0.9453,
+      "step": 20200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.441418470393814e-08,
+      "loss": 0.8535,
+      "step": 20201
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.412847803541235e-08,
+      "loss": 0.8525,
+      "step": 20202
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.384325466652021e-08,
+      "loss": 0.7954,
+      "step": 20203
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.355851460417841e-08,
+      "loss": 0.9297,
+      "step": 20204
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.32742578553014e-08,
+      "loss": 0.7822,
+      "step": 20205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.29904844267837e-08,
+      "loss": 0.9141,
+      "step": 20206
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.270719432551311e-08,
+      "loss": 0.8945,
+      "step": 20207
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.242438755836413e-08,
+      "loss": 0.9756,
+      "step": 20208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.214206413220238e-08,
+      "loss": 0.8984,
+      "step": 20209
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.18602240538735e-08,
+      "loss": 0.7412,
+      "step": 20210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.157886733022092e-08,
+      "loss": 0.9102,
+      "step": 20211
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.129799396807026e-08,
+      "loss": 0.6799,
+      "step": 20212
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.101760397424052e-08,
+      "loss": 0.8594,
+      "step": 20213
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.073769735553294e-08,
+      "loss": 1.0049,
+      "step": 20214
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.045827411873985e-08,
+      "loss": 0.7236,
+      "step": 20215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.017933427064472e-08,
+      "loss": 0.8232,
+      "step": 20216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.990087781801325e-08,
+      "loss": 0.8115,
+      "step": 20217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.96229047676067e-08,
+      "loss": 0.916,
+      "step": 20218
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.934541512616856e-08,
+      "loss": 0.7783,
+      "step": 20219
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.906840890043122e-08,
+      "loss": 0.7793,
+      "step": 20220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.879188609711818e-08,
+      "loss": 0.8496,
+      "step": 20221
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.851584672293965e-08,
+      "loss": 0.8604,
+      "step": 20222
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.824029078459694e-08,
+      "loss": 0.7959,
+      "step": 20223
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.796521828877135e-08,
+      "loss": 0.6738,
+      "step": 20224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.769062924214422e-08,
+      "loss": 0.8213,
+      "step": 20225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.741652365137464e-08,
+      "loss": 0.6875,
+      "step": 20226
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.71429015231151e-08,
+      "loss": 0.8818,
+      "step": 20227
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.686976286400694e-08,
+      "loss": 0.9395,
+      "step": 20228
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.659710768067819e-08,
+      "loss": 0.8066,
+      "step": 20229
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.632493597974577e-08,
+      "loss": 0.9336,
+      "step": 20230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.605324776781331e-08,
+      "loss": 0.7295,
+      "step": 20231
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.57820430514733e-08,
+      "loss": 0.7236,
+      "step": 20232
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.551132183730936e-08,
+      "loss": 0.957,
+      "step": 20233
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.524108413188957e-08,
+      "loss": 0.8057,
+      "step": 20234
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.497132994177091e-08,
+      "loss": 0.7266,
+      "step": 20235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.470205927350149e-08,
+      "loss": 0.8535,
+      "step": 20236
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.443327213361385e-08,
+      "loss": 0.8955,
+      "step": 20237
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.416496852863386e-08,
+      "loss": 0.792,
+      "step": 20238
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.389714846506746e-08,
+      "loss": 0.8154,
+      "step": 20239
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.362981194942054e-08,
+      "loss": 0.9375,
+      "step": 20240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.336295898817458e-08,
+      "loss": 0.9717,
+      "step": 20241
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.309658958780663e-08,
+      "loss": 0.9873,
+      "step": 20242
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.283070375478262e-08,
+      "loss": 0.832,
+      "step": 20243
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.256530149555297e-08,
+      "loss": 0.7471,
+      "step": 20244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.230038281655915e-08,
+      "loss": 0.8047,
+      "step": 20245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.203594772423161e-08,
+      "loss": 0.646,
+      "step": 20246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.177199622498299e-08,
+      "loss": 0.8555,
+      "step": 20247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.150852832522148e-08,
+      "loss": 0.8047,
+      "step": 20248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.124554403134198e-08,
+      "loss": 0.8721,
+      "step": 20249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.098304334972383e-08,
+      "loss": 0.7988,
+      "step": 20250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.072102628673749e-08,
+      "loss": 0.8682,
+      "step": 20251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.045949284874232e-08,
+      "loss": 0.7148,
+      "step": 20252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.019844304208656e-08,
+      "loss": 0.7354,
+      "step": 20253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.993787687310072e-08,
+      "loss": 0.9043,
+      "step": 20254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.967779434811084e-08,
+      "loss": 0.792,
+      "step": 20255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.941819547342965e-08,
+      "loss": 0.9229,
+      "step": 20256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.915908025535211e-08,
+      "loss": 0.9287,
+      "step": 20257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.890044870017099e-08,
+      "loss": 0.5854,
+      "step": 20258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.864230081416123e-08,
+      "loss": 0.8691,
+      "step": 20259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.838463660358673e-08,
+      "loss": 0.8057,
+      "step": 20260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.812745607470028e-08,
+      "loss": 0.8984,
+      "step": 20261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.787075923374131e-08,
+      "loss": 0.7422,
+      "step": 20262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.761454608694262e-08,
+      "loss": 0.8232,
+      "step": 20263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.735881664051924e-08,
+      "loss": 0.8779,
+      "step": 20264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.710357090067732e-08,
+      "loss": 0.7832,
+      "step": 20265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.68488088736119e-08,
+      "loss": 0.8037,
+      "step": 20266
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.659453056550247e-08,
+      "loss": 0.9453,
+      "step": 20267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.63407359825241e-08,
+      "loss": 0.7783,
+      "step": 20268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.608742513083188e-08,
+      "loss": 0.7363,
+      "step": 20269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.583459801657644e-08,
+      "loss": 0.8906,
+      "step": 20270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.558225464588841e-08,
+      "loss": 0.875,
+      "step": 20271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.533039502489624e-08,
+      "loss": 0.8057,
+      "step": 20272
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.507901915970838e-08,
+      "loss": 0.7451,
+      "step": 20273
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.482812705642661e-08,
+      "loss": 0.7222,
+      "step": 20274
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.457771872113716e-08,
+      "loss": 0.7808,
+      "step": 20275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.432779415991963e-08,
+      "loss": 0.7383,
+      "step": 20276
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.407835337884027e-08,
+      "loss": 0.9482,
+      "step": 20277
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.382939638394758e-08,
+      "loss": 0.8887,
+      "step": 20278
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.358092318128561e-08,
+      "loss": 0.9062,
+      "step": 20279
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.333293377688288e-08,
+      "loss": 0.8594,
+      "step": 20280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.3085428176759e-08,
+      "loss": 0.7725,
+      "step": 20281
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.28384063869203e-08,
+      "loss": 0.8877,
+      "step": 20282
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.259186841335973e-08,
+      "loss": 0.8184,
+      "step": 20283
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.234581426205922e-08,
+      "loss": 0.668,
+      "step": 20284
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.210024393899394e-08,
+      "loss": 0.7031,
+      "step": 20285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.185515745011917e-08,
+      "loss": 0.8535,
+      "step": 20286
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.161055480138345e-08,
+      "loss": 0.8184,
+      "step": 20287
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.136643599872205e-08,
+      "loss": 0.9092,
+      "step": 20288
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.112280104806134e-08,
+      "loss": 1.0049,
+      "step": 20289
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.087964995531214e-08,
+      "loss": 0.9756,
+      "step": 20290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.063698272637419e-08,
+      "loss": 0.835,
+      "step": 20291
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.03947993671361e-08,
+      "loss": 0.793,
+      "step": 20292
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.015309988347762e-08,
+      "loss": 0.915,
+      "step": 20293
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.991188428126293e-08,
+      "loss": 0.7168,
+      "step": 20294
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.967115256634514e-08,
+      "loss": 0.8594,
+      "step": 20295
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.943090474456403e-08,
+      "loss": 0.834,
+      "step": 20296
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.91911408217527e-08,
+      "loss": 0.6953,
+      "step": 20297
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.8951860803728724e-08,
+      "loss": 1.0283,
+      "step": 20298
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.871306469629856e-08,
+      "loss": 0.6372,
+      "step": 20299
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.847475250525758e-08,
+      "loss": 0.9375,
+      "step": 20300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.823692423638783e-08,
+      "loss": 0.7891,
+      "step": 20301
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.799957989546246e-08,
+      "loss": 0.9072,
+      "step": 20302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.776271948823908e-08,
+      "loss": 0.6367,
+      "step": 20303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.752634302046645e-08,
+      "loss": 0.791,
+      "step": 20304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.7290450497879954e-08,
+      "loss": 0.792,
+      "step": 20305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.705504192620614e-08,
+      "loss": 0.6929,
+      "step": 20306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.6820117311156e-08,
+      "loss": 1.0283,
+      "step": 20307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.658567665843162e-08,
+      "loss": 0.9414,
+      "step": 20308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.6351719973719576e-08,
+      "loss": 0.6826,
+      "step": 20309
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.611824726269977e-08,
+      "loss": 0.874,
+      "step": 20310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.5885258531038766e-08,
+      "loss": 0.834,
+      "step": 20311
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.5652753784387614e-08,
+      "loss": 0.8574,
+      "step": 20312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.5420733028390686e-08,
+      "loss": 1.0205,
+      "step": 20313
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.51891962686768e-08,
+      "loss": 0.9453,
+      "step": 20314
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.4958143510863706e-08,
+      "loss": 0.7783,
+      "step": 20315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.472757476056245e-08,
+      "loss": 0.9463,
+      "step": 20316
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.449749002336635e-08,
+      "loss": 0.9639,
+      "step": 20317
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.4267889304859824e-08,
+      "loss": 0.7578,
+      "step": 20318
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.4038772610609526e-08,
+      "loss": 0.5864,
+      "step": 20319
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.3810139946182116e-08,
+      "loss": 0.9375,
+      "step": 20320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.358199131712205e-08,
+      "loss": 0.8057,
+      "step": 20321
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.335432672896712e-08,
+      "loss": 0.9492,
+      "step": 20322
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.31271461872418e-08,
+      "loss": 0.8125,
+      "step": 20323
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.2900449697457224e-08,
+      "loss": 0.6934,
+      "step": 20324
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.267423726511789e-08,
+      "loss": 0.9824,
+      "step": 20325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.2448508895712735e-08,
+      "loss": 0.7598,
+      "step": 20326
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.2223264594717385e-08,
+      "loss": 0.915,
+      "step": 20327
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.199850436759857e-08,
+      "loss": 0.8291,
+      "step": 20328
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1774228219814146e-08,
+      "loss": 0.7397,
+      "step": 20329
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1550436156799776e-08,
+      "loss": 0.8496,
+      "step": 20330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.13271281839911e-08,
+      "loss": 0.73,
+      "step": 20331
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.110430430680824e-08,
+      "loss": 0.8877,
+      "step": 20332
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.088196453065353e-08,
+      "loss": 0.8438,
+      "step": 20333
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.066010886092709e-08,
+      "loss": 0.9014,
+      "step": 20334
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.0438737303009074e-08,
+      "loss": 0.813,
+      "step": 20335
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.021784986227518e-08,
+      "loss": 0.6196,
+      "step": 20336
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.999744654408334e-08,
+      "loss": 0.6362,
+      "step": 20337
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.977752735378261e-08,
+      "loss": 0.9033,
+      "step": 20338
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9558092296708715e-08,
+      "loss": 0.9473,
+      "step": 20339
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9339141378188516e-08,
+      "loss": 0.7783,
+      "step": 20340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9120674603535534e-08,
+      "loss": 0.9316,
+      "step": 20341
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8902691978049974e-08,
+      "loss": 0.8203,
+      "step": 20342
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8685193507020946e-08,
+      "loss": 0.8447,
+      "step": 20343
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.846817919572866e-08,
+      "loss": 0.8623,
+      "step": 20344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.82516490494378e-08,
+      "loss": 0.7773,
+      "step": 20345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8035603073404156e-08,
+      "loss": 0.7368,
+      "step": 20346
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.782004127286799e-08,
+      "loss": 0.7695,
+      "step": 20347
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.760496365306511e-08,
+      "loss": 0.9131,
+      "step": 20348
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.739037021920912e-08,
+      "loss": 0.8984,
+      "step": 20349
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.71762609765114e-08,
+      "loss": 0.8574,
+      "step": 20350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.696263593016781e-08,
+      "loss": 0.7129,
+      "step": 20351
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.674949508535864e-08,
+      "loss": 0.9385,
+      "step": 20352
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.653683844726198e-08,
+      "loss": 0.8379,
+      "step": 20353
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.632466602103369e-08,
+      "loss": 0.8145,
+      "step": 20354
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6112977811823e-08,
+      "loss": 0.9746,
+      "step": 20355
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5901773824770235e-08,
+      "loss": 0.8633,
+      "step": 20356
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.569105406499574e-08,
+      "loss": 0.9395,
+      "step": 20357
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.548081853761765e-08,
+      "loss": 0.8262,
+      "step": 20358
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.527106724773633e-08,
+      "loss": 0.917,
+      "step": 20359
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.506180020044104e-08,
+      "loss": 1.0479,
+      "step": 20360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.485301740080994e-08,
+      "loss": 0.8838,
+      "step": 20361
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.4644718853910084e-08,
+      "loss": 0.8115,
+      "step": 20362
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.443690456479743e-08,
+      "loss": 0.613,
+      "step": 20363
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.4229574538512396e-08,
+      "loss": 0.7998,
+      "step": 20364
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.402272878008873e-08,
+      "loss": 0.7451,
+      "step": 20365
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.381636729454464e-08,
+      "loss": 0.9043,
+      "step": 20366
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.361049008688944e-08,
+      "loss": 0.9814,
+      "step": 20367
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3405097162116936e-08,
+      "loss": 0.8174,
+      "step": 20368
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.320018852521424e-08,
+      "loss": 0.8857,
+      "step": 20369
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.299576418114848e-08,
+      "loss": 0.7822,
+      "step": 20370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.279182413488681e-08,
+      "loss": 0.7852,
+      "step": 20371
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.2588368391376366e-08,
+      "loss": 0.8135,
+      "step": 20372
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.2385396955550996e-08,
+      "loss": 0.7676,
+      "step": 20373
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.218290983234008e-08,
+      "loss": 0.7734,
+      "step": 20374
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.198090702665525e-08,
+      "loss": 0.8955,
+      "step": 20375
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1779388543401465e-08,
+      "loss": 0.9268,
+      "step": 20376
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1578354387465935e-08,
+      "loss": 0.6836,
+      "step": 20377
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1377804563726974e-08,
+      "loss": 0.834,
+      "step": 20378
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1177739077054024e-08,
+      "loss": 0.8086,
+      "step": 20379
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0978157932298756e-08,
+      "loss": 0.8252,
+      "step": 20380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.07790611343084e-08,
+      "loss": 0.9287,
+      "step": 20381
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0580448687910225e-08,
+      "loss": 0.9375,
+      "step": 20382
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0382320597927016e-08,
+      "loss": 0.8164,
+      "step": 20383
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.018467686916605e-08,
+      "loss": 0.9004,
+      "step": 20384
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.998751750642349e-08,
+      "loss": 0.9258,
+      "step": 20385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.979084251448217e-08,
+      "loss": 0.7607,
+      "step": 20386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9594651898116064e-08,
+      "loss": 0.8018,
+      "step": 20387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9398945662088015e-08,
+      "loss": 0.7393,
+      "step": 20388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9203723811145345e-08,
+      "loss": 0.7344,
+      "step": 20389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.900898635002426e-08,
+      "loss": 0.8262,
+      "step": 20390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8814733283454306e-08,
+      "loss": 0.707,
+      "step": 20391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.862096461614506e-08,
+      "loss": 0.7729,
+      "step": 20392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8427680352801646e-08,
+      "loss": 0.8193,
+      "step": 20393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8234880498111416e-08,
+      "loss": 0.709,
+      "step": 20394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.80425650567573e-08,
+      "loss": 0.8945,
+      "step": 20395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7850734033404445e-08,
+      "loss": 0.9209,
+      "step": 20396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7659387432706914e-08,
+      "loss": 0.8828,
+      "step": 20397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.746852525930766e-08,
+      "loss": 0.792,
+      "step": 20398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.727814751784298e-08,
+      "loss": 0.9072,
+      "step": 20399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7088254212926946e-08,
+      "loss": 0.9092,
+      "step": 20400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.689884534917143e-08,
+      "loss": 0.7373,
+      "step": 20401
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.670992093117054e-08,
+      "loss": 0.6094,
+      "step": 20402
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6521480963509494e-08,
+      "loss": 0.8242,
+      "step": 20403
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.633352545076241e-08,
+      "loss": 0.7129,
+      "step": 20404
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6146054397490085e-08,
+      "loss": 0.7783,
+      "step": 20405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5959067808239986e-08,
+      "loss": 0.7363,
+      "step": 20406
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.577256568755294e-08,
+      "loss": 0.9189,
+      "step": 20407
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5586548039951984e-08,
+      "loss": 1.0312,
+      "step": 20408
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.540101486995129e-08,
+      "loss": 0.9033,
+      "step": 20409
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.521596618205614e-08,
+      "loss": 1.0273,
+      "step": 20410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5031401980754055e-08,
+      "loss": 0.832,
+      "step": 20411
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4847322270525897e-08,
+      "loss": 0.9307,
+      "step": 20412
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.466372705583698e-08,
+      "loss": 0.8936,
+      "step": 20413
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.448061634114375e-08,
+      "loss": 0.917,
+      "step": 20414
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4297990130889305e-08,
+      "loss": 0.6904,
+      "step": 20415
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.411584842950566e-08,
+      "loss": 0.9414,
+      "step": 20416
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3934191241411506e-08,
+      "loss": 0.8252,
+      "step": 20417
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375301857101887e-08,
+      "loss": 0.7988,
+      "step": 20418
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.357233042271979e-08,
+      "loss": 0.7227,
+      "step": 20419
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.339212680090187e-08,
+      "loss": 0.6772,
+      "step": 20420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.321240770993939e-08,
+      "loss": 0.5575,
+      "step": 20421
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3033173154188856e-08,
+      "loss": 0.9678,
+      "step": 20422
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2854423138002357e-08,
+      "loss": 0.8828,
+      "step": 20423
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2676157665720854e-08,
+      "loss": 0.8096,
+      "step": 20424
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2498376741665336e-08,
+      "loss": 0.8101,
+      "step": 20425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2321080370150137e-08,
+      "loss": 0.8799,
+      "step": 20426
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.214426855548069e-08,
+      "loss": 0.875,
+      "step": 20427
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1967941301946913e-08,
+      "loss": 0.9219,
+      "step": 20428
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1792098613827594e-08,
+      "loss": 0.8193,
+      "step": 20429
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.161674049539043e-08,
+      "loss": 0.748,
+      "step": 20430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.144186695088758e-08,
+      "loss": 0.7588,
+      "step": 20431
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1267477984566753e-08,
+      "loss": 0.8623,
+      "step": 20432
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.109357360066012e-08,
+      "loss": 0.9307,
+      "step": 20433
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.092015380338209e-08,
+      "loss": 0.8623,
+      "step": 20434
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.074721859694707e-08,
+      "loss": 0.8271,
+      "step": 20435
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.057476798554948e-08,
+      "loss": 0.9131,
+      "step": 20436
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0402801973374863e-08,
+      "loss": 0.8701,
+      "step": 20437
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.023132056459543e-08,
+      "loss": 0.8535,
+      "step": 20438
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.006032376337231e-08,
+      "loss": 0.9043,
+      "step": 20439
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.988981157385551e-08,
+      "loss": 0.8242,
+      "step": 20440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9719784000183937e-08,
+      "loss": 1.0332,
+      "step": 20441
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9550241046483187e-08,
+      "loss": 0.8486,
+      "step": 20442
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.938118271686552e-08,
+      "loss": 0.7583,
+      "step": 20443
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9212609015436543e-08,
+      "loss": 0.8896,
+      "step": 20444
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.90445199462841e-08,
+      "loss": 0.8955,
+      "step": 20445
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8876915513489366e-08,
+      "loss": 0.7705,
+      "step": 20446
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8709795721117984e-08,
+      "loss": 0.833,
+      "step": 20447
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8543160573228923e-08,
+      "loss": 0.873,
+      "step": 20448
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8377010073861177e-08,
+      "loss": 0.9395,
+      "step": 20449
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8211344227049297e-08,
+      "loss": 0.8184,
+      "step": 20450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.804616303681229e-08,
+      "loss": 0.9121,
+      "step": 20451
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7881466507162502e-08,
+      "loss": 0.6475,
+      "step": 20452
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7717254642090074e-08,
+      "loss": 0.9004,
+      "step": 20453
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7553527445585147e-08,
+      "loss": 0.7666,
+      "step": 20454
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.73902849216201e-08,
+      "loss": 0.7812,
+      "step": 20455
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.722752707415621e-08,
+      "loss": 0.7012,
+      "step": 20456
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.706525390713921e-08,
+      "loss": 0.7441,
+      "step": 20457
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.690346542451261e-08,
+      "loss": 0.8857,
+      "step": 20458
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6742161630199938e-08,
+      "loss": 0.6216,
+      "step": 20459
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.658134252811584e-08,
+      "loss": 0.8506,
+      "step": 20460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.642100812216164e-08,
+      "loss": 0.7129,
+      "step": 20461
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.626115841622978e-08,
+      "loss": 0.9482,
+      "step": 20462
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6101793414199384e-08,
+      "loss": 0.8047,
+      "step": 20463
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.594291311993624e-08,
+      "loss": 0.8711,
+      "step": 20464
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5784517537299492e-08,
+      "loss": 0.8086,
+      "step": 20465
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5626606670128285e-08,
+      "loss": 0.6919,
+      "step": 20466
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.546918052225733e-08,
+      "loss": 0.8662,
+      "step": 20467
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5312239097505796e-08,
+      "loss": 0.8457,
+      "step": 20468
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.515578239968397e-08,
+      "loss": 0.8818,
+      "step": 20469
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.499981043258659e-08,
+      "loss": 0.7871,
+      "step": 20470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.484432319999952e-08,
+      "loss": 0.6602,
+      "step": 20471
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4689320705697516e-08,
+      "loss": 0.7578,
+      "step": 20472
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4534802953437574e-08,
+      "loss": 0.876,
+      "step": 20473
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4380769946974468e-08,
+      "loss": 0.8369,
+      "step": 20474
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.422722169004299e-08,
+      "loss": 0.9492,
+      "step": 20475
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4074158186371265e-08,
+      "loss": 0.7881,
+      "step": 20476
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3921579439671882e-08,
+      "loss": 0.8799,
+      "step": 20477
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3769485453648544e-08,
+      "loss": 0.9482,
+      "step": 20478
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.361787623199163e-08,
+      "loss": 0.8223,
+      "step": 20479
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3466751778378205e-08,
+      "loss": 0.7397,
+      "step": 20480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3316112096480882e-08,
+      "loss": 0.7983,
+      "step": 20481
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3165957189950073e-08,
+      "loss": 0.8423,
+      "step": 20482
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3016287062433973e-08,
+      "loss": 0.8955,
+      "step": 20483
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2867101717560793e-08,
+      "loss": 0.8926,
+      "step": 20484
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2718401158952076e-08,
+      "loss": 0.6719,
+      "step": 20485
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2570185390218267e-08,
+      "loss": 0.8223,
+      "step": 20486
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2422454414954277e-08,
+      "loss": 0.7969,
+      "step": 20487
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2275208236743895e-08,
+      "loss": 0.7725,
+      "step": 20488
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2128446859162043e-08,
+      "loss": 0.7314,
+      "step": 20489
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.198217028577254e-08,
+      "loss": 0.7598,
+      "step": 20490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1836378520119217e-08,
+      "loss": 1.0127,
+      "step": 20491
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1691071565745903e-08,
+      "loss": 0.8301,
+      "step": 20492
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1546249426174225e-08,
+      "loss": 0.7744,
+      "step": 20493
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1401912104921373e-08,
+      "loss": 0.7383,
+      "step": 20494
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.125805960548899e-08,
+      "loss": 1.0137,
+      "step": 20495
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1114691931367615e-08,
+      "loss": 0.8027,
+      "step": 20496
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0971809086038906e-08,
+      "loss": 0.833,
+      "step": 20497
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0829411072966766e-08,
+      "loss": 0.8916,
+      "step": 20498
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0687497895608422e-08,
+      "loss": 0.8271,
+      "step": 20499
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.054606955740779e-08,
+      "loss": 0.833,
+      "step": 20500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0405126061795454e-08,
+      "loss": 1.0215,
+      "step": 20501
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0264667412193128e-08,
+      "loss": 0.8252,
+      "step": 20502
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0124693612009195e-08,
+      "loss": 0.8633,
+      "step": 20503
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9985204664640935e-08,
+      "loss": 0.7769,
+      "step": 20504
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9846200573472307e-08,
+      "loss": 0.8662,
+      "step": 20505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9707681341876174e-08,
+      "loss": 0.9014,
+      "step": 20506
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9569646973216506e-08,
+      "loss": 0.4761,
+      "step": 20507
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9432097470839516e-08,
+      "loss": 1.0293,
+      "step": 20508
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.929503283808476e-08,
+      "loss": 1.0674,
+      "step": 20509
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.915845307827846e-08,
+      "loss": 0.876,
+      "step": 20510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9022358194735745e-08,
+      "loss": 0.9375,
+      "step": 20511
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.888674819075842e-08,
+      "loss": 0.7407,
+      "step": 20512
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8751623069637183e-08,
+      "loss": 0.7793,
+      "step": 20513
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.861698283464941e-08,
+      "loss": 0.8594,
+      "step": 20514
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8482827489065824e-08,
+      "loss": 0.8291,
+      "step": 20515
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8349157036141597e-08,
+      "loss": 0.5879,
+      "step": 20516
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8215971479118578e-08,
+      "loss": 0.7812,
+      "step": 20517
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8083270821229737e-08,
+      "loss": 0.7568,
+      "step": 20518
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.795105506569472e-08,
+      "loss": 0.9277,
+      "step": 20519
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.78193242157243e-08,
+      "loss": 0.9932,
+      "step": 20520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7688078274511468e-08,
+      "loss": 0.8086,
+      "step": 20521
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7557317245242566e-08,
+      "loss": 0.916,
+      "step": 20522
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7427041131092835e-08,
+      "loss": 0.8604,
+      "step": 20523
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.729724993522197e-08,
+      "loss": 0.8359,
+      "step": 20524
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7167943660778562e-08,
+      "loss": 0.917,
+      "step": 20525
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7039122310902322e-08,
+      "loss": 1.0234,
+      "step": 20526
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.691078588871964e-08,
+      "loss": 0.8223,
+      "step": 20527
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6782934397343577e-08,
+      "loss": 0.8994,
+      "step": 20528
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6655567839878316e-08,
+      "loss": 0.7725,
+      "step": 20529
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6528686219412504e-08,
+      "loss": 0.7578,
+      "step": 20530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6402289539025896e-08,
+      "loss": 1.0605,
+      "step": 20531
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6276377801784928e-08,
+      "loss": 0.9463,
+      "step": 20532
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6150951010747152e-08,
+      "loss": 0.6963,
+      "step": 20533
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6026009168956802e-08,
+      "loss": 0.9902,
+      "step": 20534
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5901552279444787e-08,
+      "loss": 1.0166,
+      "step": 20535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5777580345228695e-08,
+      "loss": 0.8779,
+      "step": 20536
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5654093369319447e-08,
+      "loss": 0.7939,
+      "step": 20537
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5531091354714645e-08,
+      "loss": 0.8896,
+      "step": 20538
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.540857430439857e-08,
+      "loss": 0.7686,
+      "step": 20539
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5286542221342182e-08,
+      "loss": 0.7139,
+      "step": 20540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5164995108509773e-08,
+      "loss": 0.7998,
+      "step": 20541
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5043932968847873e-08,
+      "loss": 0.9365,
+      "step": 20542
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.492335580529858e-08,
+      "loss": 0.9346,
+      "step": 20543
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4803263620783991e-08,
+      "loss": 0.8965,
+      "step": 20544
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.468365641821956e-08,
+      "loss": 0.6665,
+      "step": 20545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4564534200509628e-08,
+      "loss": 0.7407,
+      "step": 20546
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4445896970542994e-08,
+      "loss": 0.9385,
+      "step": 20547
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4327744731197357e-08,
+      "loss": 0.8301,
+      "step": 20548
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4210077485341533e-08,
+      "loss": 0.8008,
+      "step": 20549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4092895235833236e-08,
+      "loss": 0.7656,
+      "step": 20550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3976197985512419e-08,
+      "loss": 0.873,
+      "step": 20551
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3859985737212368e-08,
+      "loss": 0.751,
+      "step": 20552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.374425849375527e-08,
+      "loss": 0.9414,
+      "step": 20553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3629016257945548e-08,
+      "loss": 0.7354,
+      "step": 20554
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3514259032583188e-08,
+      "loss": 1.0088,
+      "step": 20555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3399986820450406e-08,
+      "loss": 0.8096,
+      "step": 20556
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.328619962432276e-08,
+      "loss": 0.8389,
+      "step": 20557
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3172897446958044e-08,
+      "loss": 0.8818,
+      "step": 20558
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3060080291109612e-08,
+      "loss": 0.6934,
+      "step": 20559
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2947748159513051e-08,
+      "loss": 0.8633,
+      "step": 20560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2835901054895072e-08,
+      "loss": 0.7695,
+      "step": 20561
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2724538979969059e-08,
+      "loss": 0.8779,
+      "step": 20562
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2613661937439514e-08,
+      "loss": 0.7124,
+      "step": 20563
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2503269929995398e-08,
+      "loss": 0.915,
+      "step": 20564
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.239336296031457e-08,
+      "loss": 0.8887,
+      "step": 20565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2283941031066004e-08,
+      "loss": 0.8281,
+      "step": 20566
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2175004144903135e-08,
+      "loss": 0.8271,
+      "step": 20567
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2066552304472734e-08,
+      "loss": 0.9268,
+      "step": 20568
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1958585512403808e-08,
+      "loss": 0.8428,
+      "step": 20569
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1851103771318706e-08,
+      "loss": 0.8096,
+      "step": 20570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.174410708382201e-08,
+      "loss": 0.8936,
+      "step": 20571
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1637595452513862e-08,
+      "loss": 0.9834,
+      "step": 20572
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.153156887997664e-08,
+      "loss": 0.7812,
+      "step": 20573
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1426027368786063e-08,
+      "loss": 0.8022,
+      "step": 20574
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1320970921500085e-08,
+      "loss": 0.7529,
+      "step": 20575
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1216399540669998e-08,
+      "loss": 0.8945,
+      "step": 20576
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.111231322883155e-08,
+      "loss": 0.8652,
+      "step": 20577
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1008711988513832e-08,
+      "loss": 0.8633,
+      "step": 20578
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0905595822228165e-08,
+      "loss": 0.8535,
+      "step": 20579
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0802964732479215e-08,
+      "loss": 0.8555,
+      "step": 20580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.070081872175388e-08,
+      "loss": 0.7334,
+      "step": 20581
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0599157792536841e-08,
+      "loss": 0.9414,
+      "step": 20582
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0497981947288349e-08,
+      "loss": 0.9473,
+      "step": 20583
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0397291188466441e-08,
+      "loss": 0.5444,
+      "step": 20584
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0297085518518046e-08,
+      "loss": 1.001,
+      "step": 20585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0197364939870113e-08,
+      "loss": 0.8887,
+      "step": 20586
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0098129454942928e-08,
+      "loss": 0.7549,
+      "step": 20587
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999379066147896e-09,
+      "loss": 1.0596,
+      "step": 20588
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.901113775878658e-09,
+      "loss": 0.7129,
+      "step": 20589
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.80333358652219e-09,
+      "loss": 0.4868,
+      "step": 20590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.706038500447712e-09,
+      "loss": 0.8545,
+      "step": 20591
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.609228520017778e-09,
+      "loss": 0.9395,
+      "step": 20592
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.512903647583837e-09,
+      "loss": 0.8193,
+      "step": 20593
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.417063885481804e-09,
+      "loss": 0.8154,
+      "step": 20594
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.321709236038702e-09,
+      "loss": 0.9717,
+      "step": 20595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.226839701563794e-09,
+      "loss": 0.9014,
+      "step": 20596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.132455284364127e-09,
+      "loss": 0.9912,
+      "step": 20597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.0385559867312e-09,
+      "loss": 0.7949,
+      "step": 20598
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.945141810938751e-09,
+      "loss": 0.8682,
+      "step": 20599
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.852212759256073e-09,
+      "loss": 0.833,
+      "step": 20600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.759768833939142e-09,
+      "loss": 0.79,
+      "step": 20601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.667810037230606e-09,
+      "loss": 0.8662,
+      "step": 20602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.576336371362014e-09,
+      "loss": 0.7812,
+      "step": 20603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.485347838551594e-09,
+      "loss": 0.8174,
+      "step": 20604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.394844441010907e-09,
+      "loss": 0.6904,
+      "step": 20605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.304826180931536e-09,
+      "loss": 0.8701,
+      "step": 20606
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.215293060502837e-09,
+      "loss": 0.8174,
+      "step": 20607
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.12624508189419e-09,
+      "loss": 0.7939,
+      "step": 20608
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.037682247268308e-09,
+      "loss": 0.8584,
+      "step": 20609
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.949604558772362e-09,
+      "loss": 0.8521,
+      "step": 20610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.86201201854686e-09,
+      "loss": 0.874,
+      "step": 20611
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.774904628714552e-09,
+      "loss": 0.8486,
+      "step": 20612
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.688282391389301e-09,
+      "loss": 0.835,
+      "step": 20613
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.60214530867609e-09,
+      "loss": 0.9121,
+      "step": 20614
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.516493382662138e-09,
+      "loss": 0.7744,
+      "step": 20615
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.4313266154280024e-09,
+      "loss": 0.9551,
+      "step": 20616
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.3466450090387e-09,
+      "loss": 0.8877,
+      "step": 20617
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.262448565550362e-09,
+      "loss": 0.9434,
+      "step": 20618
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.178737287005799e-09,
+      "loss": 0.8975,
+      "step": 20619
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.09551117543672e-09,
+      "loss": 0.8477,
+      "step": 20620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.01277023286151e-09,
+      "loss": 0.8135,
+      "step": 20621
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.930514461289672e-09,
+      "loss": 0.8564,
+      "step": 20622
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.848743862717388e-09,
+      "loss": 0.8916,
+      "step": 20623
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.7674584391275166e-09,
+      "loss": 0.8604,
+      "step": 20624
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.686658192494033e-09,
+      "loss": 0.8096,
+      "step": 20625
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.606343124777592e-09,
+      "loss": 0.8389,
+      "step": 20626
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.5265132379255245e-09,
+      "loss": 0.8975,
+      "step": 20627
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.44716853387628e-09,
+      "loss": 0.6035,
+      "step": 20628
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3683090145572054e-09,
+      "loss": 0.7324,
+      "step": 20629
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.289934681877885e-09,
+      "loss": 0.8662,
+      "step": 20630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.212045537743461e-09,
+      "loss": 0.9385,
+      "step": 20631
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.134641584043533e-09,
+      "loss": 0.6692,
+      "step": 20632
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.0577228226565974e-09,
+      "loss": 0.8623,
+      "step": 20633
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.981289255447831e-09,
+      "loss": 0.9521,
+      "step": 20634
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.905340884271305e-09,
+      "loss": 0.9824,
+      "step": 20635
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.829877710972209e-09,
+      "loss": 0.8389,
+      "step": 20636
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.754899737382413e-09,
+      "loss": 0.7705,
+      "step": 20637
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.680406965320462e-09,
+      "loss": 0.7363,
+      "step": 20638
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.606399396593798e-09,
+      "loss": 0.7412,
+      "step": 20639
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.532877032996542e-09,
+      "loss": 0.7236,
+      "step": 20640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.459839876316153e-09,
+      "loss": 0.8984,
+      "step": 20641
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.387287928324547e-09,
+      "loss": 0.875,
+      "step": 20642
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.315221190780318e-09,
+      "loss": 0.7197,
+      "step": 20643
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.243639665433176e-09,
+      "loss": 0.6963,
+      "step": 20644
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.172543354021731e-09,
+      "loss": 0.8604,
+      "step": 20645
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.101932258269049e-09,
+      "loss": 0.8379,
+      "step": 20646
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.031806379889315e-09,
+      "loss": 0.6719,
+      "step": 20647
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.962165720583389e-09,
+      "loss": 0.7432,
+      "step": 20648
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8930102820454736e-09,
+      "loss": 0.9229,
+      "step": 20649
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.824340065947564e-09,
+      "loss": 0.7441,
+      "step": 20650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.756155073961655e-09,
+      "loss": 0.9189,
+      "step": 20651
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.688455307737538e-09,
+      "loss": 0.8594,
+      "step": 20652
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.621240768922786e-09,
+      "loss": 0.7852,
+      "step": 20653
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.554511459144984e-09,
+      "loss": 0.6943,
+      "step": 20654
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4882673800250575e-09,
+      "loss": 0.8193,
+      "step": 20655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.422508533170611e-09,
+      "loss": 0.7861,
+      "step": 20656
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.357234920175924e-09,
+      "loss": 0.7725,
+      "step": 20657
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2924465426286144e-09,
+      "loss": 1.0801,
+      "step": 20658
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.228143402096319e-09,
+      "loss": 0.8398,
+      "step": 20659
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.16432550014223e-09,
+      "loss": 0.9038,
+      "step": 20660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.100992838313999e-09,
+      "loss": 0.7939,
+      "step": 20661
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.038145418148176e-09,
+      "loss": 0.8682,
+      "step": 20662
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.975783241170206e-09,
+      "loss": 0.8926,
+      "step": 20663
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.913906308894433e-09,
+      "loss": 0.9082,
+      "step": 20664
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.852514622821879e-09,
+      "loss": 0.8711,
+      "step": 20665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.791608184442463e-09,
+      "loss": 0.834,
+      "step": 20666
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7311869952350034e-09,
+      "loss": 0.6533,
+      "step": 20667
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6712510566627724e-09,
+      "loss": 0.9102,
+      "step": 20668
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6118003701823835e-09,
+      "loss": 0.9092,
+      "step": 20669
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5528349372371263e-09,
+      "loss": 0.9531,
+      "step": 20670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.494354759256968e-09,
+      "loss": 0.957,
+      "step": 20671
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4363598376629948e-09,
+      "loss": 0.9805,
+      "step": 20672
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.378850173858528e-09,
+      "loss": 0.7847,
+      "step": 20673
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3218257692424484e-09,
+      "loss": 0.7822,
+      "step": 20674
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2652866251958736e-09,
+      "loss": 0.96,
+      "step": 20675
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.20923274309326e-09,
+      "loss": 0.8105,
+      "step": 20676
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.153664124295741e-09,
+      "loss": 0.8848,
+      "step": 20677
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0985807701489068e-09,
+      "loss": 0.8438,
+      "step": 20678
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0439826819894658e-09,
+      "loss": 0.8955,
+      "step": 20679
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9898698611452447e-09,
+      "loss": 0.5713,
+      "step": 20680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9362423089285277e-09,
+      "loss": 0.7637,
+      "step": 20681
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8831000266382747e-09,
+      "loss": 0.8291,
+      "step": 20682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.830443015566786e-09,
+      "loss": 0.8936,
+      "step": 20683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7782712769908183e-09,
+      "loss": 0.8203,
+      "step": 20684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.726584812176025e-09,
+      "loss": 0.8242,
+      "step": 20685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6753836223769592e-09,
+      "loss": 0.835,
+      "step": 20686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.62466770883707e-09,
+      "loss": 0.9014,
+      "step": 20687
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.574437072786484e-09,
+      "loss": 1.0176,
+      "step": 20688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.524691715442007e-09,
+      "loss": 0.8467,
+      "step": 20689
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4754316380137812e-09,
+      "loss": 0.8809,
+      "step": 20690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.426656841698627e-09,
+      "loss": 1.0527,
+      "step": 20691
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3783673276756016e-09,
+      "loss": 0.9072,
+      "step": 20692
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3305630971171002e-09,
+      "loss": 0.8555,
+      "step": 20693
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2832441511866366e-09,
+      "loss": 0.9199,
+      "step": 20694
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2364104910321817e-09,
+      "loss": 0.8682,
+      "step": 20695
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1900621177861625e-09,
+      "loss": 1.0361,
+      "step": 20696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.144199032576566e-09,
+      "loss": 0.7256,
+      "step": 20697
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.098821236518056e-09,
+      "loss": 0.8857,
+      "step": 20698
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0539287307053126e-09,
+      "loss": 0.8506,
+      "step": 20699
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0095215162352356e-09,
+      "loss": 0.9072,
+      "step": 20700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9655995941803007e-09,
+      "loss": 0.9277,
+      "step": 20701
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.922162965608543e-09,
+      "loss": 0.7607,
+      "step": 20702
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8792116315746733e-09,
+      "loss": 0.8486,
+      "step": 20703
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8367455931200817e-09,
+      "loss": 0.8682,
+      "step": 20704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7947648512728343e-09,
+      "loss": 0.6484,
+      "step": 20705
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.753269407056557e-09,
+      "loss": 0.9062,
+      "step": 20706
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7122592614748912e-09,
+      "loss": 1.002,
+      "step": 20707
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6717344155225967e-09,
+      "loss": 0.959,
+      "step": 20708
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6316948701855517e-09,
+      "loss": 0.7939,
+      "step": 20709
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5921406264340911e-09,
+      "loss": 0.9053,
+      "step": 20710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5530716852274474e-09,
+      "loss": 0.7197,
+      "step": 20711
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5144880475159719e-09,
+      "loss": 0.6572,
+      "step": 20712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4763897142322515e-09,
+      "loss": 0.625,
+      "step": 20713
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4387766863044327e-09,
+      "loss": 0.8926,
+      "step": 20714
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4016489646428987e-09,
+      "loss": 1.0361,
+      "step": 20715
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.365006550151371e-09,
+      "loss": 1.0361,
+      "step": 20716
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3288494437158072e-09,
+      "loss": 0.9238,
+      "step": 20717
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2931776462155043e-09,
+      "loss": 0.9131,
+      "step": 20718
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2579911585164362e-09,
+      "loss": 0.7666,
+      "step": 20719
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.223289981471254e-09,
+      "loss": 0.7705,
+      "step": 20720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1890741159215068e-09,
+      "loss": 0.8623,
+      "step": 20721
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.155343562699862e-09,
+      "loss": 0.8096,
+      "step": 20722
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1220983226234439e-09,
+      "loss": 0.6855,
+      "step": 20723
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.089338396498274e-09,
+      "loss": 0.8838,
+      "step": 20724
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0570637851192723e-09,
+      "loss": 0.7852,
+      "step": 20725
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0252744892724765e-09,
+      "loss": 0.8438,
+      "step": 20726
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.939705097261609e-10,
+      "loss": 0.8662,
+      "step": 20727
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.631518472419388e-10,
+      "loss": 1.0576,
+      "step": 20728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.328185025681003e-10,
+      "loss": 0.7588,
+      "step": 20729
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.029704764373926e-10,
+      "loss": 0.9082,
+      "step": 20730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.73607769578122e-10,
+      "loss": 0.9287,
+      "step": 20731
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.447303827008313e-10,
+      "loss": 0.6953,
+      "step": 20732
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.16338316504961e-10,
+      "loss": 0.8525,
+      "step": 20733
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.884315716832902e-10,
+      "loss": 1.0029,
+      "step": 20734
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.610101489086142e-10,
+      "loss": 0.9736,
+      "step": 20735
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.34074048849287e-10,
+      "loss": 0.8311,
+      "step": 20736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.0762327215812e-10,
+      "loss": 1.082,
+      "step": 20737
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.816578194790424e-10,
+      "loss": 0.8545,
+      "step": 20738
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.561776914382201e-10,
+      "loss": 0.7373,
+      "step": 20739
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.311828886573779e-10,
+      "loss": 0.9717,
+      "step": 20740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.066734117404772e-10,
+      "loss": 0.7686,
+      "step": 20741
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.826492612825974e-10,
+      "loss": 0.8198,
+      "step": 20742
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.591104378699363e-10,
+      "loss": 0.8164,
+      "step": 20743
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.360569420709283e-10,
+      "loss": 1.0,
+      "step": 20744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.134887744451256e-10,
+      "loss": 0.7842,
+      "step": 20745
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.914059355409783e-10,
+      "loss": 0.7676,
+      "step": 20746
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.698084258936142e-10,
+      "loss": 0.833,
+      "step": 20747
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.486962460270583e-10,
+      "loss": 0.7227,
+      "step": 20748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.280693964542337e-10,
+      "loss": 0.8154,
+      "step": 20749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0792787767474083e-10,
+      "loss": 0.8398,
+      "step": 20750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.882716901792982e-10,
+      "loss": 0.8682,
+      "step": 20751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.691008344430813e-10,
+      "loss": 0.7217,
+      "step": 20752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.504153109323838e-10,
+      "loss": 0.9336,
+      "step": 20753
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3221512010017663e-10,
+      "loss": 0.8145,
+      "step": 20754
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1450026238832863e-10,
+      "loss": 0.8105,
+      "step": 20755
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9727073822760634e-10,
+      "loss": 0.8115,
+      "step": 20756
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.805265480332331e-10,
+      "loss": 0.7266,
+      "step": 20757
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.642676922159915e-10,
+      "loss": 1.0059,
+      "step": 20758
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4849417116445953e-10,
+      "loss": 0.6846,
+      "step": 20759
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.332059852672153e-10,
+      "loss": 0.8613,
+      "step": 20760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.184031348928528e-10,
+      "loss": 0.7354,
+      "step": 20761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0408562039886393e-10,
+      "loss": 0.7388,
+      "step": 20762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.902534421360791e-10,
+      "loss": 0.7979,
+      "step": 20763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7690660043756525e-10,
+      "loss": 0.7139,
+      "step": 20764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6404509562972793e-10,
+      "loss": 0.8232,
+      "step": 20765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5166892802120915e-10,
+      "loss": 0.7222,
+      "step": 20766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3977809791621e-10,
+      "loss": 0.9531,
+      "step": 20767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.283726055989476e-10,
+      "loss": 0.8867,
+      "step": 20768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1745245134919814e-10,
+      "loss": 0.8525,
+      "step": 20769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0701763543341514e-10,
+      "loss": 0.6528,
+      "step": 20770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.706815810028858e-11,
+      "loss": 0.9336,
+      "step": 20771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.76040195940675e-11,
+      "loss": 0.6934,
+      "step": 20772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.86252201456783e-11,
+      "loss": 0.7202,
+      "step": 20773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.013175997050425e-11,
+      "loss": 0.9248,
+      "step": 20774
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.212363927504684e-11,
+      "loss": 0.7275,
+      "step": 20775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.4600858254705294e-11,
+      "loss": 0.7803,
+      "step": 20776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7563417091556205e-11,
+      "loss": 0.835,
+      "step": 20777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.101131595657393e-11,
+      "loss": 0.8643,
+      "step": 20778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.494455500741012e-11,
+      "loss": 0.8779,
+      "step": 20779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9363134392834667e-11,
+      "loss": 0.9746,
+      "step": 20780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4267054248294785e-11,
+      "loss": 0.8828,
+      "step": 20781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9656314695915003e-11,
+      "loss": 0.7456,
+      "step": 20782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5530915848938066e-11,
+      "loss": 0.8809,
+      "step": 20783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.189085780728405e-11,
+      "loss": 0.8896,
+      "step": 20784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.736140659770798e-12,
+      "loss": 0.7695,
+      "step": 20785
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.0667644818934725e-12,
+      "loss": 0.9189,
+      "step": 20786
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.882729338045011e-12,
+      "loss": 0.7241,
+      "step": 20787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.184035283736563e-12,
+      "loss": 0.916,
+      "step": 20788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.706823589361592e-13,
+      "loss": 0.8311,
+      "step": 20789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.426705902891513e-13,
+      "loss": 0.9385,
+      "step": 20790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.9941,
+      "step": 20791
+    },
+    {
+      "epoch": 1.0,
+      "step": 20791,
+      "total_flos": 1137093643567104.0,
+      "train_loss": 0.9329719317778726,
+      "train_runtime": 33469.7194,
+      "train_samples_per_second": 19.878,
+      "train_steps_per_second": 0.621
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 20791,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 2000,
+  "total_flos": 1137093643567104.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}