diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,62400 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999519022654033,
+  "eval_steps": 500,
+  "global_step": 10395,
+  "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": 2.436,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-07,
+      "loss": 2.3057,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-07,
+      "loss": 2.0269,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-07,
+      "loss": 2.2324,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.205128205128205e-07,
+      "loss": 2.3618,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 2.2393,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4871794871794876e-07,
+      "loss": 2.2236,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 2.3057,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.76923076923077e-07,
+      "loss": 2.1582,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.41025641025641e-07,
+      "loss": 2.1094,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.051282051282052e-07,
+      "loss": 2.2866,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 2.0894,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 2.2949,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.974358974358975e-07,
+      "loss": 2.2075,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.615384615384617e-07,
+      "loss": 2.2051,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 2.0312,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0897435897435899e-06,
+      "loss": 2.2637,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 2.1577,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.217948717948718e-06,
+      "loss": 1.7861,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 2.1133,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3461538461538462e-06,
+      "loss": 2.1226,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4102564102564104e-06,
+      "loss": 2.0601,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4743589743589745e-06,
+      "loss": 1.876,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 2.314,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.602564102564103e-06,
+      "loss": 1.9697,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.9395,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7307692307692308e-06,
+      "loss": 1.7041,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.794871794871795e-06,
+      "loss": 1.9536,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8589743589743592e-06,
+      "loss": 1.9541,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 1.7793,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.987179487179487e-06,
+      "loss": 1.894,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 1.7075,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1153846153846155e-06,
+      "loss": 1.8022,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1794871794871797e-06,
+      "loss": 1.9175,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.243589743589744e-06,
+      "loss": 1.8721,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.9487,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.371794871794872e-06,
+      "loss": 1.7212,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.435897435897436e-06,
+      "loss": 1.6406,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.7046,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 1.7998,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6282051282051286e-06,
+      "loss": 1.6396,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 1.9443,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.756410256410257e-06,
+      "loss": 1.7871,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8205128205128207e-06,
+      "loss": 1.6636,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8846153846153845e-06,
+      "loss": 1.6248,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.948717948717949e-06,
+      "loss": 1.5166,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.012820512820513e-06,
+      "loss": 1.9668,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.584,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.141025641025641e-06,
+      "loss": 1.7358,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.205128205128206e-06,
+      "loss": 1.6348,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2692307692307696e-06,
+      "loss": 1.5571,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.5557,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.397435897435898e-06,
+      "loss": 1.3689,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 1.4067,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5256410256410263e-06,
+      "loss": 1.3965,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 1.3401,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.653846153846154e-06,
+      "loss": 1.3494,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7179487179487184e-06,
+      "loss": 1.4937,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.782051282051282e-06,
+      "loss": 1.5039,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.4094,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.910256410256411e-06,
+      "loss": 1.3586,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.974358974358974e-06,
+      "loss": 1.3997,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0384615384615385e-06,
+      "loss": 1.439,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 1.3481,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.4487,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 1.3621,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.294871794871795e-06,
+      "loss": 1.4016,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.358974358974359e-06,
+      "loss": 1.312,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.423076923076924e-06,
+      "loss": 1.4023,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.487179487179488e-06,
+      "loss": 1.4106,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.551282051282052e-06,
+      "loss": 1.3621,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.3799,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6794871794871795e-06,
+      "loss": 1.4634,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.743589743589744e-06,
+      "loss": 1.3538,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.807692307692308e-06,
+      "loss": 1.4287,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 1.415,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.935897435897436e-06,
+      "loss": 1.4119,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 1.3582,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.064102564102565e-06,
+      "loss": 1.3604,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 1.4375,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.192307692307693e-06,
+      "loss": 1.3301,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.256410256410257e-06,
+      "loss": 1.3433,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.320512820512821e-06,
+      "loss": 1.2976,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 1.2434,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.448717948717949e-06,
+      "loss": 1.3792,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.512820512820514e-06,
+      "loss": 1.4016,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.576923076923077e-06,
+      "loss": 1.3638,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 1.3008,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.705128205128206e-06,
+      "loss": 1.4065,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 1.3975,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 1.3591,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.897435897435898e-06,
+      "loss": 1.1565,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.961538461538462e-06,
+      "loss": 1.2424,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.025641025641026e-06,
+      "loss": 1.3926,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.08974358974359e-06,
+      "loss": 1.2324,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.4419,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.217948717948718e-06,
+      "loss": 1.2852,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.282051282051282e-06,
+      "loss": 1.3481,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.3461538461538466e-06,
+      "loss": 1.3777,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.410256410256412e-06,
+      "loss": 1.3704,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.474358974358975e-06,
+      "loss": 1.2434,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 1.4548,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.602564102564103e-06,
+      "loss": 1.2148,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.3762,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.730769230769232e-06,
+      "loss": 1.4204,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.794871794871796e-06,
+      "loss": 1.2253,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.858974358974359e-06,
+      "loss": 1.2385,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 1.2915,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.9871794871794876e-06,
+      "loss": 1.2759,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.051282051282053e-06,
+      "loss": 1.2637,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.115384615384616e-06,
+      "loss": 1.3323,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.17948717948718e-06,
+      "loss": 1.3442,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.243589743589744e-06,
+      "loss": 1.1411,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 1.302,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.371794871794873e-06,
+      "loss": 1.238,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.435897435897437e-06,
+      "loss": 1.2688,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.2505,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.564102564102564e-06,
+      "loss": 1.1877,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.6282051282051286e-06,
+      "loss": 1.5371,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.3169,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.756410256410258e-06,
+      "loss": 1.4087,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.820512820512822e-06,
+      "loss": 1.1409,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.884615384615384e-06,
+      "loss": 1.2908,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.948717948717949e-06,
+      "loss": 1.3101,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.012820512820515e-06,
+      "loss": 1.3047,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 1.3071,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.141025641025641e-06,
+      "loss": 1.2261,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 1.2385,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.26923076923077e-06,
+      "loss": 1.3047,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.2817,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.397435897435898e-06,
+      "loss": 1.1111,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 1.2891,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.525641025641026e-06,
+      "loss": 1.3037,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.58974358974359e-06,
+      "loss": 1.2515,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.653846153846155e-06,
+      "loss": 1.1282,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 1.1975,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.782051282051283e-06,
+      "loss": 1.1467,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 1.1946,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.910256410256411e-06,
+      "loss": 1.2048,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 1.1936,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.03846153846154e-06,
+      "loss": 1.2542,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.102564102564104e-06,
+      "loss": 1.1541,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 1.1716,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.2261,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.294871794871796e-06,
+      "loss": 1.1714,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.358974358974359e-06,
+      "loss": 1.2175,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.423076923076923e-06,
+      "loss": 1.1838,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.487179487179487e-06,
+      "loss": 1.1311,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.551282051282053e-06,
+      "loss": 1.2922,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 1.1436,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.67948717948718e-06,
+      "loss": 1.262,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 1.2297,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.807692307692308e-06,
+      "loss": 1.1665,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.871794871794872e-06,
+      "loss": 1.2415,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.935897435897437e-06,
+      "loss": 1.1196,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 1.2974,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0064102564102565e-05,
+      "loss": 1.1123,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.012820512820513e-05,
+      "loss": 1.1326,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0192307692307692e-05,
+      "loss": 1.3125,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 1.1968,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0320512820512822e-05,
+      "loss": 1.0801,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 1.2676,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.044871794871795e-05,
+      "loss": 1.187,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0512820512820514e-05,
+      "loss": 1.1406,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0576923076923078e-05,
+      "loss": 1.2202,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0641025641025643e-05,
+      "loss": 1.1707,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0705128205128205e-05,
+      "loss": 1.1921,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 1.2415,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 1.1633,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0897435897435898e-05,
+      "loss": 1.1904,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0961538461538464e-05,
+      "loss": 1.1851,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1025641025641028e-05,
+      "loss": 1.1807,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1089743589743592e-05,
+      "loss": 1.1736,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 1.2605,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1217948717948719e-05,
+      "loss": 1.2668,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 1.2854,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1346153846153847e-05,
+      "loss": 1.2764,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1410256410256411e-05,
+      "loss": 1.1372,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1474358974358974e-05,
+      "loss": 1.1782,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 1.2639,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1602564102564104e-05,
+      "loss": 1.2065,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 1.2166,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1730769230769232e-05,
+      "loss": 1.2864,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1794871794871796e-05,
+      "loss": 1.1951,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.185897435897436e-05,
+      "loss": 1.0073,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 1.1921,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1987179487179487e-05,
+      "loss": 1.2627,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2051282051282051e-05,
+      "loss": 1.2427,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2115384615384615e-05,
+      "loss": 1.2683,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.217948717948718e-05,
+      "loss": 1.1621,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2243589743589746e-05,
+      "loss": 1.2351,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 1.1963,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2371794871794874e-05,
+      "loss": 1.2834,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2435897435897436e-05,
+      "loss": 1.1492,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25e-05,
+      "loss": 1.2656,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2564102564102565e-05,
+      "loss": 1.2349,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2628205128205129e-05,
+      "loss": 1.3167,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 1.2893,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2756410256410257e-05,
+      "loss": 1.2085,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2820512820512823e-05,
+      "loss": 1.3438,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2884615384615386e-05,
+      "loss": 1.1499,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.294871794871795e-05,
+      "loss": 1.1754,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3012820512820514e-05,
+      "loss": 1.2122,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 1.1838,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3141025641025642e-05,
+      "loss": 1.2297,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3205128205128207e-05,
+      "loss": 1.2388,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3269230769230769e-05,
+      "loss": 1.1506,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.1121,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3397435897435897e-05,
+      "loss": 1.2085,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3461538461538463e-05,
+      "loss": 1.1868,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3525641025641028e-05,
+      "loss": 1.2183,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3589743589743592e-05,
+      "loss": 1.2356,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3653846153846156e-05,
+      "loss": 1.1726,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3717948717948718e-05,
+      "loss": 1.2366,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3782051282051283e-05,
+      "loss": 1.29,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.3413,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3910256410256411e-05,
+      "loss": 1.1355,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3974358974358975e-05,
+      "loss": 1.2588,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.403846153846154e-05,
+      "loss": 1.0417,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4102564102564105e-05,
+      "loss": 1.1238,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.416666666666667e-05,
+      "loss": 1.0784,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4230769230769232e-05,
+      "loss": 1.1882,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4294871794871796e-05,
+      "loss": 1.1323,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.435897435897436e-05,
+      "loss": 1.3323,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4423076923076924e-05,
+      "loss": 1.2373,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4487179487179489e-05,
+      "loss": 1.2375,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4551282051282051e-05,
+      "loss": 1.134,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 1.1357,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.467948717948718e-05,
+      "loss": 1.2446,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4743589743589745e-05,
+      "loss": 1.1296,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.480769230769231e-05,
+      "loss": 1.167,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4871794871794874e-05,
+      "loss": 1.1394,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4935897435897438e-05,
+      "loss": 1.0975,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.2185,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5064102564102565e-05,
+      "loss": 1.3164,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5128205128205129e-05,
+      "loss": 1.2065,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5192307692307693e-05,
+      "loss": 1.2522,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5256410256410257e-05,
+      "loss": 1.2043,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5320512820512823e-05,
+      "loss": 1.2605,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.1772,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.544871794871795e-05,
+      "loss": 1.1482,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5512820512820516e-05,
+      "loss": 1.2375,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.557692307692308e-05,
+      "loss": 1.1045,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5641025641025644e-05,
+      "loss": 1.2935,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5705128205128205e-05,
+      "loss": 1.1788,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 1.1555,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 1.1301,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 1.1174,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5961538461538465e-05,
+      "loss": 1.1975,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.602564102564103e-05,
+      "loss": 1.2952,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6089743589743593e-05,
+      "loss": 1.2534,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 1.1604,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6217948717948718e-05,
+      "loss": 1.2953,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6282051282051282e-05,
+      "loss": 1.1282,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6346153846153847e-05,
+      "loss": 1.2537,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 1.3164,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6474358974358975e-05,
+      "loss": 1.2729,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 1.2759,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6602564102564103e-05,
+      "loss": 1.2434,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.209,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.673076923076923e-05,
+      "loss": 1.2397,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6794871794871796e-05,
+      "loss": 1.1731,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.685897435897436e-05,
+      "loss": 1.1892,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 1.1611,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.698717948717949e-05,
+      "loss": 1.0686,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7051282051282053e-05,
+      "loss": 1.1089,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7115384615384617e-05,
+      "loss": 1.1809,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.717948717948718e-05,
+      "loss": 1.1606,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7243589743589745e-05,
+      "loss": 1.1426,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 1.0637,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7371794871794873e-05,
+      "loss": 1.2288,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7435897435897438e-05,
+      "loss": 1.24,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 1.3469,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7564102564102566e-05,
+      "loss": 1.0889,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.762820512820513e-05,
+      "loss": 1.1707,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 1.2235,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.775641025641026e-05,
+      "loss": 1.2327,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7820512820512823e-05,
+      "loss": 1.1531,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7884615384615387e-05,
+      "loss": 1.1553,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 1.0991,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8012820512820515e-05,
+      "loss": 1.2859,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.807692307692308e-05,
+      "loss": 1.2101,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8141025641025644e-05,
+      "loss": 1.22,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8205128205128208e-05,
+      "loss": 1.144,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.826923076923077e-05,
+      "loss": 1.2017,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 1.2104,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8397435897435897e-05,
+      "loss": 1.2524,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.2244,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.852564102564103e-05,
+      "loss": 1.0933,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8589743589743593e-05,
+      "loss": 1.1643,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8653846153846157e-05,
+      "loss": 1.1799,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8717948717948718e-05,
+      "loss": 1.1392,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8782051282051282e-05,
+      "loss": 1.2273,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 1.1692,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.891025641025641e-05,
+      "loss": 1.2371,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 1.2842,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.903846153846154e-05,
+      "loss": 1.2175,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9102564102564106e-05,
+      "loss": 1.248,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.916666666666667e-05,
+      "loss": 1.2266,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 1.1565,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9294871794871796e-05,
+      "loss": 1.106,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.935897435897436e-05,
+      "loss": 1.2118,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9423076923076924e-05,
+      "loss": 1.2209,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 1.3008,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9551282051282052e-05,
+      "loss": 1.1704,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 1.1021,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.967948717948718e-05,
+      "loss": 1.2571,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9743589743589745e-05,
+      "loss": 1.1099,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.980769230769231e-05,
+      "loss": 1.1294,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9871794871794873e-05,
+      "loss": 1.2903,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9935897435897437e-05,
+      "loss": 1.0612,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.1438,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999999514610685e-05,
+      "loss": 1.2036,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999805844279e-05,
+      "loss": 1.25,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999563149645e-05,
+      "loss": 1.1934,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999992233771905e-05,
+      "loss": 1.1448,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999987865269485e-05,
+      "loss": 1.1653,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998252598961e-05,
+      "loss": 1.1953,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999976215932804e-05,
+      "loss": 1.2004,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999968935099675e-05,
+      "loss": 1.1938,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999960683490935e-05,
+      "loss": 1.2634,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999995146110738e-05,
+      "loss": 1.136,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999941267949908e-05,
+      "loss": 1.2068,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999930104019505e-05,
+      "loss": 1.1655,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999991796931726e-05,
+      "loss": 1.2256,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999904863844345e-05,
+      "loss": 1.1343,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999989078760204e-05,
+      "loss": 1.1519,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999875740591707e-05,
+      "loss": 1.1217,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999859722814806e-05,
+      "loss": 1.3066,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999842734272892e-05,
+      "loss": 1.1404,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999824774967615e-05,
+      "loss": 1.2615,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999980584490072e-05,
+      "loss": 1.1494,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999978594407404e-05,
+      "loss": 1.2786,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999765072489512e-05,
+      "loss": 1.2008,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999743230149164e-05,
+      "loss": 1.2004,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999972041705511e-05,
+      "loss": 1.1105,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999696633209563e-05,
+      "loss": 1.1926,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999967187861484e-05,
+      "loss": 1.2229,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999646153273343e-05,
+      "loss": 1.2634,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999961945718756e-05,
+      "loss": 1.2627,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999591790360097e-05,
+      "loss": 1.1238,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999956315279363e-05,
+      "loss": 1.0569,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999953354449094e-05,
+      "loss": 1.2024,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999502965454902e-05,
+      "loss": 1.1558,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999471415688484e-05,
+      "loss": 1.1516,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999943889519475e-05,
+      "loss": 1.198,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999940540397686e-05,
+      "loss": 1.1533,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999370942038065e-05,
+      "loss": 1.1689,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99993355093817e-05,
+      "loss": 1.1125,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999299106011216e-05,
+      "loss": 1.1792,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999926173193014e-05,
+      "loss": 1.1624,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999922338714211e-05,
+      "loss": 1.1958,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999918407165084e-05,
+      "loss": 1.2288,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999143785460148e-05,
+      "loss": 1.2166,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999102528573945e-05,
+      "loss": 1.146,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999060300996234e-05,
+      "loss": 1.1013,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999901710273112e-05,
+      "loss": 1.0286,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998972933782795e-05,
+      "loss": 1.2229,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998927794155543e-05,
+      "loss": 1.2183,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999888168385375e-05,
+      "loss": 1.1025,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999883460288189e-05,
+      "loss": 1.0955,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998786551244533e-05,
+      "loss": 1.2258,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999873752894635e-05,
+      "loss": 1.1282,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998687535992088e-05,
+      "loss": 1.1052,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999863657238661e-05,
+      "loss": 1.2012,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998584638134864e-05,
+      "loss": 1.0906,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998531733241884e-05,
+      "loss": 1.23,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998477857712815e-05,
+      "loss": 1.0996,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998423011552877e-05,
+      "loss": 1.1388,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998367194767404e-05,
+      "loss": 1.1909,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998310407361806e-05,
+      "loss": 1.1293,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998252649341607e-05,
+      "loss": 1.052,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99981939207124e-05,
+      "loss": 1.1824,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998134221479895e-05,
+      "loss": 1.2185,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999807355164989e-05,
+      "loss": 1.1772,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998011911228267e-05,
+      "loss": 1.165,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997949300221014e-05,
+      "loss": 1.1196,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999788571863421e-05,
+      "loss": 1.0695,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997821166474023e-05,
+      "loss": 1.0977,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997755643746724e-05,
+      "loss": 1.2996,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997689150458676e-05,
+      "loss": 1.1344,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997621686616326e-05,
+      "loss": 1.219,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997553252226226e-05,
+      "loss": 1.1021,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997483847295026e-05,
+      "loss": 1.2317,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999741347182946e-05,
+      "loss": 1.2017,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997342125836354e-05,
+      "loss": 1.1472,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997269809322638e-05,
+      "loss": 1.1182,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997196522295336e-05,
+      "loss": 1.1941,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999712226476156e-05,
+      "loss": 1.292,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997047036728518e-05,
+      "loss": 1.2649,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999697083820351e-05,
+      "loss": 1.2034,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999689366919394e-05,
+      "loss": 1.1938,
+      "step": 392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996815529707296e-05,
+      "loss": 1.127,
+      "step": 393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996736419751164e-05,
+      "loss": 1.1565,
+      "step": 394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996656339333224e-05,
+      "loss": 1.1481,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996575288461247e-05,
+      "loss": 1.1718,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996493267143107e-05,
+      "loss": 1.1444,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999641027538676e-05,
+      "loss": 1.1624,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996326313200266e-05,
+      "loss": 1.1256,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999624138059178e-05,
+      "loss": 1.2976,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999615547756954e-05,
+      "loss": 1.158,
+      "step": 401
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996068604141885e-05,
+      "loss": 1.2205,
+      "step": 402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995980760317256e-05,
+      "loss": 1.176,
+      "step": 403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995891946104175e-05,
+      "loss": 1.0532,
+      "step": 404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995802161511265e-05,
+      "loss": 1.0725,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999571140654724e-05,
+      "loss": 1.3059,
+      "step": 406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995619681220916e-05,
+      "loss": 1.2456,
+      "step": 407
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999552698554119e-05,
+      "loss": 1.1921,
+      "step": 408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999543331951707e-05,
+      "loss": 1.1702,
+      "step": 409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995338683157638e-05,
+      "loss": 1.2273,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999524307647209e-05,
+      "loss": 1.1643,
+      "step": 411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995146499469705e-05,
+      "loss": 1.1729,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999504895215986e-05,
+      "loss": 1.1423,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994950434552016e-05,
+      "loss": 1.111,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994850946655745e-05,
+      "loss": 1.1672,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994750488480706e-05,
+      "loss": 1.271,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994649060036642e-05,
+      "loss": 1.2192,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994546661333414e-05,
+      "loss": 1.3079,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999444329238095e-05,
+      "loss": 1.2097,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994338953189293e-05,
+      "loss": 1.0671,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999423364376856e-05,
+      "loss": 1.1743,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999412736412899e-05,
+      "loss": 1.1677,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994020114280892e-05,
+      "loss": 1.199,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999391189423468e-05,
+      "loss": 1.3171,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993802704000857e-05,
+      "loss": 1.1743,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999369254359002e-05,
+      "loss": 1.1462,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999358141301287e-05,
+      "loss": 1.1609,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993469312280193e-05,
+      "loss": 1.1404,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993356241402876e-05,
+      "loss": 1.2893,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993242200391883e-05,
+      "loss": 1.149,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993127189258298e-05,
+      "loss": 1.1848,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999301120801328e-05,
+      "loss": 1.2773,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992894256668086e-05,
+      "loss": 1.2485,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999277633523408e-05,
+      "loss": 1.281,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992657443722697e-05,
+      "loss": 1.2305,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999253758214548e-05,
+      "loss": 1.1687,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992416750514074e-05,
+      "loss": 1.239,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992294948840204e-05,
+      "loss": 1.1993,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999217217713569e-05,
+      "loss": 1.1658,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992048435412457e-05,
+      "loss": 1.1436,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991923723682514e-05,
+      "loss": 1.1886,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991798041957972e-05,
+      "loss": 1.1191,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991671390251024e-05,
+      "loss": 1.2583,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991543768573974e-05,
+      "loss": 1.0568,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991415176939203e-05,
+      "loss": 1.2642,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99912856153592e-05,
+      "loss": 1.1758,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991155083846537e-05,
+      "loss": 1.1537,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991023582413897e-05,
+      "loss": 1.1719,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990891111074037e-05,
+      "loss": 1.2172,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990757669839812e-05,
+      "loss": 1.1746,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999062325872419e-05,
+      "loss": 1.28,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999048787774021e-05,
+      "loss": 1.1895,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990351526901012e-05,
+      "loss": 1.3025,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990214206219843e-05,
+      "loss": 1.1443,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990075915710028e-05,
+      "loss": 0.9993,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998993665538499e-05,
+      "loss": 1.1802,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998979642525825e-05,
+      "loss": 1.239,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989655225343422e-05,
+      "loss": 1.1726,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998951305565421e-05,
+      "loss": 1.2587,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998936991620442e-05,
+      "loss": 1.0608,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989225807007947e-05,
+      "loss": 1.1021,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989080728078778e-05,
+      "loss": 1.1599,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988934679431e-05,
+      "loss": 1.0774,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998878766107879e-05,
+      "loss": 1.136,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998863967303642e-05,
+      "loss": 1.0879,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988490715318253e-05,
+      "loss": 1.1838,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988340787938754e-05,
+      "loss": 1.1663,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988189890912482e-05,
+      "loss": 1.0837,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988038024254077e-05,
+      "loss": 1.082,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987885187978283e-05,
+      "loss": 1.1316,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998773138209994e-05,
+      "loss": 1.136,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987576606633978e-05,
+      "loss": 1.1653,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987420861595425e-05,
+      "loss": 1.1897,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987264146999396e-05,
+      "loss": 1.1941,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987106462861107e-05,
+      "loss": 1.238,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986947809195865e-05,
+      "loss": 1.135,
+      "step": 476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998678818601907e-05,
+      "loss": 1.199,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998662759334622e-05,
+      "loss": 1.0879,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998646603119291e-05,
+      "loss": 1.0963,
+      "step": 479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998630349957481e-05,
+      "loss": 1.1997,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986139998507715e-05,
+      "loss": 1.2057,
+      "step": 481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985975528007488e-05,
+      "loss": 1.1992,
+      "step": 482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985810088090093e-05,
+      "loss": 1.1899,
+      "step": 483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985643678771595e-05,
+      "loss": 1.0757,
+      "step": 484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985476300068152e-05,
+      "loss": 1.1694,
+      "step": 485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985307951996005e-05,
+      "loss": 1.1649,
+      "step": 486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985138634571504e-05,
+      "loss": 1.1847,
+      "step": 487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984968347811083e-05,
+      "loss": 1.1748,
+      "step": 488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984797091731267e-05,
+      "loss": 1.2598,
+      "step": 489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998462486634869e-05,
+      "loss": 1.1277,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998445167168007e-05,
+      "loss": 1.1296,
+      "step": 491
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998427750774222e-05,
+      "loss": 1.2749,
+      "step": 492
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984102374552043e-05,
+      "loss": 1.2632,
+      "step": 493
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983926272126544e-05,
+      "loss": 1.1794,
+      "step": 494
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983749200482816e-05,
+      "loss": 1.1355,
+      "step": 495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983571159638053e-05,
+      "loss": 1.1765,
+      "step": 496
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983392149609536e-05,
+      "loss": 1.2373,
+      "step": 497
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983212170414648e-05,
+      "loss": 1.166,
+      "step": 498
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998303122207085e-05,
+      "loss": 1.1577,
+      "step": 499
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998284930459572e-05,
+      "loss": 1.293,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982666418006912e-05,
+      "loss": 1.0942,
+      "step": 501
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998248256232218e-05,
+      "loss": 1.321,
+      "step": 502
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982297737559372e-05,
+      "loss": 1.2964,
+      "step": 503
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982111943736435e-05,
+      "loss": 1.0769,
+      "step": 504
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981925180871398e-05,
+      "loss": 1.1973,
+      "step": 505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99817374489824e-05,
+      "loss": 1.2668,
+      "step": 506
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998154874808766e-05,
+      "loss": 1.1724,
+      "step": 507
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99813590782055e-05,
+      "loss": 1.2295,
+      "step": 508
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998116843935433e-05,
+      "loss": 1.0977,
+      "step": 509
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980976831552653e-05,
+      "loss": 1.1987,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998078425481908e-05,
+      "loss": 1.1118,
+      "step": 511
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99805907091723e-05,
+      "loss": 1.2422,
+      "step": 512
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980396194631104e-05,
+      "loss": 1.2214,
+      "step": 513
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980200711214367e-05,
+      "loss": 1.2073,
+      "step": 514
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980004258941078e-05,
+      "loss": 1.1877,
+      "step": 515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979806837830305e-05,
+      "loss": 1.1594,
+      "step": 516
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979608447901207e-05,
+      "loss": 1.1843,
+      "step": 517
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997940908917305e-05,
+      "loss": 1.1321,
+      "step": 518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997920876166518e-05,
+      "loss": 1.3511,
+      "step": 519
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979007465397054e-05,
+      "loss": 1.2559,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978805200388205e-05,
+      "loss": 1.1484,
+      "step": 521
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978601966658277e-05,
+      "loss": 1.1266,
+      "step": 522
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997839776422699e-05,
+      "loss": 1.0881,
+      "step": 523
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978192593114173e-05,
+      "loss": 1.2341,
+      "step": 524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997798645333974e-05,
+      "loss": 1.2275,
+      "step": 525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977779344923708e-05,
+      "loss": 1.147,
+      "step": 526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997757126788618e-05,
+      "loss": 1.1997,
+      "step": 527
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977362222247353e-05,
+      "loss": 1.0985,
+      "step": 528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977152208027525e-05,
+      "loss": 1.2214,
+      "step": 529
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976941225247085e-05,
+      "loss": 1.1187,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997672927392651e-05,
+      "loss": 1.252,
+      "step": 531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976516354086374e-05,
+      "loss": 1.0767,
+      "step": 532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976302465747354e-05,
+      "loss": 1.2463,
+      "step": 533
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976087608930208e-05,
+      "loss": 1.0693,
+      "step": 534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975871783655794e-05,
+      "loss": 1.1904,
+      "step": 535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997565498994507e-05,
+      "loss": 1.1526,
+      "step": 536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975437227819076e-05,
+      "loss": 1.1802,
+      "step": 537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975218497298953e-05,
+      "loss": 1.1584,
+      "step": 538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974998798405933e-05,
+      "loss": 1.1147,
+      "step": 539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997477813116135e-05,
+      "loss": 1.0684,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997455649558662e-05,
+      "loss": 1.1384,
+      "step": 541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974333891703262e-05,
+      "loss": 1.0752,
+      "step": 542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974110319532886e-05,
+      "loss": 1.0859,
+      "step": 543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973885779097192e-05,
+      "loss": 1.2695,
+      "step": 544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973660270417984e-05,
+      "loss": 1.2207,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997343379351715e-05,
+      "loss": 1.1808,
+      "step": 546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973206348416676e-05,
+      "loss": 1.0273,
+      "step": 547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997297793513864e-05,
+      "loss": 1.2007,
+      "step": 548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972748553705225e-05,
+      "loss": 1.1956,
+      "step": 549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997251820413869e-05,
+      "loss": 1.0073,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972286886461396e-05,
+      "loss": 1.1284,
+      "step": 551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972054600695802e-05,
+      "loss": 1.1318,
+      "step": 552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997182134686446e-05,
+      "loss": 1.1024,
+      "step": 553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997158712499001e-05,
+      "loss": 1.1716,
+      "step": 554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971351935095194e-05,
+      "loss": 1.156,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971115777202836e-05,
+      "loss": 1.1646,
+      "step": 556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997087865133587e-05,
+      "loss": 1.1958,
+      "step": 557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970640557517314e-05,
+      "loss": 1.2031,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970401495770275e-05,
+      "loss": 1.1431,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997016146611797e-05,
+      "loss": 1.123,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969920468583693e-05,
+      "loss": 1.1523,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969678503190845e-05,
+      "loss": 1.1475,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969435569962915e-05,
+      "loss": 1.1149,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996919166892348e-05,
+      "loss": 1.0894,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968946800096226e-05,
+      "loss": 1.2256,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968700963504918e-05,
+      "loss": 1.2017,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968454159173426e-05,
+      "loss": 1.177,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99682063871257e-05,
+      "loss": 1.1152,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967957647385806e-05,
+      "loss": 1.1744,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967707939977887e-05,
+      "loss": 1.1816,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996745726492618e-05,
+      "loss": 1.1682,
+      "step": 571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967205622255022e-05,
+      "loss": 1.1409,
+      "step": 572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966953011988842e-05,
+      "loss": 1.1096,
+      "step": 573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966699434152162e-05,
+      "loss": 1.1177,
+      "step": 574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99664448887696e-05,
+      "loss": 1.2786,
+      "step": 575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966189375865868e-05,
+      "loss": 1.1235,
+      "step": 576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965932895465768e-05,
+      "loss": 1.0657,
+      "step": 577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99656754475942e-05,
+      "loss": 1.072,
+      "step": 578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965417032276157e-05,
+      "loss": 1.145,
+      "step": 579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965157649536725e-05,
+      "loss": 1.1794,
+      "step": 580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964897299401086e-05,
+      "loss": 1.188,
+      "step": 581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964635981894505e-05,
+      "loss": 1.3494,
+      "step": 582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964373697042366e-05,
+      "loss": 1.1405,
+      "step": 583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964110444870118e-05,
+      "loss": 1.093,
+      "step": 584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996384622540332e-05,
+      "loss": 1.1396,
+      "step": 585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963581038667624e-05,
+      "loss": 1.1572,
+      "step": 586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963314884688777e-05,
+      "loss": 1.3357,
+      "step": 587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996304776349261e-05,
+      "loss": 1.1106,
+      "step": 588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962779675105056e-05,
+      "loss": 1.2175,
+      "step": 589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962510619552144e-05,
+      "loss": 1.2131,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962240596859987e-05,
+      "loss": 1.0178,
+      "step": 591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961969607054806e-05,
+      "loss": 1.0991,
+      "step": 592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961697650162904e-05,
+      "loss": 1.0664,
+      "step": 593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996142472621068e-05,
+      "loss": 1.1748,
+      "step": 594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961150835224635e-05,
+      "loss": 1.1692,
+      "step": 595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996087597723135e-05,
+      "loss": 1.2031,
+      "step": 596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960600152257513e-05,
+      "loss": 1.1199,
+      "step": 597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960323360329902e-05,
+      "loss": 1.0674,
+      "step": 598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996004560147538e-05,
+      "loss": 1.2393,
+      "step": 599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959766875720915e-05,
+      "loss": 1.1934,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995948718309357e-05,
+      "loss": 1.1848,
+      "step": 601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959206523620494e-05,
+      "loss": 1.145,
+      "step": 602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958924897328925e-05,
+      "loss": 1.1239,
+      "step": 603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958642304246215e-05,
+      "loss": 1.1736,
+      "step": 604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958358744399788e-05,
+      "loss": 1.2903,
+      "step": 605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995807421781718e-05,
+      "loss": 1.1353,
+      "step": 606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957788724526e-05,
+      "loss": 1.3271,
+      "step": 607
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995750226455398e-05,
+      "loss": 1.1245,
+      "step": 608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957214837928913e-05,
+      "loss": 1.1168,
+      "step": 609
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956926444678707e-05,
+      "loss": 1.1135,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956637084831362e-05,
+      "loss": 1.1252,
+      "step": 611
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956346758414968e-05,
+      "loss": 1.2421,
+      "step": 612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956055465457707e-05,
+      "loss": 1.1582,
+      "step": 613
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955763205987857e-05,
+      "loss": 0.9917,
+      "step": 614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955469980033787e-05,
+      "loss": 1.3059,
+      "step": 615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955175787623974e-05,
+      "loss": 1.1165,
+      "step": 616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954880628786963e-05,
+      "loss": 1.1912,
+      "step": 617
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954584503551414e-05,
+      "loss": 1.1127,
+      "step": 618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995428741194608e-05,
+      "loss": 1.2344,
+      "step": 619
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995398935399979e-05,
+      "loss": 1.2507,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995369032974149e-05,
+      "loss": 1.1931,
+      "step": 621
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953390339200202e-05,
+      "loss": 1.3354,
+      "step": 622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953089382405052e-05,
+      "loss": 1.2476,
+      "step": 623
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952787459385254e-05,
+      "loss": 1.155,
+      "step": 624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952484570170117e-05,
+      "loss": 1.1409,
+      "step": 625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952180714789048e-05,
+      "loss": 1.0698,
+      "step": 626
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995187589327154e-05,
+      "loss": 1.0837,
+      "step": 627
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951570105647192e-05,
+      "loss": 1.134,
+      "step": 628
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995126335194568e-05,
+      "loss": 1.1809,
+      "step": 629
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950955632196792e-05,
+      "loss": 1.1052,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950646946430394e-05,
+      "loss": 1.04,
+      "step": 631
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950337294676456e-05,
+      "loss": 1.2085,
+      "step": 632
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950026676965037e-05,
+      "loss": 1.0796,
+      "step": 633
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994971509332629e-05,
+      "loss": 1.1091,
+      "step": 634
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949402543790468e-05,
+      "loss": 1.1195,
+      "step": 635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949089028387908e-05,
+      "loss": 1.2446,
+      "step": 636
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948774547149044e-05,
+      "loss": 1.1829,
+      "step": 637
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948459100104412e-05,
+      "loss": 1.1501,
+      "step": 638
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948142687284625e-05,
+      "loss": 1.1504,
+      "step": 639
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994782530872041e-05,
+      "loss": 1.1412,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947506964442567e-05,
+      "loss": 1.0911,
+      "step": 641
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994718765448201e-05,
+      "loss": 1.1545,
+      "step": 642
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946867378869733e-05,
+      "loss": 1.2974,
+      "step": 643
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946546137636825e-05,
+      "loss": 1.105,
+      "step": 644
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946223930814475e-05,
+      "loss": 1.1919,
+      "step": 645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945900758433963e-05,
+      "loss": 1.0002,
+      "step": 646
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945576620526656e-05,
+      "loss": 1.1958,
+      "step": 647
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945251517124026e-05,
+      "loss": 1.1296,
+      "step": 648
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944925448257633e-05,
+      "loss": 1.1885,
+      "step": 649
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994459841395913e-05,
+      "loss": 1.1282,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944270414260263e-05,
+      "loss": 1.1304,
+      "step": 651
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943941449192878e-05,
+      "loss": 1.1355,
+      "step": 652
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99436115187889e-05,
+      "loss": 1.3105,
+      "step": 653
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994328062308037e-05,
+      "loss": 1.1506,
+      "step": 654
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994294876209941e-05,
+      "loss": 1.1326,
+      "step": 655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942615935878225e-05,
+      "loss": 1.1898,
+      "step": 656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942282144449138e-05,
+      "loss": 1.1741,
+      "step": 657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941947387844543e-05,
+      "loss": 1.1913,
+      "step": 658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941611666096948e-05,
+      "loss": 1.1177,
+      "step": 659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994127497923893e-05,
+      "loss": 1.2856,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940937327303186e-05,
+      "loss": 1.1223,
+      "step": 661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994059871032249e-05,
+      "loss": 1.1592,
+      "step": 662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940259128329714e-05,
+      "loss": 1.228,
+      "step": 663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939918581357826e-05,
+      "loss": 1.1292,
+      "step": 664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939577069439878e-05,
+      "loss": 1.1533,
+      "step": 665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939234592609034e-05,
+      "loss": 1.1824,
+      "step": 666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938891150898535e-05,
+      "loss": 1.1919,
+      "step": 667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938546744341724e-05,
+      "loss": 1.1836,
+      "step": 668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993820137297203e-05,
+      "loss": 1.2063,
+      "step": 669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937855036822987e-05,
+      "loss": 1.0452,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937507735928214e-05,
+      "loss": 1.3882,
+      "step": 671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937159470321424e-05,
+      "loss": 1.1294,
+      "step": 672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993681024003643e-05,
+      "loss": 1.1536,
+      "step": 673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993646004510714e-05,
+      "loss": 1.293,
+      "step": 674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936108885567534e-05,
+      "loss": 1.1807,
+      "step": 675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935756761451716e-05,
+      "loss": 1.1941,
+      "step": 676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993540367279386e-05,
+      "loss": 1.2561,
+      "step": 677
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935049619628252e-05,
+      "loss": 1.1072,
+      "step": 678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934694601989254e-05,
+      "loss": 1.0468,
+      "step": 679
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993433861991134e-05,
+      "loss": 1.3171,
+      "step": 680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933981673429056e-05,
+      "loss": 1.1761,
+      "step": 681
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933623762577062e-05,
+      "loss": 1.1833,
+      "step": 682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933264887390106e-05,
+      "loss": 1.2925,
+      "step": 683
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932905047903017e-05,
+      "loss": 1.1353,
+      "step": 684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932544244150735e-05,
+      "loss": 1.1223,
+      "step": 685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993218247616828e-05,
+      "loss": 1.1998,
+      "step": 686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993181974399078e-05,
+      "loss": 1.1465,
+      "step": 687
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931456047653444e-05,
+      "loss": 1.1624,
+      "step": 688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931091387191576e-05,
+      "loss": 1.1335,
+      "step": 689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993072576264058e-05,
+      "loss": 1.1655,
+      "step": 690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930359174035946e-05,
+      "loss": 1.0709,
+      "step": 691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992999162141327e-05,
+      "loss": 1.1361,
+      "step": 692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929623104808224e-05,
+      "loss": 1.2207,
+      "step": 693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929253624256584e-05,
+      "loss": 1.2278,
+      "step": 694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928883179794227e-05,
+      "loss": 1.27,
+      "step": 695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928511771457108e-05,
+      "loss": 1.22,
+      "step": 696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928139399281282e-05,
+      "loss": 1.1643,
+      "step": 697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99277660633029e-05,
+      "loss": 1.2324,
+      "step": 698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927391763558203e-05,
+      "loss": 1.1176,
+      "step": 699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992701650008353e-05,
+      "loss": 1.1689,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992664027291531e-05,
+      "loss": 1.155,
+      "step": 701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926263082090064e-05,
+      "loss": 1.2051,
+      "step": 702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992588492764441e-05,
+      "loss": 1.1056,
+      "step": 703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992550580961506e-05,
+      "loss": 1.2285,
+      "step": 704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925125728038816e-05,
+      "loss": 1.2271,
+      "step": 705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992474468295258e-05,
+      "loss": 1.1382,
+      "step": 706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924362674393336e-05,
+      "loss": 1.1069,
+      "step": 707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923979702398172e-05,
+      "loss": 1.3269,
+      "step": 708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923595767004268e-05,
+      "loss": 1.1578,
+      "step": 709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923210868248893e-05,
+      "loss": 1.2024,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992282500616941e-05,
+      "loss": 1.0826,
+      "step": 711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922438180803284e-05,
+      "loss": 1.1782,
+      "step": 712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922050392188062e-05,
+      "loss": 1.3306,
+      "step": 713
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992166164036139e-05,
+      "loss": 1.1208,
+      "step": 714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992127192536101e-05,
+      "loss": 1.1746,
+      "step": 715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920881247224756e-05,
+      "loss": 1.1675,
+      "step": 716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992048960599055e-05,
+      "loss": 1.1387,
+      "step": 717
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992009700169641e-05,
+      "loss": 1.0557,
+      "step": 718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919703434380457e-05,
+      "loss": 1.0665,
+      "step": 719
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919308904080893e-05,
+      "loss": 1.1611,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918913410836013e-05,
+      "loss": 1.282,
+      "step": 721
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991851695468422e-05,
+      "loss": 1.1396,
+      "step": 722
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918119535664e-05,
+      "loss": 1.2949,
+      "step": 723
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917721153813924e-05,
+      "loss": 1.1443,
+      "step": 724
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917321809172676e-05,
+      "loss": 1.2791,
+      "step": 725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916921501779018e-05,
+      "loss": 1.1265,
+      "step": 726
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916520231671815e-05,
+      "loss": 1.1541,
+      "step": 727
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991611799889002e-05,
+      "loss": 1.2107,
+      "step": 728
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991571480347268e-05,
+      "loss": 1.2175,
+      "step": 729
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915310645458936e-05,
+      "loss": 1.0814,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914905524888027e-05,
+      "loss": 1.2098,
+      "step": 731
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914499441799274e-05,
+      "loss": 1.2151,
+      "step": 732
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914092396232104e-05,
+      "loss": 1.1514,
+      "step": 733
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991368438822603e-05,
+      "loss": 1.1292,
+      "step": 734
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991327541782066e-05,
+      "loss": 1.2275,
+      "step": 735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99128654850557e-05,
+      "loss": 1.3208,
+      "step": 736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991245458997094e-05,
+      "loss": 1.1665,
+      "step": 737
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912042732606274e-05,
+      "loss": 1.1414,
+      "step": 738
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991162991300168e-05,
+      "loss": 1.3042,
+      "step": 739
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911216131197235e-05,
+      "loss": 1.1101,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991080138723311e-05,
+      "loss": 1.2094,
+      "step": 741
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991038568114956e-05,
+      "loss": 1.3147,
+      "step": 742
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909969012986952e-05,
+      "loss": 1.2395,
+      "step": 743
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990955138278573e-05,
+      "loss": 1.1697,
+      "step": 744
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909132790586438e-05,
+      "loss": 1.1282,
+      "step": 745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908713236429708e-05,
+      "loss": 1.2642,
+      "step": 746
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990829272035627e-05,
+      "loss": 1.2688,
+      "step": 747
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907871242406952e-05,
+      "loss": 1.1604,
+      "step": 748
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907448802622667e-05,
+      "loss": 1.1675,
+      "step": 749
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907025401044425e-05,
+      "loss": 1.168,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990660103771333e-05,
+      "loss": 1.1577,
+      "step": 751
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990617571267057e-05,
+      "loss": 1.2,
+      "step": 752
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905749425957445e-05,
+      "loss": 1.1626,
+      "step": 753
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905322177615332e-05,
+      "loss": 1.1285,
+      "step": 754
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904893967685717e-05,
+      "loss": 1.1685,
+      "step": 755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904464796210158e-05,
+      "loss": 1.2178,
+      "step": 756
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990403466323032e-05,
+      "loss": 1.1826,
+      "step": 757
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903603568787962e-05,
+      "loss": 1.2131,
+      "step": 758
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903171512924938e-05,
+      "loss": 1.1758,
+      "step": 759
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902738495683185e-05,
+      "loss": 1.126,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990230451710474e-05,
+      "loss": 1.1831,
+      "step": 761
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901869577231732e-05,
+      "loss": 1.2329,
+      "step": 762
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901433676106394e-05,
+      "loss": 1.1145,
+      "step": 763
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900996813771024e-05,
+      "loss": 1.1777,
+      "step": 764
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900558990268047e-05,
+      "loss": 1.1091,
+      "step": 765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900120205639963e-05,
+      "loss": 1.2563,
+      "step": 766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989968045992936e-05,
+      "loss": 1.2422,
+      "step": 767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899239753178936e-05,
+      "loss": 1.1311,
+      "step": 768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898798085431473e-05,
+      "loss": 1.2095,
+      "step": 769
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898355456729844e-05,
+      "loss": 1.174,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989791186711702e-05,
+      "loss": 1.1257,
+      "step": 771
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897467316636064e-05,
+      "loss": 1.1726,
+      "step": 772
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897021805330134e-05,
+      "loss": 1.0715,
+      "step": 773
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896575333242477e-05,
+      "loss": 1.2903,
+      "step": 774
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896127900416434e-05,
+      "loss": 1.1545,
+      "step": 775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989567950689544e-05,
+      "loss": 1.2456,
+      "step": 776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895230152723033e-05,
+      "loss": 1.1614,
+      "step": 777
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894779837942827e-05,
+      "loss": 1.1527,
+      "step": 778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989432856259854e-05,
+      "loss": 1.1611,
+      "step": 779
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893876326733974e-05,
+      "loss": 1.199,
+      "step": 780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893423130393043e-05,
+      "loss": 1.1775,
+      "step": 781
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892968973619734e-05,
+      "loss": 1.1335,
+      "step": 782
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989251385645814e-05,
+      "loss": 1.2456,
+      "step": 783
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989205777895244e-05,
+      "loss": 1.228,
+      "step": 784
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891600741146915e-05,
+      "loss": 1.2322,
+      "step": 785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891142743085923e-05,
+      "loss": 1.1628,
+      "step": 786
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890683784813935e-05,
+      "loss": 1.1833,
+      "step": 787
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98902238663755e-05,
+      "loss": 1.1748,
+      "step": 788
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889762987815264e-05,
+      "loss": 1.1128,
+      "step": 789
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889301149177974e-05,
+      "loss": 1.1204,
+      "step": 790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888838350508463e-05,
+      "loss": 1.0784,
+      "step": 791
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888374591851658e-05,
+      "loss": 1.2329,
+      "step": 792
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988790987325258e-05,
+      "loss": 1.1323,
+      "step": 793
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988744419475634e-05,
+      "loss": 1.1917,
+      "step": 794
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886977556408145e-05,
+      "loss": 1.1018,
+      "step": 795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98865099582533e-05,
+      "loss": 1.2493,
+      "step": 796
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886041400337198e-05,
+      "loss": 1.1404,
+      "step": 797
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885571882705322e-05,
+      "loss": 1.2705,
+      "step": 798
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988510140540325e-05,
+      "loss": 1.2366,
+      "step": 799
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988462996847666e-05,
+      "loss": 1.1385,
+      "step": 800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884157571971318e-05,
+      "loss": 1.1609,
+      "step": 801
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988368421593308e-05,
+      "loss": 1.2827,
+      "step": 802
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98832099004079e-05,
+      "loss": 1.2588,
+      "step": 803
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988273462544182e-05,
+      "loss": 1.2666,
+      "step": 804
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882258391080986e-05,
+      "loss": 1.1572,
+      "step": 805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881781197371624e-05,
+      "loss": 1.1384,
+      "step": 806
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881303044360065e-05,
+      "loss": 1.092,
+      "step": 807
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880823932092718e-05,
+      "loss": 1.1746,
+      "step": 808
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880343860616104e-05,
+      "loss": 1.157,
+      "step": 809
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879862829976816e-05,
+      "loss": 1.1058,
+      "step": 810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879380840221562e-05,
+      "loss": 1.082,
+      "step": 811
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987889789139713e-05,
+      "loss": 1.178,
+      "step": 812
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878413983550397e-05,
+      "loss": 1.1985,
+      "step": 813
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877929116728344e-05,
+      "loss": 1.167,
+      "step": 814
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987744329097805e-05,
+      "loss": 1.1985,
+      "step": 815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987695650634666e-05,
+      "loss": 1.1055,
+      "step": 816
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876468762881445e-05,
+      "loss": 1.1951,
+      "step": 817
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875980060629744e-05,
+      "loss": 1.188,
+      "step": 818
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875490399639007e-05,
+      "loss": 1.1257,
+      "step": 819
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874999779956767e-05,
+      "loss": 1.1531,
+      "step": 820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987450820163065e-05,
+      "loss": 1.1138,
+      "step": 821
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874015664708378e-05,
+      "loss": 1.1763,
+      "step": 822
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873522169237768e-05,
+      "loss": 1.2222,
+      "step": 823
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873027715266725e-05,
+      "loss": 1.1184,
+      "step": 824
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987253230284325e-05,
+      "loss": 1.125,
+      "step": 825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872035932015435e-05,
+      "loss": 1.2205,
+      "step": 826
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987153860283147e-05,
+      "loss": 1.1425,
+      "step": 827
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871040315339632e-05,
+      "loss": 1.2234,
+      "step": 828
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870541069588294e-05,
+      "loss": 1.1716,
+      "step": 829
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870040865625924e-05,
+      "loss": 1.0603,
+      "step": 830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869539703501083e-05,
+      "loss": 1.1154,
+      "step": 831
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869037583262415e-05,
+      "loss": 1.073,
+      "step": 832
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868534504958665e-05,
+      "loss": 1.1838,
+      "step": 833
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868030468638677e-05,
+      "loss": 1.2217,
+      "step": 834
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986752547435138e-05,
+      "loss": 1.1392,
+      "step": 835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98670195221458e-05,
+      "loss": 0.9056,
+      "step": 836
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866512612071046e-05,
+      "loss": 1.1194,
+      "step": 837
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866004744176334e-05,
+      "loss": 1.2097,
+      "step": 838
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865495918510965e-05,
+      "loss": 1.1851,
+      "step": 839
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864986135124336e-05,
+      "loss": 1.1409,
+      "step": 840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864475394065932e-05,
+      "loss": 1.1766,
+      "step": 841
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986396369538534e-05,
+      "loss": 1.2202,
+      "step": 842
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863451039132227e-05,
+      "loss": 1.2302,
+      "step": 843
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986293742535637e-05,
+      "loss": 1.126,
+      "step": 844
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862422854107626e-05,
+      "loss": 1.1018,
+      "step": 845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861907325435946e-05,
+      "loss": 1.2217,
+      "step": 846
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861390839391374e-05,
+      "loss": 1.1887,
+      "step": 847
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986087339602406e-05,
+      "loss": 1.0132,
+      "step": 848
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986035499538422e-05,
+      "loss": 1.0957,
+      "step": 849
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98598356375222e-05,
+      "loss": 1.1128,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859315322488397e-05,
+      "loss": 1.0925,
+      "step": 851
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858794050333338e-05,
+      "loss": 1.1045,
+      "step": 852
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858271821107617e-05,
+      "loss": 1.1628,
+      "step": 853
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857748634861937e-05,
+      "loss": 1.2898,
+      "step": 854
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857224491647087e-05,
+      "loss": 1.2114,
+      "step": 855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856699391513946e-05,
+      "loss": 1.0853,
+      "step": 856
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856173334513495e-05,
+      "loss": 1.1338,
+      "step": 857
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855646320696797e-05,
+      "loss": 1.074,
+      "step": 858
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855118350115012e-05,
+      "loss": 1.0165,
+      "step": 859
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854589422819404e-05,
+      "loss": 1.1855,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854059538861312e-05,
+      "loss": 1.0957,
+      "step": 861
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985352869829218e-05,
+      "loss": 1.1892,
+      "step": 862
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852996901163533e-05,
+      "loss": 1.1685,
+      "step": 863
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985246414752701e-05,
+      "loss": 1.042,
+      "step": 864
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985193043743432e-05,
+      "loss": 1.1484,
+      "step": 865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851395770937273e-05,
+      "loss": 1.2539,
+      "step": 866
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985086014808778e-05,
+      "loss": 1.199,
+      "step": 867
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850323568937836e-05,
+      "loss": 1.248,
+      "step": 868
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849786033539533e-05,
+      "loss": 1.0972,
+      "step": 869
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849247541945044e-05,
+      "loss": 1.0947,
+      "step": 870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848708094206656e-05,
+      "loss": 1.1283,
+      "step": 871
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848167690376732e-05,
+      "loss": 1.2175,
+      "step": 872
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847626330507734e-05,
+      "loss": 1.187,
+      "step": 873
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847084014652214e-05,
+      "loss": 1.1228,
+      "step": 874
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846540742862826e-05,
+      "loss": 1.1565,
+      "step": 875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98459965151923e-05,
+      "loss": 1.2668,
+      "step": 876
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845451331693478e-05,
+      "loss": 1.1968,
+      "step": 877
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844905192419276e-05,
+      "loss": 1.1216,
+      "step": 878
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984435809742272e-05,
+      "loss": 1.0542,
+      "step": 879
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843810046756915e-05,
+      "loss": 1.2437,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984326104047507e-05,
+      "loss": 1.1409,
+      "step": 881
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842711078630477e-05,
+      "loss": 1.1128,
+      "step": 882
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842160161276522e-05,
+      "loss": 1.1399,
+      "step": 883
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841608288466698e-05,
+      "loss": 1.1423,
+      "step": 884
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841055460254572e-05,
+      "loss": 1.1729,
+      "step": 885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840501676693805e-05,
+      "loss": 1.1462,
+      "step": 886
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983994693783817e-05,
+      "loss": 1.1133,
+      "step": 887
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839391243741514e-05,
+      "loss": 1.1499,
+      "step": 888
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838834594457785e-05,
+      "loss": 1.1001,
+      "step": 889
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838276990041017e-05,
+      "loss": 1.1687,
+      "step": 890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837718430545343e-05,
+      "loss": 1.1506,
+      "step": 891
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983715891602499e-05,
+      "loss": 1.2175,
+      "step": 892
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983659844653427e-05,
+      "loss": 1.3157,
+      "step": 893
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836037022127593e-05,
+      "loss": 1.0994,
+      "step": 894
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983547464285946e-05,
+      "loss": 1.1548,
+      "step": 895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834911308784467e-05,
+      "loss": 1.2002,
+      "step": 896
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834347019957302e-05,
+      "loss": 1.1963,
+      "step": 897
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833781776432745e-05,
+      "loss": 1.1697,
+      "step": 898
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983321557826567e-05,
+      "loss": 1.1438,
+      "step": 899
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832648425511038e-05,
+      "loss": 1.1482,
+      "step": 900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983208031822391e-05,
+      "loss": 1.0991,
+      "step": 901
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831511256459433e-05,
+      "loss": 1.1033,
+      "step": 902
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830941240272858e-05,
+      "loss": 1.0358,
+      "step": 903
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830370269719516e-05,
+      "loss": 1.1243,
+      "step": 904
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829798344854833e-05,
+      "loss": 1.2461,
+      "step": 905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829225465734335e-05,
+      "loss": 1.2507,
+      "step": 906
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982865163241363e-05,
+      "loss": 1.2726,
+      "step": 907
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828076844948435e-05,
+      "loss": 1.2207,
+      "step": 908
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827501103394536e-05,
+      "loss": 1.104,
+      "step": 909
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826924407807838e-05,
+      "loss": 1.1584,
+      "step": 910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826346758244314e-05,
+      "loss": 1.1511,
+      "step": 911
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825768154760046e-05,
+      "loss": 1.135,
+      "step": 912
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825188597411204e-05,
+      "loss": 1.3721,
+      "step": 913
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824608086254048e-05,
+      "loss": 1.1946,
+      "step": 914
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824026621344935e-05,
+      "loss": 1.124,
+      "step": 915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823444202740315e-05,
+      "loss": 1.0701,
+      "step": 916
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982286083049672e-05,
+      "loss": 1.26,
+      "step": 917
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982227650467079e-05,
+      "loss": 1.0924,
+      "step": 918
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821691225319247e-05,
+      "loss": 1.1543,
+      "step": 919
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821104992498906e-05,
+      "loss": 1.1052,
+      "step": 920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820517806266682e-05,
+      "loss": 1.2128,
+      "step": 921
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819929666679578e-05,
+      "loss": 1.1902,
+      "step": 922
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819340573794682e-05,
+      "loss": 1.2159,
+      "step": 923
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818750527669194e-05,
+      "loss": 1.2119,
+      "step": 924
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818159528360383e-05,
+      "loss": 1.1993,
+      "step": 925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817567575925627e-05,
+      "loss": 1.201,
+      "step": 926
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816974670422392e-05,
+      "loss": 1.1802,
+      "step": 927
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816380811908233e-05,
+      "loss": 1.1948,
+      "step": 928
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815786000440802e-05,
+      "loss": 0.9954,
+      "step": 929
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815190236077846e-05,
+      "loss": 1.1438,
+      "step": 930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814593518877197e-05,
+      "loss": 1.0094,
+      "step": 931
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981399584889678e-05,
+      "loss": 1.2314,
+      "step": 932
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981339722619462e-05,
+      "loss": 1.1353,
+      "step": 933
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812797650828827e-05,
+      "loss": 1.1379,
+      "step": 934
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981219712285761e-05,
+      "loss": 1.0356,
+      "step": 935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981159564233926e-05,
+      "loss": 1.0862,
+      "step": 936
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981099320933218e-05,
+      "loss": 1.1484,
+      "step": 937
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810389823894846e-05,
+      "loss": 1.0737,
+      "step": 938
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809785486085825e-05,
+      "loss": 1.1824,
+      "step": 939
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98091801959638e-05,
+      "loss": 1.1633,
+      "step": 940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980857395358752e-05,
+      "loss": 1.1824,
+      "step": 941
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980796675901584e-05,
+      "loss": 1.2034,
+      "step": 942
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980735861230771e-05,
+      "loss": 1.2058,
+      "step": 943
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806749513522163e-05,
+      "loss": 1.0354,
+      "step": 944
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980613946271833e-05,
+      "loss": 1.1055,
+      "step": 945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805528459955433e-05,
+      "loss": 1.2327,
+      "step": 946
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980491650529279e-05,
+      "loss": 1.1846,
+      "step": 947
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804303598789804e-05,
+      "loss": 1.1838,
+      "step": 948
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980368974050598e-05,
+      "loss": 1.1099,
+      "step": 949
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98030749305009e-05,
+      "loss": 1.1228,
+      "step": 950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980245916883426e-05,
+      "loss": 1.2273,
+      "step": 951
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980184245556583e-05,
+      "loss": 1.1699,
+      "step": 952
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801224790755484e-05,
+      "loss": 1.2217,
+      "step": 953
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800606174463177e-05,
+      "loss": 1.1875,
+      "step": 954
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979998660674897e-05,
+      "loss": 1.0984,
+      "step": 955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799366087673e-05,
+      "loss": 0.9904,
+      "step": 956
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979874461729552e-05,
+      "loss": 1.1855,
+      "step": 957
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798122195676845e-05,
+      "loss": 1.1307,
+      "step": 958
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797498822877407e-05,
+      "loss": 1.0989,
+      "step": 959
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979687449895772e-05,
+      "loss": 1.1575,
+      "step": 960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796249223978397e-05,
+      "loss": 1.2004,
+      "step": 961
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795622998000132e-05,
+      "loss": 1.2197,
+      "step": 962
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979499582108372e-05,
+      "loss": 1.2329,
+      "step": 963
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794367693290045e-05,
+      "loss": 1.3401,
+      "step": 964
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793738614680084e-05,
+      "loss": 1.116,
+      "step": 965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979310858531491e-05,
+      "loss": 1.2104,
+      "step": 966
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979247760525568e-05,
+      "loss": 1.1699,
+      "step": 967
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979184567456365e-05,
+      "loss": 1.0215,
+      "step": 968
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791212793300173e-05,
+      "loss": 1.125,
+      "step": 969
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790578961526675e-05,
+      "loss": 1.1064,
+      "step": 970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789944179304697e-05,
+      "loss": 1.24,
+      "step": 971
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978930844669586e-05,
+      "loss": 1.1968,
+      "step": 972
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788671763761877e-05,
+      "loss": 1.2124,
+      "step": 973
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978803413056456e-05,
+      "loss": 1.1084,
+      "step": 974
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787395547165804e-05,
+      "loss": 1.1265,
+      "step": 975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786756013627606e-05,
+      "loss": 1.0875,
+      "step": 976
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978611553001205e-05,
+      "loss": 1.2063,
+      "step": 977
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978547409638131e-05,
+      "loss": 1.175,
+      "step": 978
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978483171279765e-05,
+      "loss": 1.0568,
+      "step": 979
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784188379323444e-05,
+      "loss": 1.2085,
+      "step": 980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978354409602114e-05,
+      "loss": 1.0833,
+      "step": 981
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978289886295328e-05,
+      "loss": 1.1487,
+      "step": 982
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782252680182504e-05,
+      "loss": 1.1177,
+      "step": 983
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781605547771546e-05,
+      "loss": 1.0869,
+      "step": 984
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780957465783223e-05,
+      "loss": 1.0334,
+      "step": 985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978030843428045e-05,
+      "loss": 1.2369,
+      "step": 986
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779658453326234e-05,
+      "loss": 1.1895,
+      "step": 987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977900752298368e-05,
+      "loss": 1.1791,
+      "step": 988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778355643315967e-05,
+      "loss": 1.0928,
+      "step": 989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777702814386387e-05,
+      "loss": 1.1797,
+      "step": 990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777049036258314e-05,
+      "loss": 1.1765,
+      "step": 991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776394308995213e-05,
+      "loss": 1.123,
+      "step": 992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775738632660647e-05,
+      "loss": 1.2341,
+      "step": 993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775082007318264e-05,
+      "loss": 1.1094,
+      "step": 994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774424433031805e-05,
+      "loss": 1.217,
+      "step": 995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773765909865114e-05,
+      "loss": 1.1738,
+      "step": 996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773106437882114e-05,
+      "loss": 1.1426,
+      "step": 997
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772446017146827e-05,
+      "loss": 1.1145,
+      "step": 998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771784647723363e-05,
+      "loss": 1.1174,
+      "step": 999
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771122329675928e-05,
+      "loss": 1.1873,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977045906306882e-05,
+      "loss": 1.2009,
+      "step": 1001
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769794847966427e-05,
+      "loss": 1.1145,
+      "step": 1002
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769129684433226e-05,
+      "loss": 1.0862,
+      "step": 1003
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768463572533792e-05,
+      "loss": 1.134,
+      "step": 1004
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976779651233279e-05,
+      "loss": 1.2864,
+      "step": 1005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767128503894976e-05,
+      "loss": 1.1628,
+      "step": 1006
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766459547285198e-05,
+      "loss": 1.124,
+      "step": 1007
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97657896425684e-05,
+      "loss": 1.114,
+      "step": 1008
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765118789809615e-05,
+      "loss": 1.0854,
+      "step": 1009
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976444698907396e-05,
+      "loss": 1.1719,
+      "step": 1010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763774240426664e-05,
+      "loss": 1.1443,
+      "step": 1011
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763100543933026e-05,
+      "loss": 1.134,
+      "step": 1012
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976242589965845e-05,
+      "loss": 1.2415,
+      "step": 1013
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761750307668436e-05,
+      "loss": 1.1106,
+      "step": 1014
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976107376802856e-05,
+      "loss": 1.0411,
+      "step": 1015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97603962808045e-05,
+      "loss": 1.2781,
+      "step": 1016
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975971784606203e-05,
+      "loss": 1.1792,
+      "step": 1017
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759038463867008e-05,
+      "loss": 1.0916,
+      "step": 1018
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758358134285384e-05,
+      "loss": 1.1411,
+      "step": 1019
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975767685738321e-05,
+      "loss": 1.1132,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975699463322662e-05,
+      "loss": 1.2505,
+      "step": 1021
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756311461881838e-05,
+      "loss": 1.0483,
+      "step": 1022
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975562734341519e-05,
+      "loss": 1.1802,
+      "step": 1023
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754942277893083e-05,
+      "loss": 1.1833,
+      "step": 1024
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754256265382034e-05,
+      "loss": 1.1216,
+      "step": 1025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753569305948624e-05,
+      "loss": 1.2012,
+      "step": 1026
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752881399659553e-05,
+      "loss": 1.074,
+      "step": 1027
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752192546581598e-05,
+      "loss": 1.1315,
+      "step": 1028
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975150274678163e-05,
+      "loss": 1.2428,
+      "step": 1029
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750812000326616e-05,
+      "loss": 1.2368,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750120307283608e-05,
+      "loss": 1.1021,
+      "step": 1031
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749427667719757e-05,
+      "loss": 1.2058,
+      "step": 1032
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748734081702304e-05,
+      "loss": 1.0847,
+      "step": 1033
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974803954929858e-05,
+      "loss": 1.1792,
+      "step": 1034
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747344070576005e-05,
+      "loss": 1.2659,
+      "step": 1035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746647645602102e-05,
+      "loss": 1.0952,
+      "step": 1036
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745950274444472e-05,
+      "loss": 1.0776,
+      "step": 1037
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745251957170817e-05,
+      "loss": 1.2585,
+      "step": 1038
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744552693848925e-05,
+      "loss": 1.0457,
+      "step": 1039
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974385248454669e-05,
+      "loss": 1.1331,
+      "step": 1040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743151329332072e-05,
+      "loss": 1.2209,
+      "step": 1041
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742449228273143e-05,
+      "loss": 1.1617,
+      "step": 1042
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741746181438063e-05,
+      "loss": 1.1606,
+      "step": 1043
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741042188895084e-05,
+      "loss": 1.1948,
+      "step": 1044
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740337250712546e-05,
+      "loss": 1.125,
+      "step": 1045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973963136695888e-05,
+      "loss": 1.0845,
+      "step": 1046
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738924537702618e-05,
+      "loss": 1.1331,
+      "step": 1047
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738216763012372e-05,
+      "loss": 1.1919,
+      "step": 1048
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737508042956853e-05,
+      "loss": 1.1379,
+      "step": 1049
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736798377604864e-05,
+      "loss": 1.053,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736087767025295e-05,
+      "loss": 1.1536,
+      "step": 1051
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735376211287132e-05,
+      "loss": 1.1497,
+      "step": 1052
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734663710459454e-05,
+      "loss": 1.0791,
+      "step": 1053
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973395026461142e-05,
+      "loss": 1.1755,
+      "step": 1054
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733235873812302e-05,
+      "loss": 1.1597,
+      "step": 1055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732520538131442e-05,
+      "loss": 1.1256,
+      "step": 1056
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731804257638285e-05,
+      "loss": 1.1169,
+      "step": 1057
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973108703240237e-05,
+      "loss": 1.0361,
+      "step": 1058
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973036886249332e-05,
+      "loss": 1.1479,
+      "step": 1059
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729649747980856e-05,
+      "loss": 1.2842,
+      "step": 1060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972892968893479e-05,
+      "loss": 1.1223,
+      "step": 1061
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972820868542501e-05,
+      "loss": 0.9668,
+      "step": 1062
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972748673752153e-05,
+      "loss": 1.1797,
+      "step": 1063
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726763845294423e-05,
+      "loss": 1.0664,
+      "step": 1064
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726040008813866e-05,
+      "loss": 1.1042,
+      "step": 1065
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725315228150134e-05,
+      "loss": 1.1589,
+      "step": 1066
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724589503373575e-05,
+      "loss": 1.1619,
+      "step": 1067
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723862834554656e-05,
+      "loss": 1.2385,
+      "step": 1068
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972313522176391e-05,
+      "loss": 1.2451,
+      "step": 1069
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722406665071973e-05,
+      "loss": 1.209,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972167716454958e-05,
+      "loss": 1.2454,
+      "step": 1071
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720946720267537e-05,
+      "loss": 1.137,
+      "step": 1072
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972021533229676e-05,
+      "loss": 1.1206,
+      "step": 1073
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719483000708255e-05,
+      "loss": 1.1377,
+      "step": 1074
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971874972557311e-05,
+      "loss": 1.2126,
+      "step": 1075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718015506962508e-05,
+      "loss": 1.1675,
+      "step": 1076
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971728034494773e-05,
+      "loss": 1.0553,
+      "step": 1077
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716544239600142e-05,
+      "loss": 1.137,
+      "step": 1078
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715807190991202e-05,
+      "loss": 1.2126,
+      "step": 1079
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715069199192465e-05,
+      "loss": 1.1191,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971433026427557e-05,
+      "loss": 1.2031,
+      "step": 1081
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713590386312252e-05,
+      "loss": 1.0563,
+      "step": 1082
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712849565374336e-05,
+      "loss": 1.1145,
+      "step": 1083
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712107801533745e-05,
+      "loss": 1.2815,
+      "step": 1084
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971136509486248e-05,
+      "loss": 1.2148,
+      "step": 1085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710621445432646e-05,
+      "loss": 1.1094,
+      "step": 1086
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709876853316434e-05,
+      "loss": 1.2483,
+      "step": 1087
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709131318586126e-05,
+      "loss": 1.1892,
+      "step": 1088
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708384841314098e-05,
+      "loss": 1.1459,
+      "step": 1089
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970763742157282e-05,
+      "loss": 1.1246,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9706889059434844e-05,
+      "loss": 1.1697,
+      "step": 1091
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970613975497282e-05,
+      "loss": 1.0834,
+      "step": 1092
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970538950825949e-05,
+      "loss": 1.0942,
+      "step": 1093
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970463831936769e-05,
+      "loss": 1.2434,
+      "step": 1094
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970388618837034e-05,
+      "loss": 1.1108,
+      "step": 1095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703133115340458e-05,
+      "loss": 1.2285,
+      "step": 1096
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702379100351146e-05,
+      "loss": 1.156,
+      "step": 1097
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701624143475608e-05,
+      "loss": 1.1088,
+      "step": 1098
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700868244787127e-05,
+      "loss": 1.1226,
+      "step": 1099
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970011140435909e-05,
+      "loss": 1.2234,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699353622264967e-05,
+      "loss": 1.1765,
+      "step": 1101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969859489857832e-05,
+      "loss": 1.1704,
+      "step": 1102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969783523337281e-05,
+      "loss": 1.0847,
+      "step": 1103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969707462672218e-05,
+      "loss": 1.0555,
+      "step": 1104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696313078700262e-05,
+      "loss": 1.1556,
+      "step": 1105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695550589380993e-05,
+      "loss": 1.2002,
+      "step": 1106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694787158838394e-05,
+      "loss": 1.0222,
+      "step": 1107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694022787146578e-05,
+      "loss": 1.1475,
+      "step": 1108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969325747437974e-05,
+      "loss": 1.1248,
+      "step": 1109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692491220612188e-05,
+      "loss": 1.2256,
+      "step": 1110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691724025918296e-05,
+      "loss": 1.1277,
+      "step": 1111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969095589037255e-05,
+      "loss": 1.1143,
+      "step": 1112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690186814049513e-05,
+      "loss": 1.1399,
+      "step": 1113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968941679702385e-05,
+      "loss": 1.1855,
+      "step": 1114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688645839370307e-05,
+      "loss": 1.2036,
+      "step": 1115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687873941163737e-05,
+      "loss": 1.1899,
+      "step": 1116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687101102479063e-05,
+      "loss": 1.1985,
+      "step": 1117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686327323391317e-05,
+      "loss": 1.072,
+      "step": 1118
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685552603975612e-05,
+      "loss": 1.1094,
+      "step": 1119
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684776944307162e-05,
+      "loss": 1.2788,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968400034446126e-05,
+      "loss": 1.1755,
+      "step": 1121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96832228045133e-05,
+      "loss": 0.9209,
+      "step": 1122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682444324538766e-05,
+      "loss": 1.1899,
+      "step": 1123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681664904613224e-05,
+      "loss": 1.1172,
+      "step": 1124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680884544812347e-05,
+      "loss": 1.1233,
+      "step": 1125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680103245211884e-05,
+      "loss": 1.189,
+      "step": 1126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679321005887683e-05,
+      "loss": 1.0896,
+      "step": 1127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678537826915687e-05,
+      "loss": 1.1138,
+      "step": 1128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967775370837192e-05,
+      "loss": 1.2944,
+      "step": 1129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676968650332504e-05,
+      "loss": 1.1348,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676182652873653e-05,
+      "loss": 1.1956,
+      "step": 1131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675395716071666e-05,
+      "loss": 1.1135,
+      "step": 1132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674607840002942e-05,
+      "loss": 1.0525,
+      "step": 1133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967381902474396e-05,
+      "loss": 1.1561,
+      "step": 1134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673029270371303e-05,
+      "loss": 1.0669,
+      "step": 1135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672238576961634e-05,
+      "loss": 1.1584,
+      "step": 1136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671446944591714e-05,
+      "loss": 1.1584,
+      "step": 1137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670654373338393e-05,
+      "loss": 1.3101,
+      "step": 1138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669860863278612e-05,
+      "loss": 1.1213,
+      "step": 1139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96690664144894e-05,
+      "loss": 1.0251,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668271027047886e-05,
+      "loss": 1.2859,
+      "step": 1141
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667474701031284e-05,
+      "loss": 1.1647,
+      "step": 1142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666677436516894e-05,
+      "loss": 1.2222,
+      "step": 1143
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966587923358212e-05,
+      "loss": 1.1475,
+      "step": 1144
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665080092304444e-05,
+      "loss": 1.0754,
+      "step": 1145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664280012761445e-05,
+      "loss": 1.1118,
+      "step": 1146
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663478995030795e-05,
+      "loss": 1.0835,
+      "step": 1147
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662677039190257e-05,
+      "loss": 1.1895,
+      "step": 1148
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966187414531768e-05,
+      "loss": 1.2009,
+      "step": 1149
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661070313491012e-05,
+      "loss": 1.2114,
+      "step": 1150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966026554378828e-05,
+      "loss": 1.1604,
+      "step": 1151
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659459836287614e-05,
+      "loss": 1.1812,
+      "step": 1152
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965865319106723e-05,
+      "loss": 1.2002,
+      "step": 1153
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657845608205435e-05,
+      "loss": 1.0781,
+      "step": 1154
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657037087780626e-05,
+      "loss": 1.095,
+      "step": 1155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656227629871295e-05,
+      "loss": 1.0831,
+      "step": 1156
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655417234556018e-05,
+      "loss": 1.1331,
+      "step": 1157
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654605901913472e-05,
+      "loss": 1.3645,
+      "step": 1158
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653793632022418e-05,
+      "loss": 1.1025,
+      "step": 1159
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652980424961707e-05,
+      "loss": 1.2288,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652166280810285e-05,
+      "loss": 1.1465,
+      "step": 1161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965135119964719e-05,
+      "loss": 1.1768,
+      "step": 1162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965053518155154e-05,
+      "loss": 1.0815,
+      "step": 1163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649718226602562e-05,
+      "loss": 1.2141,
+      "step": 1164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964890033487956e-05,
+      "loss": 1.0889,
+      "step": 1165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964808150646193e-05,
+      "loss": 1.027,
+      "step": 1166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964726174142917e-05,
+      "loss": 1.1975,
+      "step": 1167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646441039860853e-05,
+      "loss": 1.1938,
+      "step": 1168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645619401836654e-05,
+      "loss": 1.1423,
+      "step": 1169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964479682743634e-05,
+      "loss": 1.1511,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643973316739758e-05,
+      "loss": 1.1533,
+      "step": 1171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643148869826854e-05,
+      "loss": 1.1812,
+      "step": 1172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642323486777668e-05,
+      "loss": 1.0732,
+      "step": 1173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641497167672323e-05,
+      "loss": 1.0293,
+      "step": 1174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640669912591036e-05,
+      "loss": 1.1104,
+      "step": 1175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963984172161412e-05,
+      "loss": 1.0891,
+      "step": 1176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639012594821968e-05,
+      "loss": 1.1475,
+      "step": 1177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638182532295066e-05,
+      "loss": 1.1094,
+      "step": 1178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637351534114008e-05,
+      "loss": 1.1737,
+      "step": 1179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636519600359456e-05,
+      "loss": 1.1177,
+      "step": 1180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635686731112173e-05,
+      "loss": 1.1528,
+      "step": 1181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634852926453016e-05,
+      "loss": 1.2268,
+      "step": 1182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634018186462924e-05,
+      "loss": 1.1443,
+      "step": 1183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963318251122294e-05,
+      "loss": 1.136,
+      "step": 1184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632345900814178e-05,
+      "loss": 1.1028,
+      "step": 1185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631508355317862e-05,
+      "loss": 1.0996,
+      "step": 1186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96306698748153e-05,
+      "loss": 1.0813,
+      "step": 1187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629830459387884e-05,
+      "loss": 1.2261,
+      "step": 1188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628990109117106e-05,
+      "loss": 1.3062,
+      "step": 1189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962814882408455e-05,
+      "loss": 1.0339,
+      "step": 1190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627306604371873e-05,
+      "loss": 1.1284,
+      "step": 1191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962646345006085e-05,
+      "loss": 1.1589,
+      "step": 1192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625619361233327e-05,
+      "loss": 1.1384,
+      "step": 1193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962477433797125e-05,
+      "loss": 1.2421,
+      "step": 1194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9623928380356642e-05,
+      "loss": 1.0532,
+      "step": 1195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623081488471636e-05,
+      "loss": 1.2732,
+      "step": 1196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622233662398444e-05,
+      "loss": 1.0594,
+      "step": 1197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621384902219368e-05,
+      "loss": 1.0979,
+      "step": 1198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962053520801681e-05,
+      "loss": 1.0464,
+      "step": 1199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619684579873254e-05,
+      "loss": 1.1082,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618833017871273e-05,
+      "loss": 0.9886,
+      "step": 1201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617980522093544e-05,
+      "loss": 1.2454,
+      "step": 1202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961712709262281e-05,
+      "loss": 1.1809,
+      "step": 1203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616272729541936e-05,
+      "loss": 1.1951,
+      "step": 1204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615417432933855e-05,
+      "loss": 1.155,
+      "step": 1205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614561202881596e-05,
+      "loss": 1.1436,
+      "step": 1206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613704039468285e-05,
+      "loss": 1.1843,
+      "step": 1207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961284594277713e-05,
+      "loss": 1.093,
+      "step": 1208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961198691289143e-05,
+      "loss": 1.1129,
+      "step": 1209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611126949894586e-05,
+      "loss": 1.077,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610266053870074e-05,
+      "loss": 1.0573,
+      "step": 1211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960940422490147e-05,
+      "loss": 1.1218,
+      "step": 1212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960854146307244e-05,
+      "loss": 1.198,
+      "step": 1213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960767776846674e-05,
+      "loss": 1.051,
+      "step": 1214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606813141168214e-05,
+      "loss": 1.1136,
+      "step": 1215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605947581260795e-05,
+      "loss": 1.1267,
+      "step": 1216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605081088828512e-05,
+      "loss": 1.156,
+      "step": 1217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604213663955483e-05,
+      "loss": 1.1926,
+      "step": 1218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603345306725923e-05,
+      "loss": 1.1104,
+      "step": 1219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602476017224115e-05,
+      "loss": 1.147,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960160579553446e-05,
+      "loss": 1.0874,
+      "step": 1221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960073464174143e-05,
+      "loss": 1.1262,
+      "step": 1222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95998625559296e-05,
+      "loss": 1.1135,
+      "step": 1223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959898953818363e-05,
+      "loss": 1.1223,
+      "step": 1224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598115588588266e-05,
+      "loss": 1.1814,
+      "step": 1225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597240707228352e-05,
+      "loss": 1.135,
+      "step": 1226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596364894188822e-05,
+      "loss": 1.0349,
+      "step": 1227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595488149554695e-05,
+      "loss": 1.2461,
+      "step": 1228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959461047341108e-05,
+      "loss": 1.0515,
+      "step": 1229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593731865843187e-05,
+      "loss": 1.1262,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592852326936305e-05,
+      "loss": 1.2351,
+      "step": 1231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959197185677582e-05,
+      "loss": 1.1313,
+      "step": 1232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959109045544721e-05,
+      "loss": 1.0555,
+      "step": 1233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590208123036024e-05,
+      "loss": 1.0781,
+      "step": 1234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589324859627935e-05,
+      "loss": 1.1187,
+      "step": 1235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588440665308682e-05,
+      "loss": 1.1072,
+      "step": 1236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587555540164097e-05,
+      "loss": 1.1333,
+      "step": 1237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958666948428011e-05,
+      "loss": 1.182,
+      "step": 1238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585782497742735e-05,
+      "loss": 1.0576,
+      "step": 1239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958489458063808e-05,
+      "loss": 1.2512,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584005733052346e-05,
+      "loss": 1.0615,
+      "step": 1241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583115955071813e-05,
+      "loss": 1.1024,
+      "step": 1242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958222524678286e-05,
+      "loss": 1.2135,
+      "step": 1243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581333608271963e-05,
+      "loss": 1.1089,
+      "step": 1244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958044103962567e-05,
+      "loss": 0.9866,
+      "step": 1245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579547540930636e-05,
+      "loss": 1.198,
+      "step": 1246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95786531122736e-05,
+      "loss": 1.1093,
+      "step": 1247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577757753741385e-05,
+      "loss": 1.1796,
+      "step": 1248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957686146542092e-05,
+      "loss": 1.1919,
+      "step": 1249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957596424739921e-05,
+      "loss": 1.0869,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957506609976335e-05,
+      "loss": 1.0718,
+      "step": 1251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574167022600536e-05,
+      "loss": 1.0569,
+      "step": 1252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957326701599805e-05,
+      "loss": 1.1257,
+      "step": 1253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957236608004326e-05,
+      "loss": 1.2085,
+      "step": 1254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571464214823627e-05,
+      "loss": 1.2015,
+      "step": 1255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570561420426703e-05,
+      "loss": 1.155,
+      "step": 1256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569657696940124e-05,
+      "loss": 1.2065,
+      "step": 1257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956875304445163e-05,
+      "loss": 1.1172,
+      "step": 1258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956784746304904e-05,
+      "loss": 1.1321,
+      "step": 1259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956694095282026e-05,
+      "loss": 1.0275,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566033513853302e-05,
+      "loss": 1.2437,
+      "step": 1261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565125146236252e-05,
+      "loss": 1.1407,
+      "step": 1262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564215850057294e-05,
+      "loss": 1.1895,
+      "step": 1263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95633056254047e-05,
+      "loss": 1.1416,
+      "step": 1264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956239447236683e-05,
+      "loss": 1.1693,
+      "step": 1265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956148239103214e-05,
+      "loss": 1.0326,
+      "step": 1266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560569381489178e-05,
+      "loss": 1.157,
+      "step": 1267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559655443826564e-05,
+      "loss": 1.1924,
+      "step": 1268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558740578133033e-05,
+      "loss": 1.0457,
+      "step": 1269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557824784497392e-05,
+      "loss": 1.1224,
+      "step": 1270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556908063008546e-05,
+      "loss": 1.1526,
+      "step": 1271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955599041375549e-05,
+      "loss": 1.1616,
+      "step": 1272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555071836827304e-05,
+      "loss": 1.175,
+      "step": 1273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554152332313165e-05,
+      "loss": 1.1711,
+      "step": 1274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553231900302334e-05,
+      "loss": 1.1531,
+      "step": 1275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552310540884164e-05,
+      "loss": 1.1814,
+      "step": 1276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551388254148103e-05,
+      "loss": 1.1111,
+      "step": 1277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955046504018368e-05,
+      "loss": 1.1021,
+      "step": 1278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549540899080522e-05,
+      "loss": 1.0942,
+      "step": 1279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954861583092834e-05,
+      "loss": 1.187,
+      "step": 1280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547689835816936e-05,
+      "loss": 1.147,
+      "step": 1281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954676291383621e-05,
+      "loss": 1.1052,
+      "step": 1282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545835065076136e-05,
+      "loss": 1.2161,
+      "step": 1283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544906289626802e-05,
+      "loss": 1.2275,
+      "step": 1284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954397658757836e-05,
+      "loss": 1.0181,
+      "step": 1285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954304595902106e-05,
+      "loss": 1.0107,
+      "step": 1286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954211440404526e-05,
+      "loss": 1.1011,
+      "step": 1287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954118192274138e-05,
+      "loss": 1.0908,
+      "step": 1288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954024851519995e-05,
+      "loss": 1.1243,
+      "step": 1289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539314181511584e-05,
+      "loss": 1.1401,
+      "step": 1290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953837892176698e-05,
+      "loss": 1.0918,
+      "step": 1291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537442736056935e-05,
+      "loss": 1.0382,
+      "step": 1292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536505624472332e-05,
+      "loss": 1.1685,
+      "step": 1293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953556758710414e-05,
+      "loss": 1.271,
+      "step": 1294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534628624043427e-05,
+      "loss": 1.1855,
+      "step": 1295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953368873538134e-05,
+      "loss": 1.1003,
+      "step": 1296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532747921209126e-05,
+      "loss": 1.1758,
+      "step": 1297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531806181618112e-05,
+      "loss": 1.2246,
+      "step": 1298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953086351669973e-05,
+      "loss": 1.1515,
+      "step": 1299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952991992654548e-05,
+      "loss": 1.1003,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952897541124697e-05,
+      "loss": 1.1089,
+      "step": 1301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528029970895892e-05,
+      "loss": 1.2073,
+      "step": 1302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527083605584022e-05,
+      "loss": 1.1948,
+      "step": 1303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526136315403243e-05,
+      "loss": 1.1121,
+      "step": 1304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95251881004455e-05,
+      "loss": 1.0417,
+      "step": 1305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524238960802854e-05,
+      "loss": 1.219,
+      "step": 1306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523288896567445e-05,
+      "loss": 1.0663,
+      "step": 1307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522337907831497e-05,
+      "loss": 1.1055,
+      "step": 1308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952138599468734e-05,
+      "loss": 1.2549,
+      "step": 1309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520433157227374e-05,
+      "loss": 1.0596,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519479395544103e-05,
+      "loss": 1.0728,
+      "step": 1311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518524709730115e-05,
+      "loss": 1.0535,
+      "step": 1312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517569099878093e-05,
+      "loss": 1.178,
+      "step": 1313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516612566080797e-05,
+      "loss": 1.2349,
+      "step": 1314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515655108431094e-05,
+      "loss": 1.1624,
+      "step": 1315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514696727021925e-05,
+      "loss": 1.1729,
+      "step": 1316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513737421946332e-05,
+      "loss": 1.2461,
+      "step": 1317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951277719329744e-05,
+      "loss": 1.1294,
+      "step": 1318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951181604116847e-05,
+      "loss": 1.2297,
+      "step": 1319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510853965652723e-05,
+      "loss": 1.1038,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509890966843596e-05,
+      "loss": 1.1582,
+      "step": 1321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950892704483458e-05,
+      "loss": 1.1538,
+      "step": 1322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507962199719245e-05,
+      "loss": 1.1184,
+      "step": 1323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950699643159126e-05,
+      "loss": 1.1238,
+      "step": 1324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506029740544377e-05,
+      "loss": 1.1904,
+      "step": 1325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505062126672442e-05,
+      "loss": 1.2693,
+      "step": 1326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504093590069385e-05,
+      "loss": 1.1111,
+      "step": 1327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503124130829238e-05,
+      "loss": 0.9827,
+      "step": 1328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502153749046105e-05,
+      "loss": 1.1577,
+      "step": 1329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501182444814192e-05,
+      "loss": 1.1487,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500210218227794e-05,
+      "loss": 1.092,
+      "step": 1331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499237069381288e-05,
+      "loss": 1.2085,
+      "step": 1332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498262998369148e-05,
+      "loss": 1.2532,
+      "step": 1333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497288005285934e-05,
+      "loss": 1.1982,
+      "step": 1334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496312090226296e-05,
+      "loss": 1.1823,
+      "step": 1335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495335253284974e-05,
+      "loss": 1.2156,
+      "step": 1336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494357494556798e-05,
+      "loss": 1.1475,
+      "step": 1337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493378814136683e-05,
+      "loss": 1.104,
+      "step": 1338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492399212119643e-05,
+      "loss": 1.2349,
+      "step": 1339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949141868860077e-05,
+      "loss": 1.2222,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949043724367526e-05,
+      "loss": 1.0432,
+      "step": 1341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489454877438375e-05,
+      "loss": 1.2686,
+      "step": 1342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488471589985494e-05,
+      "loss": 1.2566,
+      "step": 1343
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487487381412065e-05,
+      "loss": 1.1567,
+      "step": 1344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948650225181364e-05,
+      "loss": 1.0883,
+      "step": 1345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948551620128585e-05,
+      "loss": 1.1274,
+      "step": 1346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484529229924416e-05,
+      "loss": 1.116,
+      "step": 1347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483541337825152e-05,
+      "loss": 1.1309,
+      "step": 1348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948255252508396e-05,
+      "loss": 1.1809,
+      "step": 1349
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481562791796834e-05,
+      "loss": 1.2275,
+      "step": 1350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948057213805986e-05,
+      "loss": 1.1941,
+      "step": 1351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479580563969194e-05,
+      "loss": 1.0754,
+      "step": 1352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947858806962111e-05,
+      "loss": 1.2197,
+      "step": 1353
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477594655111953e-05,
+      "loss": 1.1714,
+      "step": 1354
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476600320538158e-05,
+      "loss": 1.0969,
+      "step": 1355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475605065996258e-05,
+      "loss": 1.1946,
+      "step": 1356
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474608891582862e-05,
+      "loss": 1.1257,
+      "step": 1357
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473611797394688e-05,
+      "loss": 1.2148,
+      "step": 1358
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472613783528522e-05,
+      "loss": 1.2297,
+      "step": 1359
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471614850081256e-05,
+      "loss": 1.1833,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947061499714986e-05,
+      "loss": 1.1545,
+      "step": 1361
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469614224831397e-05,
+      "loss": 1.2205,
+      "step": 1362
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468612533223024e-05,
+      "loss": 1.0986,
+      "step": 1363
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946760992242198e-05,
+      "loss": 1.0337,
+      "step": 1364
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466606392525595e-05,
+      "loss": 1.1167,
+      "step": 1365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465601943631297e-05,
+      "loss": 1.0071,
+      "step": 1366
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464596575836588e-05,
+      "loss": 1.0472,
+      "step": 1367
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946359028923907e-05,
+      "loss": 1.1416,
+      "step": 1368
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462583083936427e-05,
+      "loss": 1.2197,
+      "step": 1369
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946157496002644e-05,
+      "loss": 1.2375,
+      "step": 1370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460565917606976e-05,
+      "loss": 1.1237,
+      "step": 1371
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459555956775992e-05,
+      "loss": 1.1423,
+      "step": 1372
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458545077631532e-05,
+      "loss": 1.144,
+      "step": 1373
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457533280271725e-05,
+      "loss": 1.1296,
+      "step": 1374
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456520564794797e-05,
+      "loss": 0.9661,
+      "step": 1375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455506931299068e-05,
+      "loss": 1.2178,
+      "step": 1376
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454492379882926e-05,
+      "loss": 1.1279,
+      "step": 1377
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945347691064487e-05,
+      "loss": 1.1265,
+      "step": 1378
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945246052368348e-05,
+      "loss": 1.0706,
+      "step": 1379
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945144321909742e-05,
+      "loss": 1.1367,
+      "step": 1380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945042499698545e-05,
+      "loss": 1.135,
+      "step": 1381
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449405857446422e-05,
+      "loss": 1.1394,
+      "step": 1382
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448385800579262e-05,
+      "loss": 1.1709,
+      "step": 1383
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447364826483e-05,
+      "loss": 1.0713,
+      "step": 1384
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446342935256747e-05,
+      "loss": 1.1736,
+      "step": 1385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445320126999715e-05,
+      "loss": 1.2996,
+      "step": 1386
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444296401811185e-05,
+      "loss": 1.1482,
+      "step": 1387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443271759790544e-05,
+      "loss": 1.1482,
+      "step": 1388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442246201037264e-05,
+      "loss": 0.9712,
+      "step": 1389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441219725650895e-05,
+      "loss": 1.1704,
+      "step": 1390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440192333731095e-05,
+      "loss": 1.1327,
+      "step": 1391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94391640253776e-05,
+      "loss": 1.1638,
+      "step": 1392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438134800690225e-05,
+      "loss": 1.0581,
+      "step": 1393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94371046597689e-05,
+      "loss": 1.1305,
+      "step": 1394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436073602713617e-05,
+      "loss": 1.0972,
+      "step": 1395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435041629624477e-05,
+      "loss": 1.114,
+      "step": 1396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434008740601654e-05,
+      "loss": 1.1636,
+      "step": 1397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432974935745426e-05,
+      "loss": 1.1851,
+      "step": 1398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431940215156152e-05,
+      "loss": 1.1685,
+      "step": 1399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430904578934275e-05,
+      "loss": 1.0697,
+      "step": 1400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429868027180333e-05,
+      "loss": 1.1477,
+      "step": 1401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942883055999496e-05,
+      "loss": 1.0969,
+      "step": 1402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9427792177478863e-05,
+      "loss": 1.2507,
+      "step": 1403
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426752879732848e-05,
+      "loss": 1.0847,
+      "step": 1404
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942571266685781e-05,
+      "loss": 1.1458,
+      "step": 1405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424671538954726e-05,
+      "loss": 1.1765,
+      "step": 1406
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423629496124674e-05,
+      "loss": 1.1919,
+      "step": 1407
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422586538468807e-05,
+      "loss": 1.2332,
+      "step": 1408
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421542666088375e-05,
+      "loss": 1.0723,
+      "step": 1409
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420497879084715e-05,
+      "loss": 1.2124,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419452177559252e-05,
+      "loss": 1.0476,
+      "step": 1411
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94184055616135e-05,
+      "loss": 1.208,
+      "step": 1412
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941735803134907e-05,
+      "loss": 1.1389,
+      "step": 1413
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941630958686764e-05,
+      "loss": 1.123,
+      "step": 1414
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415260228271e-05,
+      "loss": 1.2014,
+      "step": 1415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414209955661015e-05,
+      "loss": 1.0688,
+      "step": 1416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941315876913965e-05,
+      "loss": 1.2134,
+      "step": 1417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412106668808944e-05,
+      "loss": 1.1399,
+      "step": 1418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941105365477104e-05,
+      "loss": 1.228,
+      "step": 1419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409999727128155e-05,
+      "loss": 1.2032,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940894488598261e-05,
+      "loss": 1.0842,
+      "step": 1421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94078891314368e-05,
+      "loss": 1.1418,
+      "step": 1422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406832463593217e-05,
+      "loss": 1.1438,
+      "step": 1423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940577488255444e-05,
+      "loss": 1.1458,
+      "step": 1424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940471638842314e-05,
+      "loss": 1.1617,
+      "step": 1425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940365698130207e-05,
+      "loss": 1.1301,
+      "step": 1426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940259666129408e-05,
+      "loss": 1.0287,
+      "step": 1427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401535428502096e-05,
+      "loss": 1.1091,
+      "step": 1428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400473283029142e-05,
+      "loss": 1.1255,
+      "step": 1429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939941022497833e-05,
+      "loss": 1.2227,
+      "step": 1430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398346254452864e-05,
+      "loss": 1.146,
+      "step": 1431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397281371556026e-05,
+      "loss": 1.124,
+      "step": 1432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396215576391196e-05,
+      "loss": 1.0551,
+      "step": 1433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395148869061836e-05,
+      "loss": 1.1763,
+      "step": 1434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394081249671502e-05,
+      "loss": 1.1417,
+      "step": 1435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393012718323833e-05,
+      "loss": 1.2439,
+      "step": 1436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391943275122567e-05,
+      "loss": 1.1116,
+      "step": 1437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390872920171512e-05,
+      "loss": 1.1331,
+      "step": 1438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389801653574588e-05,
+      "loss": 1.2588,
+      "step": 1439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388729475435784e-05,
+      "loss": 1.1101,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387656385859185e-05,
+      "loss": 1.2205,
+      "step": 1441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386582384948965e-05,
+      "loss": 1.2319,
+      "step": 1442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938550747280939e-05,
+      "loss": 1.0945,
+      "step": 1443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384431649544805e-05,
+      "loss": 1.125,
+      "step": 1444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383354915259648e-05,
+      "loss": 1.1672,
+      "step": 1445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938227727005845e-05,
+      "loss": 1.1018,
+      "step": 1446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381198714045825e-05,
+      "loss": 1.2214,
+      "step": 1447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380119247326475e-05,
+      "loss": 1.1338,
+      "step": 1448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379038870005195e-05,
+      "loss": 1.022,
+      "step": 1449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377957582186862e-05,
+      "loss": 1.0571,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937687538397645e-05,
+      "loss": 1.1545,
+      "step": 1451
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375792275479014e-05,
+      "loss": 1.0593,
+      "step": 1452
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374708256799703e-05,
+      "loss": 1.1023,
+      "step": 1453
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937362332804374e-05,
+      "loss": 1.1372,
+      "step": 1454
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372537489316462e-05,
+      "loss": 1.1599,
+      "step": 1455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937145074072327e-05,
+      "loss": 1.1971,
+      "step": 1456
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937036308236967e-05,
+      "loss": 1.033,
+      "step": 1457
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936927451436125e-05,
+      "loss": 1.1362,
+      "step": 1458
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368185036803677e-05,
+      "loss": 1.0684,
+      "step": 1459
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936709464980272e-05,
+      "loss": 1.2061,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366003353464235e-05,
+      "loss": 1.0508,
+      "step": 1461
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936491114789416e-05,
+      "loss": 0.9838,
+      "step": 1462
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936381803319852e-05,
+      "loss": 1.1704,
+      "step": 1463
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936272400948344e-05,
+      "loss": 1.0322,
+      "step": 1464
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361629076855118e-05,
+      "loss": 1.2078,
+      "step": 1465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360533235419854e-05,
+      "loss": 1.0573,
+      "step": 1466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359436485284025e-05,
+      "loss": 1.1201,
+      "step": 1467
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358338826554104e-05,
+      "loss": 1.0636,
+      "step": 1468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357240259336648e-05,
+      "loss": 1.0714,
+      "step": 1469
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356140783738304e-05,
+      "loss": 1.2192,
+      "step": 1470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935504039986581e-05,
+      "loss": 1.1353,
+      "step": 1471
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353939107825984e-05,
+      "loss": 1.0935,
+      "step": 1472
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352836907725735e-05,
+      "loss": 1.0774,
+      "step": 1473
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935173379967207e-05,
+      "loss": 1.165,
+      "step": 1474
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935062978377207e-05,
+      "loss": 1.291,
+      "step": 1475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349524860132916e-05,
+      "loss": 1.2026,
+      "step": 1476
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348419028861863e-05,
+      "loss": 1.0392,
+      "step": 1477
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347312290066273e-05,
+      "loss": 1.0736,
+      "step": 1478
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934620464385358e-05,
+      "loss": 1.0134,
+      "step": 1479
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345096090331315e-05,
+      "loss": 1.1926,
+      "step": 1480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343986629607093e-05,
+      "loss": 1.0492,
+      "step": 1481
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342876261788612e-05,
+      "loss": 1.178,
+      "step": 1482
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341764986983674e-05,
+      "loss": 1.0157,
+      "step": 1483
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340652805300154e-05,
+      "loss": 1.2993,
+      "step": 1484
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339539716846016e-05,
+      "loss": 1.153,
+      "step": 1485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338425721729327e-05,
+      "loss": 1.1436,
+      "step": 1486
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933731082005822e-05,
+      "loss": 1.2825,
+      "step": 1487
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336195011940938e-05,
+      "loss": 1.25,
+      "step": 1488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933507829748579e-05,
+      "loss": 1.1208,
+      "step": 1489
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333960676801194e-05,
+      "loss": 1.2578,
+      "step": 1490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933284214999564e-05,
+      "loss": 1.1716,
+      "step": 1491
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331722717177714e-05,
+      "loss": 1.1792,
+      "step": 1492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330602378456093e-05,
+      "loss": 1.0884,
+      "step": 1493
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329481133939527e-05,
+      "loss": 1.1331,
+      "step": 1494
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328358983736873e-05,
+      "loss": 1.106,
+      "step": 1495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932723592795706e-05,
+      "loss": 1.1748,
+      "step": 1496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326111966709122e-05,
+      "loss": 1.0686,
+      "step": 1497
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932498710010216e-05,
+      "loss": 1.0686,
+      "step": 1498
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932386132824538e-05,
+      "loss": 0.9728,
+      "step": 1499
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932273465124806e-05,
+      "loss": 1.1653,
+      "step": 1500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932160706921959e-05,
+      "loss": 1.1345,
+      "step": 1501
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320478582269427e-05,
+      "loss": 1.1992,
+      "step": 1502
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931934919050712e-05,
+      "loss": 1.1748,
+      "step": 1503
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318218894042307e-05,
+      "loss": 1.1467,
+      "step": 1504
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317087692984718e-05,
+      "loss": 1.1377,
+      "step": 1505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931595558744417e-05,
+      "loss": 1.2256,
+      "step": 1506
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9314822577530554e-05,
+      "loss": 1.0908,
+      "step": 1507
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313688663353875e-05,
+      "loss": 1.0652,
+      "step": 1508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93125538450242e-05,
+      "loss": 1.2008,
+      "step": 1509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93114181226517e-05,
+      "loss": 1.0845,
+      "step": 1510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310281496346625e-05,
+      "loss": 1.1978,
+      "step": 1511
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930914396621932e-05,
+      "loss": 1.0872,
+      "step": 1512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308005532380213e-05,
+      "loss": 1.3086,
+      "step": 1513
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306866194939822e-05,
+      "loss": 1.0219,
+      "step": 1514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930572595400875e-05,
+      "loss": 1.0898,
+      "step": 1515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304584809697684e-05,
+      "loss": 1.2417,
+      "step": 1516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303442762117414e-05,
+      "loss": 1.1255,
+      "step": 1517
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302299811378797e-05,
+      "loss": 1.219,
+      "step": 1518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301155957592798e-05,
+      "loss": 1.2646,
+      "step": 1519
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300011200870453e-05,
+      "loss": 1.2583,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298865541322893e-05,
+      "loss": 1.1479,
+      "step": 1521
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929771897906134e-05,
+      "loss": 1.2815,
+      "step": 1522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92965715141971e-05,
+      "loss": 1.1665,
+      "step": 1523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295423146841562e-05,
+      "loss": 0.9619,
+      "step": 1524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294273877106207e-05,
+      "loss": 1.0519,
+      "step": 1525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929312370510261e-05,
+      "loss": 1.1453,
+      "step": 1526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929197263094242e-05,
+      "loss": 1.1475,
+      "step": 1527
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929082065473739e-05,
+      "loss": 1.0723,
+      "step": 1528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289667776599338e-05,
+      "loss": 1.2874,
+      "step": 1529
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928851399664019e-05,
+      "loss": 1.1611,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287359314971958e-05,
+      "loss": 1.1372,
+      "step": 1531
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286203731706728e-05,
+      "loss": 1.2131,
+      "step": 1532
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928504724695669e-05,
+      "loss": 1.1233,
+      "step": 1533
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92838898608341e-05,
+      "loss": 1.1133,
+      "step": 1534
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282731573451323e-05,
+      "loss": 1.1538,
+      "step": 1535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281572384920806e-05,
+      "loss": 1.1208,
+      "step": 1536
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280412295355075e-05,
+      "loss": 1.0875,
+      "step": 1537
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279251304866747e-05,
+      "loss": 1.1847,
+      "step": 1538
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278089413568537e-05,
+      "loss": 1.2349,
+      "step": 1539
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276926621573234e-05,
+      "loss": 1.0732,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275762928993714e-05,
+      "loss": 1.303,
+      "step": 1541
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274598335942958e-05,
+      "loss": 1.0881,
+      "step": 1542
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927343284253401e-05,
+      "loss": 1.0791,
+      "step": 1543
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272266448880024e-05,
+      "loss": 1.155,
+      "step": 1544
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271099155094223e-05,
+      "loss": 1.1433,
+      "step": 1545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269930961289928e-05,
+      "loss": 1.1375,
+      "step": 1546
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268761867580544e-05,
+      "loss": 1.1704,
+      "step": 1547
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926759187407957e-05,
+      "loss": 1.1047,
+      "step": 1548
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926642098090058e-05,
+      "loss": 1.0857,
+      "step": 1549
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265249188157243e-05,
+      "loss": 1.1226,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264076495963312e-05,
+      "loss": 1.0469,
+      "step": 1551
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262902904432637e-05,
+      "loss": 1.1182,
+      "step": 1552
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926172841367914e-05,
+      "loss": 1.2568,
+      "step": 1553
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926055302381684e-05,
+      "loss": 1.1736,
+      "step": 1554
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259376734959848e-05,
+      "loss": 1.1677,
+      "step": 1555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258199547222344e-05,
+      "loss": 1.1332,
+      "step": 1556
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257021460718618e-05,
+      "loss": 1.1155,
+      "step": 1557
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255842475563024e-05,
+      "loss": 1.0476,
+      "step": 1558
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254662591870028e-05,
+      "loss": 1.3037,
+      "step": 1559
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253481809754163e-05,
+      "loss": 1.1382,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252300129330058e-05,
+      "loss": 1.1184,
+      "step": 1561
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925111755071243e-05,
+      "loss": 1.1936,
+      "step": 1562
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249934074016077e-05,
+      "loss": 1.2361,
+      "step": 1563
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248749699355892e-05,
+      "loss": 1.1802,
+      "step": 1564
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247564426846855e-05,
+      "loss": 1.2546,
+      "step": 1565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246378256604023e-05,
+      "loss": 1.0815,
+      "step": 1566
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245191188742548e-05,
+      "loss": 1.1558,
+      "step": 1567
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924400322337767e-05,
+      "loss": 1.1909,
+      "step": 1568
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242814360624714e-05,
+      "loss": 1.0406,
+      "step": 1569
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924162460059909e-05,
+      "loss": 1.1208,
+      "step": 1570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240433943416304e-05,
+      "loss": 1.0563,
+      "step": 1571
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239242389191936e-05,
+      "loss": 1.1301,
+      "step": 1572
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923804993804166e-05,
+      "loss": 1.1472,
+      "step": 1573
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236856590081235e-05,
+      "loss": 1.1801,
+      "step": 1574
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235662345426513e-05,
+      "loss": 1.1423,
+      "step": 1575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923446720419343e-05,
+      "loss": 1.1392,
+      "step": 1576
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233271166498002e-05,
+      "loss": 1.0865,
+      "step": 1577
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232074232456345e-05,
+      "loss": 1.1211,
+      "step": 1578
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230876402184647e-05,
+      "loss": 1.2041,
+      "step": 1579
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9229677675799193e-05,
+      "loss": 1.0938,
+      "step": 1580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228478053416358e-05,
+      "loss": 0.9597,
+      "step": 1581
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227277535152592e-05,
+      "loss": 1.0775,
+      "step": 1582
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922607612112444e-05,
+      "loss": 1.0901,
+      "step": 1583
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922487381144854e-05,
+      "loss": 1.2034,
+      "step": 1584
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92236706062416e-05,
+      "loss": 1.2266,
+      "step": 1585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922246650562043e-05,
+      "loss": 1.0581,
+      "step": 1586
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221261509701922e-05,
+      "loss": 1.2031,
+      "step": 1587
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220055618603045e-05,
+      "loss": 1.1116,
+      "step": 1588
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218848832440878e-05,
+      "loss": 1.1545,
+      "step": 1589
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921764115133257e-05,
+      "loss": 1.1101,
+      "step": 1590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921643257539535e-05,
+      "loss": 1.229,
+      "step": 1591
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215223104746556e-05,
+      "loss": 1.1079,
+      "step": 1592
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214012739503594e-05,
+      "loss": 1.072,
+      "step": 1593
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212801479783967e-05,
+      "loss": 1.17,
+      "step": 1594
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921158932570526e-05,
+      "loss": 1.2439,
+      "step": 1595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210376277385143e-05,
+      "loss": 1.2683,
+      "step": 1596
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209162334941386e-05,
+      "loss": 1.1797,
+      "step": 1597
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207947498491823e-05,
+      "loss": 1.1558,
+      "step": 1598
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92067317681544e-05,
+      "loss": 1.1086,
+      "step": 1599
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920551514404713e-05,
+      "loss": 1.0894,
+      "step": 1600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204297626288118e-05,
+      "loss": 1.0381,
+      "step": 1601
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203079214995565e-05,
+      "loss": 1.1567,
+      "step": 1602
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920185991028775e-05,
+      "loss": 1.1458,
+      "step": 1603
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200639712283037e-05,
+      "loss": 1.2078,
+      "step": 1604
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199418621099886e-05,
+      "loss": 1.2209,
+      "step": 1605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9198196636856833e-05,
+      "loss": 1.2288,
+      "step": 1606
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196973759672506e-05,
+      "loss": 1.115,
+      "step": 1607
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919574998966562e-05,
+      "loss": 1.1497,
+      "step": 1608
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194525326954978e-05,
+      "loss": 1.196,
+      "step": 1609
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193299771659465e-05,
+      "loss": 1.1135,
+      "step": 1610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919207332389806e-05,
+      "loss": 1.1306,
+      "step": 1611
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190845983789814e-05,
+      "loss": 1.1437,
+      "step": 1612
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189617751453885e-05,
+      "loss": 1.1221,
+      "step": 1613
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91883886270095e-05,
+      "loss": 1.179,
+      "step": 1614
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187158610575985e-05,
+      "loss": 1.0515,
+      "step": 1615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185927702272746e-05,
+      "loss": 1.0422,
+      "step": 1616
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184695902219274e-05,
+      "loss": 1.1594,
+      "step": 1617
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918346321053515e-05,
+      "loss": 1.1179,
+      "step": 1618
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182229627340048e-05,
+      "loss": 1.0403,
+      "step": 1619
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180995152753713e-05,
+      "loss": 1.1304,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179759786895987e-05,
+      "loss": 1.0808,
+      "step": 1621
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178523529886802e-05,
+      "loss": 0.9904,
+      "step": 1622
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177286381846165e-05,
+      "loss": 1.179,
+      "step": 1623
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176048342894176e-05,
+      "loss": 1.1331,
+      "step": 1624
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174809413151026e-05,
+      "loss": 1.1412,
+      "step": 1625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917356959273698e-05,
+      "loss": 1.2009,
+      "step": 1626
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917232888177241e-05,
+      "loss": 1.0028,
+      "step": 1627
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917108728037775e-05,
+      "loss": 1.144,
+      "step": 1628
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169844788673534e-05,
+      "loss": 1.1802,
+      "step": 1629
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916860140678038e-05,
+      "loss": 1.1787,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167357134819e-05,
+      "loss": 1.2,
+      "step": 1631
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166111972910178e-05,
+      "loss": 1.205,
+      "step": 1632
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164865921174792e-05,
+      "loss": 1.0706,
+      "step": 1633
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163618979733808e-05,
+      "loss": 1.1194,
+      "step": 1634
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162371148708278e-05,
+      "loss": 1.1572,
+      "step": 1635
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161122428219338e-05,
+      "loss": 1.2756,
+      "step": 1636
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159872818388207e-05,
+      "loss": 1.0461,
+      "step": 1637
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91586223193362e-05,
+      "loss": 1.2234,
+      "step": 1638
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915737093118471e-05,
+      "loss": 1.1255,
+      "step": 1639
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156118654055217e-05,
+      "loss": 0.98,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154865488069294e-05,
+      "loss": 1.1108,
+      "step": 1641
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915361143334859e-05,
+      "loss": 1.2192,
+      "step": 1642
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152356490014855e-05,
+      "loss": 1.1011,
+      "step": 1643
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151100658189906e-05,
+      "loss": 1.0168,
+      "step": 1644
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149843937995665e-05,
+      "loss": 1.1697,
+      "step": 1645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148586329554123e-05,
+      "loss": 1.2271,
+      "step": 1646
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147327832987375e-05,
+      "loss": 1.1313,
+      "step": 1647
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146068448417584e-05,
+      "loss": 1.0122,
+      "step": 1648
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144808175967017e-05,
+      "loss": 1.0828,
+      "step": 1649
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143547015758014e-05,
+      "loss": 1.0415,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142284967913007e-05,
+      "loss": 1.046,
+      "step": 1651
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141022032554514e-05,
+      "loss": 1.1479,
+      "step": 1652
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139758209805132e-05,
+      "loss": 1.093,
+      "step": 1653
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913849349978756e-05,
+      "loss": 1.0483,
+      "step": 1654
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137227902624564e-05,
+      "loss": 1.2371,
+      "step": 1655
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913596141843901e-05,
+      "loss": 0.9709,
+      "step": 1656
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134694047353846e-05,
+      "loss": 1.2344,
+      "step": 1657
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133425789492106e-05,
+      "loss": 1.1411,
+      "step": 1658
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132156644976906e-05,
+      "loss": 1.2783,
+      "step": 1659
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130886613931457e-05,
+      "loss": 1.1919,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129615696479046e-05,
+      "loss": 1.1343,
+      "step": 1661
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128343892743055e-05,
+      "loss": 1.0277,
+      "step": 1662
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127071202846945e-05,
+      "loss": 1.0935,
+      "step": 1663
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912579762691427e-05,
+      "loss": 1.0872,
+      "step": 1664
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912452316506866e-05,
+      "loss": 1.061,
+      "step": 1665
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912324781743384e-05,
+      "loss": 1.186,
+      "step": 1666
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912197158413362e-05,
+      "loss": 1.085,
+      "step": 1667
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120694465291893e-05,
+      "loss": 1.1587,
+      "step": 1668
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119416461032636e-05,
+      "loss": 1.1542,
+      "step": 1669
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911813757147992e-05,
+      "loss": 1.144,
+      "step": 1670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9116857796757892e-05,
+      "loss": 1.1377,
+      "step": 1671
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115577136990792e-05,
+      "loss": 1.0867,
+      "step": 1672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114295592302944e-05,
+      "loss": 1.1484,
+      "step": 1673
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113013162818757e-05,
+      "loss": 1.1804,
+      "step": 1674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111729848662728e-05,
+      "loss": 1.0931,
+      "step": 1675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110445649959435e-05,
+      "loss": 1.0994,
+      "step": 1676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109160566833548e-05,
+      "loss": 1.2117,
+      "step": 1677
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910787459940982e-05,
+      "loss": 1.2377,
+      "step": 1678
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9106587747813092e-05,
+      "loss": 1.1929,
+      "step": 1679
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910530001216828e-05,
+      "loss": 1.1287,
+      "step": 1680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104011392600407e-05,
+      "loss": 1.1271,
+      "step": 1681
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910272188923456e-05,
+      "loss": 1.2349,
+      "step": 1682
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101431502195925e-05,
+      "loss": 1.0654,
+      "step": 1683
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910014023160977e-05,
+      "loss": 1.1487,
+      "step": 1684
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909884807760145e-05,
+      "loss": 1.1184,
+      "step": 1685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90975550402964e-05,
+      "loss": 1.0605,
+      "step": 1686
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909626111982015e-05,
+      "loss": 1.1509,
+      "step": 1687
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909496631629831e-05,
+      "loss": 1.1914,
+      "step": 1688
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093670629856577e-05,
+      "loss": 1.1954,
+      "step": 1689
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909237406062073e-05,
+      "loss": 1.2048,
+      "step": 1690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091076608716638e-05,
+      "loss": 1.1162,
+      "step": 1691
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089778274270262e-05,
+      "loss": 1.1196,
+      "step": 1692
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9088479057407632e-05,
+      "loss": 1.2356,
+      "step": 1693
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087178958254877e-05,
+      "loss": 1.1396,
+      "step": 1694
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908587797693821e-05,
+      "loss": 1.1729,
+      "step": 1695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084576113583924e-05,
+      "loss": 1.1873,
+      "step": 1696
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083273368318403e-05,
+      "loss": 1.0773,
+      "step": 1697
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081969741268117e-05,
+      "loss": 1.1738,
+      "step": 1698
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080665232559614e-05,
+      "loss": 1.0768,
+      "step": 1699
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9079359842319534e-05,
+      "loss": 1.2454,
+      "step": 1700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907805357067461e-05,
+      "loss": 1.161,
+      "step": 1701
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907674641775164e-05,
+      "loss": 1.2083,
+      "step": 1702
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075438383677523e-05,
+      "loss": 1.1975,
+      "step": 1703
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074129468579247e-05,
+      "loss": 1.082,
+      "step": 1704
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907281967258387e-05,
+      "loss": 1.1841,
+      "step": 1705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071508995818547e-05,
+      "loss": 1.2451,
+      "step": 1706
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907019743841052e-05,
+      "loss": 0.9617,
+      "step": 1707
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068885000487107e-05,
+      "loss": 1.1594,
+      "step": 1708
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906757168217572e-05,
+      "loss": 1.1377,
+      "step": 1709
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906625748360385e-05,
+      "loss": 1.1411,
+      "step": 1710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064942404899075e-05,
+      "loss": 1.0851,
+      "step": 1711
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063626446189068e-05,
+      "loss": 0.9868,
+      "step": 1712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062309607601575e-05,
+      "loss": 1.1582,
+      "step": 1713
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906099188926443e-05,
+      "loss": 1.1035,
+      "step": 1714
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9059673291305553e-05,
+      "loss": 1.1819,
+      "step": 1715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058353813852958e-05,
+      "loss": 1.0352,
+      "step": 1716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905703345703473e-05,
+      "loss": 1.1934,
+      "step": 1717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055712220979052e-05,
+      "loss": 1.1436,
+      "step": 1718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905439010581418e-05,
+      "loss": 1.1399,
+      "step": 1719
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053067111668465e-05,
+      "loss": 0.9601,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9051743238670344e-05,
+      "loss": 1.033,
+      "step": 1721
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050418486948334e-05,
+      "loss": 1.02,
+      "step": 1722
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049092856631035e-05,
+      "loss": 1.0403,
+      "step": 1723
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047766347847142e-05,
+      "loss": 1.1752,
+      "step": 1724
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046438960725428e-05,
+      "loss": 1.1301,
+      "step": 1725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904511069539475e-05,
+      "loss": 1.2234,
+      "step": 1726
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9043781551984053e-05,
+      "loss": 1.1628,
+      "step": 1727
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042451530622372e-05,
+      "loss": 1.0972,
+      "step": 1728
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041120631438818e-05,
+      "loss": 1.1772,
+      "step": 1729
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039788854562597e-05,
+      "loss": 1.2334,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903845620012299e-05,
+      "loss": 1.1575,
+      "step": 1731
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037122668249372e-05,
+      "loss": 1.0737,
+      "step": 1732
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90357882590712e-05,
+      "loss": 1.1692,
+      "step": 1733
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034452972718008e-05,
+      "loss": 1.1628,
+      "step": 1734
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903311680931943e-05,
+      "loss": 1.1799,
+      "step": 1735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903177976900518e-05,
+      "loss": 1.2107,
+      "step": 1736
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9030441851905046e-05,
+      "loss": 1.1418,
+      "step": 1737
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029103058148916e-05,
+      "loss": 1.1208,
+      "step": 1738
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027763387866757e-05,
+      "loss": 1.0586,
+      "step": 1739
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902642284118862e-05,
+      "loss": 1.1088,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025081418244645e-05,
+      "loss": 1.1699,
+      "step": 1741
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902373911916505e-05,
+      "loss": 1.0353,
+      "step": 1742
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022395944080147e-05,
+      "loss": 1.2529,
+      "step": 1743
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021051893120324e-05,
+      "loss": 1.0837,
+      "step": 1744
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019706966416068e-05,
+      "loss": 1.2061,
+      "step": 1745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901836116409793e-05,
+      "loss": 1.1025,
+      "step": 1746
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017014486296563e-05,
+      "loss": 1.1829,
+      "step": 1747
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90156669331427e-05,
+      "loss": 1.0796,
+      "step": 1748
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014318504767155e-05,
+      "loss": 1.2661,
+      "step": 1749
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012969201300833e-05,
+      "loss": 1.124,
+      "step": 1750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011619022874726e-05,
+      "loss": 1.1816,
+      "step": 1751
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90102679696199e-05,
+      "loss": 1.1172,
+      "step": 1752
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008916041667513e-05,
+      "loss": 1.2023,
+      "step": 1753
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007563239148812e-05,
+      "loss": 1.0413,
+      "step": 1754
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006209562195116e-05,
+      "loss": 1.1133,
+      "step": 1755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9004855010937847e-05,
+      "loss": 1.0742,
+      "step": 1756
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003499585508495e-05,
+      "loss": 1.1244,
+      "step": 1757
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002143286038642e-05,
+      "loss": 1.2314,
+      "step": 1758
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900078611265996e-05,
+      "loss": 1.2017,
+      "step": 1759
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8999428065504195e-05,
+      "loss": 1.1392,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998069144703187e-05,
+      "loss": 1.2256,
+      "step": 1761
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996709350388855e-05,
+      "loss": 1.0303,
+      "step": 1762
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8995348682693203e-05,
+      "loss": 1.0162,
+      "step": 1763
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993987141748324e-05,
+      "loss": 1.2554,
+      "step": 1764
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992624727686395e-05,
+      "loss": 1.1,
+      "step": 1765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991261440639674e-05,
+      "loss": 1.1554,
+      "step": 1766
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898989728074051e-05,
+      "loss": 1.196,
+      "step": 1767
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988532248121324e-05,
+      "loss": 1.0864,
+      "step": 1768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987166342914638e-05,
+      "loss": 1.1194,
+      "step": 1769
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985799565253052e-05,
+      "loss": 1.1069,
+      "step": 1770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984431915269243e-05,
+      "loss": 1.1194,
+      "step": 1771
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983063393095983e-05,
+      "loss": 1.2424,
+      "step": 1772
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898169399886613e-05,
+      "loss": 1.1917,
+      "step": 1773
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898032373271261e-05,
+      "loss": 0.9967,
+      "step": 1774
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978952594768464e-05,
+      "loss": 1.198,
+      "step": 1775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897758058516678e-05,
+      "loss": 1.2039,
+      "step": 1776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897620770404076e-05,
+      "loss": 1.1216,
+      "step": 1777
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897483395152368e-05,
+      "loss": 1.1854,
+      "step": 1778
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973459327748897e-05,
+      "loss": 1.1423,
+      "step": 1779
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972083832849863e-05,
+      "loss": 1.2104,
+      "step": 1780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89707074669601e-05,
+      "loss": 1.2061,
+      "step": 1781
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969330230213225e-05,
+      "loss": 0.9805,
+      "step": 1782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967952122742942e-05,
+      "loss": 1.1389,
+      "step": 1783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8966573144683028e-05,
+      "loss": 1.012,
+      "step": 1784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896519329616736e-05,
+      "loss": 1.0605,
+      "step": 1785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896381257732988e-05,
+      "loss": 1.2031,
+      "step": 1786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962430988304638e-05,
+      "loss": 1.0989,
+      "step": 1787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896104852922574e-05,
+      "loss": 1.0911,
+      "step": 1788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959665200227406e-05,
+      "loss": 1.1427,
+      "step": 1789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895828100144392e-05,
+      "loss": 1.0125,
+      "step": 1790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895689593300966e-05,
+      "loss": 1.064,
+      "step": 1791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895550999505908e-05,
+      "loss": 1.147,
+      "step": 1792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954123187726728e-05,
+      "loss": 1.2439,
+      "step": 1793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8952735511147234e-05,
+      "loss": 1.1077,
+      "step": 1794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951346965455307e-05,
+      "loss": 1.1853,
+      "step": 1795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949957550785748e-05,
+      "loss": 1.1047,
+      "step": 1796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948567267273433e-05,
+      "loss": 1.0723,
+      "step": 1797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8947176115053334e-05,
+      "loss": 1.1755,
+      "step": 1798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945784094260497e-05,
+      "loss": 1.1814,
+      "step": 1799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944391205030057e-05,
+      "loss": 1.1208,
+      "step": 1800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894299744749723e-05,
+      "loss": 1.134,
+      "step": 1801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894160282179733e-05,
+      "loss": 1.073,
+      "step": 1802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940207328065728e-05,
+      "loss": 1.1863,
+      "step": 1803
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938810966437908e-05,
+      "loss": 1.0916,
+      "step": 1804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893741373704942e-05,
+      "loss": 1.2048,
+      "step": 1805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936015640035907e-05,
+      "loss": 1.1409,
+      "step": 1806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8934616675533092e-05,
+      "loss": 1.067,
+      "step": 1807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933216843676787e-05,
+      "loss": 1.0117,
+      "step": 1808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931816144602874e-05,
+      "loss": 1.0813,
+      "step": 1809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893041457844734e-05,
+      "loss": 1.1995,
+      "step": 1810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8929012145346243e-05,
+      "loss": 1.1252,
+      "step": 1811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927608845435728e-05,
+      "loss": 1.2004,
+      "step": 1812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926204678852027e-05,
+      "loss": 1.0232,
+      "step": 1813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892479964573145e-05,
+      "loss": 1.2356,
+      "step": 1814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923393746210395e-05,
+      "loss": 1.1289,
+      "step": 1815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921986980425344e-05,
+      "loss": 1.1931,
+      "step": 1816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920579348512864e-05,
+      "loss": 1.1621,
+      "step": 1817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919170850609606e-05,
+      "loss": 1.2292,
+      "step": 1818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89177614868523e-05,
+      "loss": 1.189,
+      "step": 1819
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916351257377764e-05,
+      "loss": 1.1516,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914940162322907e-05,
+      "loss": 1.1172,
+      "step": 1821
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913528201824706e-05,
+      "loss": 1.2312,
+      "step": 1822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891211537602024e-05,
+      "loss": 1.1108,
+      "step": 1823
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8910701685046656e-05,
+      "loss": 1.0115,
+      "step": 1824
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890928712904119e-05,
+      "loss": 1.147,
+      "step": 1825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907871708141177e-05,
+      "loss": 1.1257,
+      "step": 1826
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890645542248401e-05,
+      "loss": 1.228,
+      "step": 1827
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905038272207187e-05,
+      "loss": 1.1755,
+      "step": 1828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903620257448273e-05,
+      "loss": 1.2417,
+      "step": 1829
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902201378344938e-05,
+      "loss": 1.054,
+      "step": 1830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900781635034914e-05,
+      "loss": 1.1643,
+      "step": 1831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889936102765603e-05,
+      "loss": 1.139,
+      "step": 1832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897939556346196e-05,
+      "loss": 1.1465,
+      "step": 1833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896517221243404e-05,
+      "loss": 1.1838,
+      "step": 1834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895094022485735e-05,
+      "loss": 1.0713,
+      "step": 1835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8893669960211342e-05,
+      "loss": 1.0867,
+      "step": 1836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892245034558478e-05,
+      "loss": 1.1213,
+      "step": 1837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889081924566547e-05,
+      "loss": 1.0383,
+      "step": 1838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888939259367073e-05,
+      "loss": 1.1069,
+      "step": 1839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888796507871275e-05,
+      "loss": 1.2305,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886536700930116e-05,
+      "loss": 1.0681,
+      "step": 1841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888510746046149e-05,
+      "loss": 1.219,
+      "step": 1842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883677357445617e-05,
+      "loss": 1.2148,
+      "step": 1843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888224639202133e-05,
+      "loss": 1.0417,
+      "step": 1844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880814564327548e-05,
+      "loss": 1.1843,
+      "step": 1845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879381874503265e-05,
+      "loss": 1.0103,
+      "step": 1846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877948322687568e-05,
+      "loss": 1.085,
+      "step": 1847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8876513909019615e-05,
+      "loss": 1.2163,
+      "step": 1848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875078633638664e-05,
+      "loss": 1.0673,
+      "step": 1849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873642496684047e-05,
+      "loss": 1.1226,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872205498295174e-05,
+      "loss": 1.1565,
+      "step": 1851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870767638611555e-05,
+      "loss": 1.0286,
+      "step": 1852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886932891777277e-05,
+      "loss": 1.2434,
+      "step": 1853
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867889335918487e-05,
+      "loss": 1.0925,
+      "step": 1854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866448893188458e-05,
+      "loss": 1.1699,
+      "step": 1855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886500758972252e-05,
+      "loss": 1.0942,
+      "step": 1856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886356542566059e-05,
+      "loss": 1.2136,
+      "step": 1857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862122401142668e-05,
+      "loss": 1.1982,
+      "step": 1858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886067851630884e-05,
+      "loss": 1.0713,
+      "step": 1859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885923377129928e-05,
+      "loss": 1.1294,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8857788166254236e-05,
+      "loss": 1.0933,
+      "step": 1861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8856341701314047e-05,
+      "loss": 1.1624,
+      "step": 1862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885489437661913e-05,
+      "loss": 1.2009,
+      "step": 1863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853446192309988e-05,
+      "loss": 1.0957,
+      "step": 1864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851997148527217e-05,
+      "loss": 1.2244,
+      "step": 1865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8850547245411476e-05,
+      "loss": 1.1273,
+      "step": 1866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849096483103518e-05,
+      "loss": 0.9983,
+      "step": 1867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847644861744188e-05,
+      "loss": 0.9473,
+      "step": 1868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846192381474404e-05,
+      "loss": 1.2284,
+      "step": 1869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8844739042435164e-05,
+      "loss": 1.2065,
+      "step": 1870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8843284844767564e-05,
+      "loss": 0.9564,
+      "step": 1871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841829788612762e-05,
+      "loss": 1.123,
+      "step": 1872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840373874112024e-05,
+      "loss": 1.1821,
+      "step": 1873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838917101406683e-05,
+      "loss": 1.1995,
+      "step": 1874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8837459470638158e-05,
+      "loss": 1.0649,
+      "step": 1875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883600098194795e-05,
+      "loss": 1.156,
+      "step": 1876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834541635477652e-05,
+      "loss": 1.1082,
+      "step": 1877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8833081431368932e-05,
+      "loss": 1.0835,
+      "step": 1878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883162036976354e-05,
+      "loss": 1.1479,
+      "step": 1879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883015845080332e-05,
+      "loss": 1.1643,
+      "step": 1880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828695674630183e-05,
+      "loss": 1.1614,
+      "step": 1881
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882723204138614e-05,
+      "loss": 1.0531,
+      "step": 1882
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8825767551213275e-05,
+      "loss": 1.2429,
+      "step": 1883
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824302204253756e-05,
+      "loss": 1.1511,
+      "step": 1884
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8822836000649837e-05,
+      "loss": 1.0839,
+      "step": 1885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821368940543855e-05,
+      "loss": 1.1023,
+      "step": 1886
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819901024078225e-05,
+      "loss": 1.1594,
+      "step": 1887
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818432251395452e-05,
+      "loss": 1.1982,
+      "step": 1888
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881696262263812e-05,
+      "loss": 1.1948,
+      "step": 1889
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88154921379489e-05,
+      "loss": 1.1667,
+      "step": 1890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881402079747054e-05,
+      "loss": 1.2085,
+      "step": 1891
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881254860134588e-05,
+      "loss": 1.1313,
+      "step": 1892
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811075549717835e-05,
+      "loss": 1.1851,
+      "step": 1893
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88096016427294e-05,
+      "loss": 1.2429,
+      "step": 1894
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8808126880523666e-05,
+      "loss": 1.1296,
+      "step": 1895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88066512632438e-05,
+      "loss": 0.9797,
+      "step": 1896
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805174791033044e-05,
+      "loss": 1.0759,
+      "step": 1897
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880369746403474e-05,
+      "loss": 1.228,
+      "step": 1898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88022192823923e-05,
+      "loss": 1.0044,
+      "step": 1899
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880074024624922e-05,
+      "loss": 1.228,
+      "step": 1900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799260355749086e-05,
+      "loss": 1.1326,
+      "step": 1901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879777961103556e-05,
+      "loss": 1.0789,
+      "step": 1902
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796298012252392e-05,
+      "loss": 1.229,
+      "step": 1903
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8794815559543408e-05,
+      "loss": 1.1024,
+      "step": 1904
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793332253052525e-05,
+      "loss": 1.1797,
+      "step": 1905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879184809292374e-05,
+      "loss": 1.1614,
+      "step": 1906
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879036307930113e-05,
+      "loss": 1.0698,
+      "step": 1907
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788877212328852e-05,
+      "loss": 1.1826,
+      "step": 1908
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787390492151162e-05,
+      "loss": 1.2134,
+      "step": 1909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878590291891238e-05,
+      "loss": 1.0498,
+      "step": 1910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878441449275692e-05,
+      "loss": 1.1902,
+      "step": 1911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878292521382927e-05,
+      "loss": 1.1599,
+      "step": 1912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8781435082274013e-05,
+      "loss": 1.0605,
+      "step": 1913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87799440982358e-05,
+      "loss": 1.1598,
+      "step": 1914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877845226185938e-05,
+      "loss": 1.0974,
+      "step": 1915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776959573289572e-05,
+      "loss": 1.2368,
+      "step": 1916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775466032671288e-05,
+      "loss": 1.0994,
+      "step": 1917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773971640149515e-05,
+      "loss": 1.0966,
+      "step": 1918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772476395869323e-05,
+      "loss": 1.2014,
+      "step": 1919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770980299975867e-05,
+      "loss": 1.0427,
+      "step": 1920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876948335261439e-05,
+      "loss": 1.1589,
+      "step": 1921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876798555393021e-05,
+      "loss": 1.2163,
+      "step": 1922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876648690406873e-05,
+      "loss": 1.1179,
+      "step": 1923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764987403175433e-05,
+      "loss": 1.2666,
+      "step": 1924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876348705139589e-05,
+      "loss": 1.1521,
+      "step": 1925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761985848875752e-05,
+      "loss": 1.1882,
+      "step": 1926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760483795760753e-05,
+      "loss": 1.1208,
+      "step": 1927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758980892196706e-05,
+      "loss": 1.1504,
+      "step": 1928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757477138329515e-05,
+      "loss": 1.0649,
+      "step": 1929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875597253430516e-05,
+      "loss": 1.0847,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87544670802697e-05,
+      "loss": 1.1655,
+      "step": 1931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752960776369283e-05,
+      "loss": 1.1172,
+      "step": 1932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751453622750142e-05,
+      "loss": 1.0994,
+      "step": 1933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749945619558585e-05,
+      "loss": 1.2329,
+      "step": 1934
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748436766941e-05,
+      "loss": 1.1226,
+      "step": 1935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746927065043876e-05,
+      "loss": 1.1455,
+      "step": 1936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8745416514013764e-05,
+      "loss": 1.189,
+      "step": 1937
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87439051139973e-05,
+      "loss": 1.0386,
+      "step": 1938
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874239286514122e-05,
+      "loss": 1.0859,
+      "step": 1939
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874087976759232e-05,
+      "loss": 1.1772,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739365821497496e-05,
+      "loss": 1.2427,
+      "step": 1941
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873785102700371e-05,
+      "loss": 1.1716,
+      "step": 1942
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736335384258022e-05,
+      "loss": 1.2163,
+      "step": 1943
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873481889340756e-05,
+      "loss": 1.165,
+      "step": 1944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873330155459955e-05,
+      "loss": 1.1367,
+      "step": 1945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731783367981283e-05,
+      "loss": 1.0112,
+      "step": 1946
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730264333700153e-05,
+      "loss": 1.0637,
+      "step": 1947
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728744451903617e-05,
+      "loss": 1.0088,
+      "step": 1948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872722372273922e-05,
+      "loss": 1.2153,
+      "step": 1949
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725702146354593e-05,
+      "loss": 1.1855,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724179722897452e-05,
+      "loss": 1.3096,
+      "step": 1951
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722656452515588e-05,
+      "loss": 1.0476,
+      "step": 1952
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872113233535687e-05,
+      "loss": 1.2063,
+      "step": 1953
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871960737156927e-05,
+      "loss": 1.1229,
+      "step": 1954
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718081561300817e-05,
+      "loss": 1.2891,
+      "step": 1955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716554904699635e-05,
+      "loss": 1.0229,
+      "step": 1956
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715027401913932e-05,
+      "loss": 1.0796,
+      "step": 1957
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871349905309199e-05,
+      "loss": 1.0881,
+      "step": 1958
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871196985838218e-05,
+      "loss": 1.1077,
+      "step": 1959
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871043981793296e-05,
+      "loss": 1.076,
+      "step": 1960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8708908931892853e-05,
+      "loss": 1.1865,
+      "step": 1961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870737720041048e-05,
+      "loss": 1.0725,
+      "step": 1962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705844623634533e-05,
+      "loss": 1.158,
+      "step": 1963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704311201713795e-05,
+      "loss": 1.1912,
+      "step": 1964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702776934797125e-05,
+      "loss": 1.0652,
+      "step": 1965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701241823033473e-05,
+      "loss": 1.0143,
+      "step": 1966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699705866571856e-05,
+      "loss": 1.2107,
+      "step": 1967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698169065561385e-05,
+      "loss": 1.1592,
+      "step": 1968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696631420151254e-05,
+      "loss": 1.1086,
+      "step": 1969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695092930490722e-05,
+      "loss": 1.0881,
+      "step": 1970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693553596729153e-05,
+      "loss": 1.1509,
+      "step": 1971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869201341901598e-05,
+      "loss": 1.0747,
+      "step": 1972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690472397500717e-05,
+      "loss": 1.1624,
+      "step": 1973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868893053233297e-05,
+      "loss": 1.0403,
+      "step": 1974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687387823662408e-05,
+      "loss": 1.1652,
+      "step": 1975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685844271638808e-05,
+      "loss": 1.0244,
+      "step": 1976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684299876412e-05,
+      "loss": 1.0562,
+      "step": 1977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682754638131926e-05,
+      "loss": 1.1733,
+      "step": 1978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681208556948583e-05,
+      "loss": 1.0608,
+      "step": 1979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8679661633012065e-05,
+      "loss": 1.1501,
+      "step": 1980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678113866472545e-05,
+      "loss": 1.1553,
+      "step": 1981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676565257480277e-05,
+      "loss": 1.1309,
+      "step": 1982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675015806185594e-05,
+      "loss": 1.1675,
+      "step": 1983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673465512738917e-05,
+      "loss": 1.1521,
+      "step": 1984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671914377290743e-05,
+      "loss": 1.0416,
+      "step": 1985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670362399991654e-05,
+      "loss": 1.1348,
+      "step": 1986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668809580992307e-05,
+      "loss": 1.1228,
+      "step": 1987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8667255920443453e-05,
+      "loss": 1.1221,
+      "step": 1988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665701418495922e-05,
+      "loss": 1.2031,
+      "step": 1989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866414607530061e-05,
+      "loss": 1.1688,
+      "step": 1990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662589891008512e-05,
+      "loss": 1.0139,
+      "step": 1991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8661032865770706e-05,
+      "loss": 1.146,
+      "step": 1992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659474999738333e-05,
+      "loss": 1.1117,
+      "step": 1993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865791629306263e-05,
+      "loss": 1.1416,
+      "step": 1994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656356745894922e-05,
+      "loss": 1.322,
+      "step": 1995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8654796358386595e-05,
+      "loss": 1.0742,
+      "step": 1996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653235130689132e-05,
+      "loss": 1.1338,
+      "step": 1997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8651673062954097e-05,
+      "loss": 1.1942,
+      "step": 1998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865011015533313e-05,
+      "loss": 1.1787,
+      "step": 1999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864854640797795e-05,
+      "loss": 1.2046,
+      "step": 2000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646981821040377e-05,
+      "loss": 1.2234,
+      "step": 2001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8645416394672278e-05,
+      "loss": 1.0551,
+      "step": 2002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643850129025635e-05,
+      "loss": 1.1199,
+      "step": 2003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864228302425249e-05,
+      "loss": 1.1067,
+      "step": 2004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864071508050498e-05,
+      "loss": 1.0933,
+      "step": 2005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8639146297935316e-05,
+      "loss": 1.1838,
+      "step": 2006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863757667669579e-05,
+      "loss": 1.1536,
+      "step": 2007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636006216938777e-05,
+      "loss": 1.2712,
+      "step": 2008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863443491881674e-05,
+      "loss": 1.2983,
+      "step": 2009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8632862782482208e-05,
+      "loss": 1.1382,
+      "step": 2010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631289808087808e-05,
+      "loss": 1.1461,
+      "step": 2011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862971599578624e-05,
+      "loss": 1.2644,
+      "step": 2012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862814134573028e-05,
+      "loss": 1.1555,
+      "step": 2013
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626565858072805e-05,
+      "loss": 1.0059,
+      "step": 2014
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8624989532966746e-05,
+      "loss": 1.134,
+      "step": 2015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623412370565135e-05,
+      "loss": 1.2097,
+      "step": 2016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862183437102108e-05,
+      "loss": 1.0776,
+      "step": 2017
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8620255534487767e-05,
+      "loss": 1.1885,
+      "step": 2018
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861867586111847e-05,
+      "loss": 1.12,
+      "step": 2019
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617095351066542e-05,
+      "loss": 1.1619,
+      "step": 2020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615514004485405e-05,
+      "loss": 0.9954,
+      "step": 2021
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613931821528588e-05,
+      "loss": 1.0239,
+      "step": 2022
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8612348802349675e-05,
+      "loss": 1.05,
+      "step": 2023
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861076494710234e-05,
+      "loss": 1.0771,
+      "step": 2024
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860918025594035e-05,
+      "loss": 1.1685,
+      "step": 2025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860759472901754e-05,
+      "loss": 1.2107,
+      "step": 2026
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8606008366487822e-05,
+      "loss": 1.106,
+      "step": 2027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860442116850521e-05,
+      "loss": 1.2058,
+      "step": 2028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602833135223773e-05,
+      "loss": 1.0967,
+      "step": 2029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860124426679768e-05,
+      "loss": 1.163,
+      "step": 2030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599654563381178e-05,
+      "loss": 1.1313,
+      "step": 2031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598064025128587e-05,
+      "loss": 1.1272,
+      "step": 2032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8596472652194313e-05,
+      "loss": 1.2456,
+      "step": 2033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594880444732846e-05,
+      "loss": 1.0377,
+      "step": 2034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593287402898755e-05,
+      "loss": 1.2134,
+      "step": 2035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8591693526846685e-05,
+      "loss": 1.0667,
+      "step": 2036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590098816731365e-05,
+      "loss": 1.1298,
+      "step": 2037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8588503272707612e-05,
+      "loss": 1.158,
+      "step": 2038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8586906894930315e-05,
+      "loss": 1.0972,
+      "step": 2039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8585309683554444e-05,
+      "loss": 1.1033,
+      "step": 2040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858371163873506e-05,
+      "loss": 1.0618,
+      "step": 2041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858211276062729e-05,
+      "loss": 1.1274,
+      "step": 2042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580513049386356e-05,
+      "loss": 1.0938,
+      "step": 2043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857891250516755e-05,
+      "loss": 1.1411,
+      "step": 2044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577311128126255e-05,
+      "loss": 1.1836,
+      "step": 2045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857570891841792e-05,
+      "loss": 1.1331,
+      "step": 2046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574105876198093e-05,
+      "loss": 1.1311,
+      "step": 2047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572502001622388e-05,
+      "loss": 1.1744,
+      "step": 2048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570897294846508e-05,
+      "loss": 1.0654,
+      "step": 2049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569291756026237e-05,
+      "loss": 1.187,
+      "step": 2050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567685385317437e-05,
+      "loss": 1.0355,
+      "step": 2051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8566078182876045e-05,
+      "loss": 1.2498,
+      "step": 2052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564470148858088e-05,
+      "loss": 1.1733,
+      "step": 2053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8562861283419672e-05,
+      "loss": 1.113,
+      "step": 2054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561251586716985e-05,
+      "loss": 1.1943,
+      "step": 2055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855964105890629e-05,
+      "loss": 1.2441,
+      "step": 2056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8558029700143926e-05,
+      "loss": 1.1929,
+      "step": 2057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855641751058633e-05,
+      "loss": 1.1428,
+      "step": 2058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855480449039001e-05,
+      "loss": 1.1978,
+      "step": 2059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8553190639711554e-05,
+      "loss": 1.1809,
+      "step": 2060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551575958707623e-05,
+      "loss": 1.209,
+      "step": 2061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549960447534975e-05,
+      "loss": 1.0928,
+      "step": 2062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854834410635044e-05,
+      "loss": 1.1543,
+      "step": 2063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8546726935310925e-05,
+      "loss": 1.1687,
+      "step": 2064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545108934573424e-05,
+      "loss": 1.1912,
+      "step": 2065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8543490104295006e-05,
+      "loss": 1.1389,
+      "step": 2066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854187044463283e-05,
+      "loss": 1.1113,
+      "step": 2067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540249955744124e-05,
+      "loss": 1.1733,
+      "step": 2068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8538628637786203e-05,
+      "loss": 1.1577,
+      "step": 2069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853700649091646e-05,
+      "loss": 1.2561,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853538351529237e-05,
+      "loss": 1.1746,
+      "step": 2071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853375971107149e-05,
+      "loss": 1.1267,
+      "step": 2072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532135078411453e-05,
+      "loss": 1.1763,
+      "step": 2073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853050961746998e-05,
+      "loss": 1.0753,
+      "step": 2074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528883328404857e-05,
+      "loss": 1.1605,
+      "step": 2075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852725621137397e-05,
+      "loss": 1.1995,
+      "step": 2076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525628266535272e-05,
+      "loss": 1.049,
+      "step": 2077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85239994940468e-05,
+      "loss": 1.094,
+      "step": 2078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522369894066674e-05,
+      "loss": 1.158,
+      "step": 2079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852073946675309e-05,
+      "loss": 1.0929,
+      "step": 2080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851910821226433e-05,
+      "loss": 1.2061,
+      "step": 2081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851747613075875e-05,
+      "loss": 1.0649,
+      "step": 2082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515843222394786e-05,
+      "loss": 1.0891,
+      "step": 2083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8514209487330964e-05,
+      "loss": 1.0713,
+      "step": 2084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512574925725878e-05,
+      "loss": 1.2319,
+      "step": 2085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510939537738212e-05,
+      "loss": 1.1168,
+      "step": 2086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850930332352672e-05,
+      "loss": 1.0754,
+      "step": 2087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507666283250246e-05,
+      "loss": 1.0901,
+      "step": 2088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506028417067712e-05,
+      "loss": 1.1626,
+      "step": 2089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8504389725138118e-05,
+      "loss": 1.0701,
+      "step": 2090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850275020762054e-05,
+      "loss": 1.1655,
+      "step": 2091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501109864674146e-05,
+      "loss": 1.167,
+      "step": 2092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849946869645817e-05,
+      "loss": 1.1333,
+      "step": 2093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8497826703131938e-05,
+      "loss": 1.0801,
+      "step": 2094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8496183884854846e-05,
+      "loss": 1.1587,
+      "step": 2095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8494540241786385e-05,
+      "loss": 1.1392,
+      "step": 2096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492895774086107e-05,
+      "loss": 1.1636,
+      "step": 2097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491250481913656e-05,
+      "loss": 1.0737,
+      "step": 2098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489604365428754e-05,
+      "loss": 1.2444,
+      "step": 2099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8487957424791206e-05,
+      "loss": 1.0579,
+      "step": 2100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486309660160885e-05,
+      "loss": 1.1887,
+      "step": 2101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848466107169776e-05,
+      "loss": 1.2678,
+      "step": 2102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848301165956187e-05,
+      "loss": 1.0227,
+      "step": 2103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481361423913338e-05,
+      "loss": 0.978,
+      "step": 2104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847971036491236e-05,
+      "loss": 1.1913,
+      "step": 2105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478058482719222e-05,
+      "loss": 1.0635,
+      "step": 2106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476405777494285e-05,
+      "loss": 1.1335,
+      "step": 2107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8474752249397992e-05,
+      "loss": 1.1506,
+      "step": 2108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847309789859086e-05,
+      "loss": 1.1077,
+      "step": 2109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8471442725233488e-05,
+      "loss": 1.1082,
+      "step": 2110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8469786729486562e-05,
+      "loss": 1.1917,
+      "step": 2111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846812991151084e-05,
+      "loss": 1.1814,
+      "step": 2112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846647227146716e-05,
+      "loss": 1.2505,
+      "step": 2113
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8464813809516447e-05,
+      "loss": 1.0376,
+      "step": 2114
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84631545258197e-05,
+      "loss": 1.1465,
+      "step": 2115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461494420537997e-05,
+      "loss": 1.1143,
+      "step": 2116
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459833493832496e-05,
+      "loss": 1.1412,
+      "step": 2117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458171745864443e-05,
+      "loss": 1.1748,
+      "step": 2118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845650917679515e-05,
+      "loss": 1.2053,
+      "step": 2119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454845786786012e-05,
+      "loss": 1.0734,
+      "step": 2120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8453181575998518e-05,
+      "loss": 0.9763,
+      "step": 2121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845151654459422e-05,
+      "loss": 0.9971,
+      "step": 2122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844985069273476e-05,
+      "loss": 1.1772,
+      "step": 2123
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448184020581847e-05,
+      "loss": 1.0952,
+      "step": 2124
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446516528297286e-05,
+      "loss": 0.957,
+      "step": 2125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844484821604295e-05,
+      "loss": 1.1228,
+      "step": 2126
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84431790839808e-05,
+      "loss": 1.2041,
+      "step": 2127
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441509132272862e-05,
+      "loss": 1.1338,
+      "step": 2128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8439838361081258e-05,
+      "loss": 1.1067,
+      "step": 2129
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8438166770568182e-05,
+      "loss": 1.1831,
+      "step": 2130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843649436089591e-05,
+      "loss": 1.2207,
+      "step": 2131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8434821132226792e-05,
+      "loss": 1.1094,
+      "step": 2132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433147084723268e-05,
+      "loss": 1.0959,
+      "step": 2133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843147221854784e-05,
+      "loss": 1.2314,
+      "step": 2134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842979653386311e-05,
+      "loss": 1.2551,
+      "step": 2135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842812003083175e-05,
+      "loss": 1.1768,
+      "step": 2136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8426442709616507e-05,
+      "loss": 1.189,
+      "step": 2137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8424764570380215e-05,
+      "loss": 1.1367,
+      "step": 2138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423085613285775e-05,
+      "loss": 1.1973,
+      "step": 2139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842140583849619e-05,
+      "loss": 1.1238,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841972524617452e-05,
+      "loss": 1.1074,
+      "step": 2141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841804383648392e-05,
+      "loss": 1.1862,
+      "step": 2142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841636160958761e-05,
+      "loss": 1.069,
+      "step": 2143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414678565648905e-05,
+      "loss": 1.0208,
+      "step": 2144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841299470483118e-05,
+      "loss": 1.1099,
+      "step": 2145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411310027297915e-05,
+      "loss": 1.1572,
+      "step": 2146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8409624533212644e-05,
+      "loss": 1.1721,
+      "step": 2147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407938222738993e-05,
+      "loss": 1.0076,
+      "step": 2148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840625109604067e-05,
+      "loss": 1.2222,
+      "step": 2149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404563153281456e-05,
+      "loss": 1.1902,
+      "step": 2150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8402874394625208e-05,
+      "loss": 0.9568,
+      "step": 2151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401184820235873e-05,
+      "loss": 1.2329,
+      "step": 2152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839949443027747e-05,
+      "loss": 1.145,
+      "step": 2153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397803224914095e-05,
+      "loss": 1.1746,
+      "step": 2154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839611120430993e-05,
+      "loss": 1.2708,
+      "step": 2155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394418368629234e-05,
+      "loss": 1.1691,
+      "step": 2156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392724718036338e-05,
+      "loss": 1.1075,
+      "step": 2157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391030252695664e-05,
+      "loss": 1.0205,
+      "step": 2158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8389334972771705e-05,
+      "loss": 1.0935,
+      "step": 2159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387638878429034e-05,
+      "loss": 1.2031,
+      "step": 2160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8385941969832303e-05,
+      "loss": 1.1458,
+      "step": 2161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8384244247146253e-05,
+      "loss": 1.1233,
+      "step": 2162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838254571053568e-05,
+      "loss": 1.186,
+      "step": 2163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8380846360165488e-05,
+      "loss": 1.0099,
+      "step": 2164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379146196200645e-05,
+      "loss": 1.1992,
+      "step": 2165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8377445218806186e-05,
+      "loss": 1.0244,
+      "step": 2166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8375743428147255e-05,
+      "loss": 1.0676,
+      "step": 2167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837404082438905e-05,
+      "loss": 1.188,
+      "step": 2168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8372337407696858e-05,
+      "loss": 1.1951,
+      "step": 2169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8370633178236037e-05,
+      "loss": 1.0864,
+      "step": 2170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836892813617204e-05,
+      "loss": 1.1575,
+      "step": 2171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367222281670384e-05,
+      "loss": 1.0376,
+      "step": 2172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8365515614896666e-05,
+      "loss": 1.1289,
+      "step": 2173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836380813601657e-05,
+      "loss": 1.1787,
+      "step": 2174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362099845195857e-05,
+      "loss": 1.2078,
+      "step": 2175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8360390742600362e-05,
+      "loss": 1.0903,
+      "step": 2176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358680828395993e-05,
+      "loss": 1.01,
+      "step": 2177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356970102748758e-05,
+      "loss": 1.1211,
+      "step": 2178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835525856582472e-05,
+      "loss": 1.1211,
+      "step": 2179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835354621779004e-05,
+      "loss": 1.1465,
+      "step": 2180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835183305881094e-05,
+      "loss": 1.1257,
+      "step": 2181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8350119089053738e-05,
+      "loss": 1.1162,
+      "step": 2182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348404308684816e-05,
+      "loss": 0.9634,
+      "step": 2183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834668871787065e-05,
+      "loss": 1.1428,
+      "step": 2184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8344972316777774e-05,
+      "loss": 1.0247,
+      "step": 2185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343255105572822e-05,
+      "loss": 1.1404,
+      "step": 2186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8341537084422495e-05,
+      "loss": 1.0376,
+      "step": 2187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8339818253493572e-05,
+      "loss": 1.0774,
+      "step": 2188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338098612952913e-05,
+      "loss": 1.1562,
+      "step": 2189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8336378162967465e-05,
+      "loss": 1.2146,
+      "step": 2190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334656903704234e-05,
+      "loss": 1.1489,
+      "step": 2191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8332934835330325e-05,
+      "loss": 1.2598,
+      "step": 2192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833121195801291e-05,
+      "loss": 1.1021,
+      "step": 2193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832948827191924e-05,
+      "loss": 1.0764,
+      "step": 2194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327763777216653e-05,
+      "loss": 1.147,
+      "step": 2195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832603847407255e-05,
+      "loss": 1.0195,
+      "step": 2196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324312362654425e-05,
+      "loss": 1.0491,
+      "step": 2197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832258544312985e-05,
+      "loss": 1.1746,
+      "step": 2198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832085771566646e-05,
+      "loss": 1.0608,
+      "step": 2199
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831912918043199e-05,
+      "loss": 1.1858,
+      "step": 2200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831739983759423e-05,
+      "loss": 1.0504,
+      "step": 2201
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831566968732107e-05,
+      "loss": 1.1638,
+      "step": 2202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313938729780472e-05,
+      "loss": 1.0955,
+      "step": 2203
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8312206965140467e-05,
+      "loss": 0.9983,
+      "step": 2204
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310474393569172e-05,
+      "loss": 1.1702,
+      "step": 2205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308741015234784e-05,
+      "loss": 0.9871,
+      "step": 2206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307006830305574e-05,
+      "loss": 0.9702,
+      "step": 2207
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830527183894989e-05,
+      "loss": 1.1436,
+      "step": 2208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8303536041336167e-05,
+      "loss": 1.1042,
+      "step": 2209
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830179943763291e-05,
+      "loss": 1.0762,
+      "step": 2210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300062028008704e-05,
+      "loss": 1.2141,
+      "step": 2211
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298323812632212e-05,
+      "loss": 1.1399,
+      "step": 2212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829658479167218e-05,
+      "loss": 1.0928,
+      "step": 2213
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294844965297426e-05,
+      "loss": 1.1418,
+      "step": 2214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829310433367685e-05,
+      "loss": 1.1755,
+      "step": 2215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829136289697943e-05,
+      "loss": 1.0425,
+      "step": 2216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8289620655374213e-05,
+      "loss": 1.1704,
+      "step": 2217
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828787760903034e-05,
+      "loss": 1.0858,
+      "step": 2218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8286133758117017e-05,
+      "loss": 1.1277,
+      "step": 2219
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284389102803542e-05,
+      "loss": 1.2324,
+      "step": 2220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828264364325927e-05,
+      "loss": 1.1965,
+      "step": 2221
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280897379653657e-05,
+      "loss": 1.0665,
+      "step": 2222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279150312156225e-05,
+      "loss": 1.0591,
+      "step": 2223
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277402440936568e-05,
+      "loss": 1.1172,
+      "step": 2224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275653766164377e-05,
+      "loss": 1.0762,
+      "step": 2225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8273904288009402e-05,
+      "loss": 1.1292,
+      "step": 2226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272154006641475e-05,
+      "loss": 1.1531,
+      "step": 2227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827040292223052e-05,
+      "loss": 1.0554,
+      "step": 2228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8268651034946522e-05,
+      "loss": 1.0593,
+      "step": 2229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826689834495955e-05,
+      "loss": 1.0864,
+      "step": 2230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265144852439754e-05,
+      "loss": 1.1685,
+      "step": 2231
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8263390557557355e-05,
+      "loss": 1.1445,
+      "step": 2232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261635460482666e-05,
+      "loss": 1.0867,
+      "step": 2233
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8259879561386058e-05,
+      "loss": 1.2017,
+      "step": 2234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8258122860437993e-05,
+      "loss": 1.1952,
+      "step": 2235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256365357809015e-05,
+      "loss": 1.1682,
+      "step": 2236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254607053669728e-05,
+      "loss": 1.064,
+      "step": 2237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825284794819082e-05,
+      "loss": 1.1465,
+      "step": 2238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825108804154308e-05,
+      "loss": 1.1299,
+      "step": 2239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824932733389734e-05,
+      "loss": 1.0916,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824756582542454e-05,
+      "loss": 1.1123,
+      "step": 2241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245803516295667e-05,
+      "loss": 1.1382,
+      "step": 2242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824404040668181e-05,
+      "loss": 1.1208,
+      "step": 2243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242276496754128e-05,
+      "loss": 1.0809,
+      "step": 2244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8240511786683858e-05,
+      "loss": 1.1292,
+      "step": 2245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8238746276642317e-05,
+      "loss": 1.1653,
+      "step": 2246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823697996680089e-05,
+      "loss": 1.1272,
+      "step": 2247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8235212857331053e-05,
+      "loss": 1.1777,
+      "step": 2248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233444948404347e-05,
+      "loss": 1.2415,
+      "step": 2249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231676240192402e-05,
+      "loss": 0.9716,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822990673286692e-05,
+      "loss": 1.1218,
+      "step": 2251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822813642659968e-05,
+      "loss": 1.2668,
+      "step": 2252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226365321562536e-05,
+      "loss": 1.0818,
+      "step": 2253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822459341792743e-05,
+      "loss": 1.1797,
+      "step": 2254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822282071586637e-05,
+      "loss": 1.0222,
+      "step": 2255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8221047215551446e-05,
+      "loss": 1.1176,
+      "step": 2256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8219272917154828e-05,
+      "loss": 1.1178,
+      "step": 2257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217497820848758e-05,
+      "loss": 1.1301,
+      "step": 2258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821572192680556e-05,
+      "loss": 1.0354,
+      "step": 2259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821394523519764e-05,
+      "loss": 1.2485,
+      "step": 2260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8212167746197464e-05,
+      "loss": 1.198,
+      "step": 2261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8210389459977594e-05,
+      "loss": 1.0854,
+      "step": 2262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8208610376710664e-05,
+      "loss": 1.1018,
+      "step": 2263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206830496569378e-05,
+      "loss": 1.0254,
+      "step": 2264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8205049819726523e-05,
+      "loss": 1.1018,
+      "step": 2265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820326834635497e-05,
+      "loss": 1.1431,
+      "step": 2266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8201486076627656e-05,
+      "loss": 1.2637,
+      "step": 2267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81997030107176e-05,
+      "loss": 1.0105,
+      "step": 2268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197919148797894e-05,
+      "loss": 1.0051,
+      "step": 2269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8196134491041723e-05,
+      "loss": 0.9805,
+      "step": 2270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194349037622324e-05,
+      "loss": 1.1162,
+      "step": 2271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8192562788713037e-05,
+      "loss": 1.1765,
+      "step": 2272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819077574448726e-05,
+      "loss": 1.1761,
+      "step": 2273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8188987905118475e-05,
+      "loss": 0.9863,
+      "step": 2274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818719927078025e-05,
+      "loss": 1.0977,
+      "step": 2275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8185409841646212e-05,
+      "loss": 1.0642,
+      "step": 2276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818361961789008e-05,
+      "loss": 1.1658,
+      "step": 2277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818182859968564e-05,
+      "loss": 1.2332,
+      "step": 2278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818003678720677e-05,
+      "loss": 1.1504,
+      "step": 2279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178244180627406e-05,
+      "loss": 1.1072,
+      "step": 2280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8176450780121574e-05,
+      "loss": 1.0422,
+      "step": 2281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8174656585863376e-05,
+      "loss": 1.0887,
+      "step": 2282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172861598026986e-05,
+      "loss": 1.2263,
+      "step": 2283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817106581678665e-05,
+      "loss": 1.1702,
+      "step": 2284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816926924231672e-05,
+      "loss": 1.2512,
+      "step": 2285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167471874791578e-05,
+      "loss": 1.1165,
+      "step": 2286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165673714385725e-05,
+      "loss": 1.1616,
+      "step": 2287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163874761273717e-05,
+      "loss": 1.1157,
+      "step": 2288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8162075015630192e-05,
+      "loss": 1.1295,
+      "step": 2289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816027447762987e-05,
+      "loss": 1.076,
+      "step": 2290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158473147447537e-05,
+      "loss": 1.1375,
+      "step": 2291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8156671025258068e-05,
+      "loss": 1.2358,
+      "step": 2292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154868111236406e-05,
+      "loss": 1.1816,
+      "step": 2293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8153064405557575e-05,
+      "loss": 1.118,
+      "step": 2294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151259908396674e-05,
+      "loss": 1.1948,
+      "step": 2295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814945461992888e-05,
+      "loss": 1.0184,
+      "step": 2296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814764854032945e-05,
+      "loss": 1.2537,
+      "step": 2297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8145841669773706e-05,
+      "loss": 1.0599,
+      "step": 2298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8144034008437062e-05,
+      "loss": 1.1112,
+      "step": 2299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8142225556495004e-05,
+      "loss": 1.2349,
+      "step": 2300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8140416314123087e-05,
+      "loss": 1.145,
+      "step": 2301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8138606281496952e-05,
+      "loss": 1.1548,
+      "step": 2302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136795458792308e-05,
+      "loss": 1.2689,
+      "step": 2303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813498384618495e-05,
+      "loss": 1.1216,
+      "step": 2304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133171443850745e-05,
+      "loss": 1.1985,
+      "step": 2305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813135825196564e-05,
+      "loss": 1.2026,
+      "step": 2306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812954427070565e-05,
+      "loss": 1.0479,
+      "step": 2307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8127729500246873e-05,
+      "loss": 1.1062,
+      "step": 2308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812591394076549e-05,
+      "loss": 1.0779,
+      "step": 2309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812409759243774e-05,
+      "loss": 1.0376,
+      "step": 2310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8122280455439963e-05,
+      "loss": 1.0917,
+      "step": 2311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8120462529948555e-05,
+      "loss": 1.0669,
+      "step": 2312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8118643816139996e-05,
+      "loss": 1.0972,
+      "step": 2313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116824314190847e-05,
+      "loss": 1.1675,
+      "step": 2314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811500402427774e-05,
+      "loss": 1.111,
+      "step": 2315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8113182946577383e-05,
+      "loss": 1.0332,
+      "step": 2316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111361081266564e-05,
+      "loss": 1.2023,
+      "step": 2317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8109538428522144e-05,
+      "loss": 1.0659,
+      "step": 2318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107714988521064e-05,
+      "loss": 1.1414,
+      "step": 2319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810589076144034e-05,
+      "loss": 1.1274,
+      "step": 2320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810406574745706e-05,
+      "loss": 1.2463,
+      "step": 2321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81022399467484e-05,
+      "loss": 1.2234,
+      "step": 2322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81004133594916e-05,
+      "loss": 1.1458,
+      "step": 2323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809858598586398e-05,
+      "loss": 1.0432,
+      "step": 2324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096757826042944e-05,
+      "loss": 1.1235,
+      "step": 2325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094928880205954e-05,
+      "loss": 1.1128,
+      "step": 2326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809309914853057e-05,
+      "loss": 1.0242,
+      "step": 2327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8091268631194417e-05,
+      "loss": 1.0552,
+      "step": 2328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089437328375193e-05,
+      "loss": 1.1003,
+      "step": 2329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8087605240250683e-05,
+      "loss": 1.0662,
+      "step": 2330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808577236699874e-05,
+      "loss": 1.1909,
+      "step": 2331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083938708797294e-05,
+      "loss": 1.11,
+      "step": 2332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8082104265824355e-05,
+      "loss": 1.1104,
+      "step": 2333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8080269038258004e-05,
+      "loss": 1.1284,
+      "step": 2334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078433026276402e-05,
+      "loss": 1.1318,
+      "step": 2335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807659623005779e-05,
+      "loss": 1.1084,
+      "step": 2336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807475864978047e-05,
+      "loss": 1.194,
+      "step": 2337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8072920285622836e-05,
+      "loss": 1.0581,
+      "step": 2338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8071081137763355e-05,
+      "loss": 1.0908,
+      "step": 2339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069241206380565e-05,
+      "loss": 1.2441,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806740049165308e-05,
+      "loss": 1.0286,
+      "step": 2341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80655589937596e-05,
+      "loss": 1.0771,
+      "step": 2342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8063716712878885e-05,
+      "loss": 1.2026,
+      "step": 2343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061873649189788e-05,
+      "loss": 1.0981,
+      "step": 2344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806002980287122e-05,
+      "loss": 1.2642,
+      "step": 2345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058185174102186e-05,
+      "loss": 1.0415,
+      "step": 2346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056339763061755e-05,
+      "loss": 1.1294,
+      "step": 2347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054493569929075e-05,
+      "loss": 1.2007,
+      "step": 2348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8052646594883375e-05,
+      "loss": 0.9423,
+      "step": 2349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8050798838103954e-05,
+      "loss": 1.1194,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804895029977018e-05,
+      "loss": 1.0601,
+      "step": 2351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8047100980061517e-05,
+      "loss": 1.1912,
+      "step": 2352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045250879157483e-05,
+      "loss": 1.0912,
+      "step": 2353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804339999723769e-05,
+      "loss": 1.1902,
+      "step": 2354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041548334481813e-05,
+      "loss": 1.1562,
+      "step": 2355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8039695891069606e-05,
+      "loss": 1.1487,
+      "step": 2356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037842667180907e-05,
+      "loss": 1.1123,
+      "step": 2357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8035988662995617e-05,
+      "loss": 1.1166,
+      "step": 2358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803413387869372e-05,
+      "loss": 1.0217,
+      "step": 2359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8032278314455273e-05,
+      "loss": 1.1648,
+      "step": 2360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030421970460417e-05,
+      "loss": 1.1267,
+      "step": 2361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8028564846889355e-05,
+      "loss": 1.2495,
+      "step": 2362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8026706943922374e-05,
+      "loss": 1.2166,
+      "step": 2363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8024848261739837e-05,
+      "loss": 1.135,
+      "step": 2364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802298880052218e-05,
+      "loss": 1.0339,
+      "step": 2365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8021128560449913e-05,
+      "loss": 1.0109,
+      "step": 2366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801926754170363e-05,
+      "loss": 1.1719,
+      "step": 2367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8017405744463988e-05,
+      "loss": 1.0767,
+      "step": 2368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8015543168911734e-05,
+      "loss": 1.0613,
+      "step": 2369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013679815227676e-05,
+      "loss": 1.0101,
+      "step": 2370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011815683592706e-05,
+      "loss": 1.042,
+      "step": 2371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800995077418779e-05,
+      "loss": 1.1938,
+      "step": 2372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008085087193975e-05,
+      "loss": 1.0649,
+      "step": 2373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8006218622792365e-05,
+      "loss": 1.2305,
+      "step": 2374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8004351381164168e-05,
+      "loss": 1.2043,
+      "step": 2375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8002483362490643e-05,
+      "loss": 1.1875,
+      "step": 2376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800061456695313e-05,
+      "loss": 1.1414,
+      "step": 2377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7998744994733055e-05,
+      "loss": 1.165,
+      "step": 2378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799687464601191e-05,
+      "loss": 1.1531,
+      "step": 2379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7995003520971264e-05,
+      "loss": 1.0662,
+      "step": 2380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799313161979276e-05,
+      "loss": 1.145,
+      "step": 2381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7991258942658122e-05,
+      "loss": 1.1276,
+      "step": 2382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798938548974914e-05,
+      "loss": 1.1311,
+      "step": 2383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798751126124769e-05,
+      "loss": 1.0392,
+      "step": 2384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7985636257335716e-05,
+      "loss": 1.1086,
+      "step": 2385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7983760478195238e-05,
+      "loss": 1.1506,
+      "step": 2386
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981883924008355e-05,
+      "loss": 1.1714,
+      "step": 2387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798000659495724e-05,
+      "loss": 1.2183,
+      "step": 2388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7978128491224137e-05,
+      "loss": 1.1313,
+      "step": 2389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7976249612991367e-05,
+      "loss": 1.2542,
+      "step": 2390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974369960441335e-05,
+      "loss": 1.2661,
+      "step": 2391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79724895337565e-05,
+      "loss": 1.147,
+      "step": 2392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797060833311943e-05,
+      "loss": 1.3125,
+      "step": 2393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7968726358712728e-05,
+      "loss": 1.0541,
+      "step": 2394
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7966843610719103e-05,
+      "loss": 1.0889,
+      "step": 2395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7964960089321324e-05,
+      "loss": 1.0323,
+      "step": 2396
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796307579470224e-05,
+      "loss": 1.1935,
+      "step": 2397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7961190727044777e-05,
+      "loss": 1.1594,
+      "step": 2398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795930488653193e-05,
+      "loss": 1.0955,
+      "step": 2399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7957418273346773e-05,
+      "loss": 1.0635,
+      "step": 2400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795553088767246e-05,
+      "loss": 1.1331,
+      "step": 2401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7953642729692204e-05,
+      "loss": 1.0332,
+      "step": 2402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7951753799589312e-05,
+      "loss": 1.0244,
+      "step": 2403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7949864097547154e-05,
+      "loss": 1.1016,
+      "step": 2404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7947973623749178e-05,
+      "loss": 1.1465,
+      "step": 2405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7946082378378906e-05,
+      "loss": 1.1938,
+      "step": 2406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944190361619937e-05,
+      "loss": 1.1702,
+      "step": 2407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7942297573655948e-05,
+      "loss": 1.059,
+      "step": 2408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7940404014670678e-05,
+      "loss": 0.9944,
+      "step": 2409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793850968484796e-05,
+      "loss": 1.1902,
+      "step": 2410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7936614584371685e-05,
+      "loss": 1.0109,
+      "step": 2411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934718713425825e-05,
+      "loss": 0.9812,
+      "step": 2412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793282207219443e-05,
+      "loss": 1.2087,
+      "step": 2413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7930924660861618e-05,
+      "loss": 1.1404,
+      "step": 2414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792902647961159e-05,
+      "loss": 1.1458,
+      "step": 2415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792712752862862e-05,
+      "loss": 1.2266,
+      "step": 2416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925227808097048e-05,
+      "loss": 1.2078,
+      "step": 2417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7923327318201295e-05,
+      "loss": 1.1155,
+      "step": 2418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7921426059125856e-05,
+      "loss": 1.0892,
+      "step": 2419
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7919524031055307e-05,
+      "loss": 1.1111,
+      "step": 2420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791762123417429e-05,
+      "loss": 1.2056,
+      "step": 2421
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791571766866752e-05,
+      "loss": 1.1169,
+      "step": 2422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7913813334719797e-05,
+      "loss": 1.2495,
+      "step": 2423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7911908232515988e-05,
+      "loss": 1.1365,
+      "step": 2424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7910002362241035e-05,
+      "loss": 1.0913,
+      "step": 2425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7908095724079954e-05,
+      "loss": 1.0889,
+      "step": 2426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7906188318217842e-05,
+      "loss": 1.24,
+      "step": 2427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7904280144839862e-05,
+      "loss": 1.1007,
+      "step": 2428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7902371204131258e-05,
+      "loss": 1.0853,
+      "step": 2429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7900461496277345e-05,
+      "loss": 1.158,
+      "step": 2430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7898551021463514e-05,
+      "loss": 1.0935,
+      "step": 2431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7896639779875226e-05,
+      "loss": 1.1443,
+      "step": 2432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7894727771698027e-05,
+      "loss": 1.2429,
+      "step": 2433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7892814997117527e-05,
+      "loss": 1.1221,
+      "step": 2434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789090145631941e-05,
+      "loss": 1.1885,
+      "step": 2435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788898714948945e-05,
+      "loss": 1.1589,
+      "step": 2436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788707207681347e-05,
+      "loss": 1.1042,
+      "step": 2437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788515623847739e-05,
+      "loss": 1.0803,
+      "step": 2438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788323963466719e-05,
+      "loss": 1.1182,
+      "step": 2439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881322265568933e-05,
+      "loss": 1.0109,
+      "step": 2440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879404131368757e-05,
+      "loss": 1.1091,
+      "step": 2441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7877485232252865e-05,
+      "loss": 1.1631,
+      "step": 2442
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787556556840754e-05,
+      "loss": 1.2397,
+      "step": 2443
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787364514001914e-05,
+      "loss": 1.0703,
+      "step": 2444
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871723947274096e-05,
+      "loss": 1.0942,
+      "step": 2445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7869801990358913e-05,
+      "loss": 1.1875,
+      "step": 2446
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786787926946017e-05,
+      "loss": 1.2109,
+      "step": 2447
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865955784764525e-05,
+      "loss": 1.1333,
+      "step": 2448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7864031536458693e-05,
+      "loss": 1.1125,
+      "step": 2449
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7862106524729494e-05,
+      "loss": 1.137,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7860180749763787e-05,
+      "loss": 1.1655,
+      "step": 2451
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785825421174854e-05,
+      "loss": 1.0703,
+      "step": 2452
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785632691087076e-05,
+      "loss": 1.0782,
+      "step": 2453
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7854398847317555e-05,
+      "loss": 1.1838,
+      "step": 2454
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852470021276092e-05,
+      "loss": 1.2224,
+      "step": 2455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7850540432933624e-05,
+      "loss": 1.1121,
+      "step": 2456
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7848610082477463e-05,
+      "loss": 1.022,
+      "step": 2457
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7846678970095013e-05,
+      "loss": 1.1093,
+      "step": 2458
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844747095973736e-05,
+      "loss": 1.1548,
+      "step": 2459
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7842814460301173e-05,
+      "loss": 1.0427,
+      "step": 2460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7840881063264944e-05,
+      "loss": 1.0204,
+      "step": 2461
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7838946905052738e-05,
+      "loss": 1.0039,
+      "step": 2462
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837011985852315e-05,
+      "loss": 1.0872,
+      "step": 2463
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835076305851518e-05,
+      "loss": 1.176,
+      "step": 2464
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833139865238257e-05,
+      "loss": 1.0403,
+      "step": 2465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831202664200518e-05,
+      "loss": 1.1838,
+      "step": 2466
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7829264702926358e-05,
+      "loss": 1.2437,
+      "step": 2467
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827325981603914e-05,
+      "loss": 1.1104,
+      "step": 2468
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7825386500421386e-05,
+      "loss": 1.0803,
+      "step": 2469
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7823446259567062e-05,
+      "loss": 1.0637,
+      "step": 2470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78215052592293e-05,
+      "loss": 1.0966,
+      "step": 2471
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7819563499596516e-05,
+      "loss": 1.176,
+      "step": 2472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781762098085722e-05,
+      "loss": 1.2175,
+      "step": 2473
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781567770319998e-05,
+      "loss": 1.1522,
+      "step": 2474
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7813733666813456e-05,
+      "loss": 1.144,
+      "step": 2475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811788871886366e-05,
+      "loss": 1.1573,
+      "step": 2476
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809843318607504e-05,
+      "loss": 1.1121,
+      "step": 2477
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780789700716574e-05,
+      "loss": 1.2039,
+      "step": 2478
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805949937750026e-05,
+      "loss": 1.0396,
+      "step": 2479
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780400211054937e-05,
+      "loss": 1.0681,
+      "step": 2480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780205352575287e-05,
+      "loss": 1.1084,
+      "step": 2481
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780010418354968e-05,
+      "loss": 1.141,
+      "step": 2482
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779815408412905e-05,
+      "loss": 1.0767,
+      "step": 2483
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796203227680283e-05,
+      "loss": 0.9744,
+      "step": 2484
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779425161439277e-05,
+      "loss": 1.011,
+      "step": 2485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7792299244455965e-05,
+      "loss": 1.1748,
+      "step": 2486
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7790346118059402e-05,
+      "loss": 1.1294,
+      "step": 2487
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7788392235392688e-05,
+      "loss": 1.176,
+      "step": 2488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786437596645496e-05,
+      "loss": 1.0371,
+      "step": 2489
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7784482202007583e-05,
+      "loss": 1.187,
+      "step": 2490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782526051668775e-05,
+      "loss": 1.123,
+      "step": 2491
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7780569145818967e-05,
+      "loss": 1.2288,
+      "step": 2492
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7778611484648135e-05,
+      "loss": 1.1287,
+      "step": 2493
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7776653068346323e-05,
+      "loss": 1.1943,
+      "step": 2494
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777469389710365e-05,
+      "loss": 1.0728,
+      "step": 2495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7772733971110305e-05,
+      "loss": 1.0815,
+      "step": 2496
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777077329055656e-05,
+      "loss": 1.1666,
+      "step": 2497
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776881185563275e-05,
+      "loss": 1.1204,
+      "step": 2498
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7766849666529286e-05,
+      "loss": 1.0814,
+      "step": 2499
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7764886723436652e-05,
+      "loss": 1.0935,
+      "step": 2500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776292302654541e-05,
+      "loss": 1.0905,
+      "step": 2501
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7760958576046192e-05,
+      "loss": 1.1775,
+      "step": 2502
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77589933721297e-05,
+      "loss": 1.1453,
+      "step": 2503
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7757027414986713e-05,
+      "loss": 0.8889,
+      "step": 2504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775506070480808e-05,
+      "loss": 1.1648,
+      "step": 2505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7753093241784727e-05,
+      "loss": 1.1128,
+      "step": 2506
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7751125026107648e-05,
+      "loss": 1.1245,
+      "step": 2507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774915605796792e-05,
+      "loss": 0.9955,
+      "step": 2508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7747186337556677e-05,
+      "loss": 1.1382,
+      "step": 2509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7745215865065146e-05,
+      "loss": 1.1293,
+      "step": 2510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743244640684605e-05,
+      "loss": 1.0977,
+      "step": 2511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7741272664606424e-05,
+      "loss": 1.1333,
+      "step": 2512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739299937022038e-05,
+      "loss": 1.2344,
+      "step": 2513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7737326458122948e-05,
+      "loss": 1.0436,
+      "step": 2514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7735352228100743e-05,
+      "loss": 1.104,
+      "step": 2515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7733377247147076e-05,
+      "loss": 1.1477,
+      "step": 2516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731401515453667e-05,
+      "loss": 1.2222,
+      "step": 2517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729425033212326e-05,
+      "loss": 1.1663,
+      "step": 2518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772744780061492e-05,
+      "loss": 1.1667,
+      "step": 2519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7725469817853394e-05,
+      "loss": 1.135,
+      "step": 2520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7723491085119767e-05,
+      "loss": 1.1489,
+      "step": 2521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772151160260613e-05,
+      "loss": 1.1616,
+      "step": 2522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771953137050465e-05,
+      "loss": 1.1149,
+      "step": 2523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771755038900756e-05,
+      "loss": 1.1326,
+      "step": 2524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771556865830717e-05,
+      "loss": 0.9323,
+      "step": 2525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713586178595865e-05,
+      "loss": 1.1194,
+      "step": 2526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7711602950066096e-05,
+      "loss": 1.2729,
+      "step": 2527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770961897291039e-05,
+      "loss": 1.1885,
+      "step": 2528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707634247321355e-05,
+      "loss": 1.1184,
+      "step": 2529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705648773491653e-05,
+      "loss": 1.087,
+      "step": 2530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770366255161404e-05,
+      "loss": 1.1862,
+      "step": 2531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7701675581881326e-05,
+      "loss": 1.1281,
+      "step": 2532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7699687864486407e-05,
+      "loss": 1.1765,
+      "step": 2533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7697699399622242e-05,
+      "loss": 1.033,
+      "step": 2534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769571018748187e-05,
+      "loss": 1.1367,
+      "step": 2535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7693720228258402e-05,
+      "loss": 1.0377,
+      "step": 2536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7691729522145012e-05,
+      "loss": 1.1487,
+      "step": 2537
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768973806933496e-05,
+      "loss": 1.1875,
+      "step": 2538
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7687745870021567e-05,
+      "loss": 0.9938,
+      "step": 2539
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7685752924398234e-05,
+      "loss": 1.0149,
+      "step": 2540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7683759232658432e-05,
+      "loss": 1.1555,
+      "step": 2541
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7681764794995704e-05,
+      "loss": 1.077,
+      "step": 2542
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679769611603665e-05,
+      "loss": 1.1614,
+      "step": 2543
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7677773682676007e-05,
+      "loss": 1.0713,
+      "step": 2544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7675777008406485e-05,
+      "loss": 1.1477,
+      "step": 2545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7673779588988937e-05,
+      "loss": 1.0325,
+      "step": 2546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767178142461726e-05,
+      "loss": 1.0283,
+      "step": 2547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766978251548544e-05,
+      "loss": 1.0895,
+      "step": 2548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7667782861787524e-05,
+      "loss": 1.0891,
+      "step": 2549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766578246371763e-05,
+      "loss": 1.0719,
+      "step": 2550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663781321469962e-05,
+      "loss": 1.2117,
+      "step": 2551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7661779435238776e-05,
+      "loss": 1.1123,
+      "step": 2552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765977680521842e-05,
+      "loss": 1.1506,
+      "step": 2553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76577734316033e-05,
+      "loss": 0.9597,
+      "step": 2554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7655769314587897e-05,
+      "loss": 1.197,
+      "step": 2555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7653764454366774e-05,
+      "loss": 1.177,
+      "step": 2556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765175885113455e-05,
+      "loss": 1.1379,
+      "step": 2557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764975250508593e-05,
+      "loss": 1.2085,
+      "step": 2558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764774541641568e-05,
+      "loss": 1.0242,
+      "step": 2559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645737585318653e-05,
+      "loss": 1.0115,
+      "step": 2560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7643729011989763e-05,
+      "loss": 1.0715,
+      "step": 2561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764171969662399e-05,
+      "loss": 1.1707,
+      "step": 2562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76397096394164e-05,
+      "loss": 1.1589,
+      "step": 2563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7637698840562127e-05,
+      "loss": 1.1611,
+      "step": 2564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635687300256374e-05,
+      "loss": 1.2048,
+      "step": 2565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763367501869441e-05,
+      "loss": 1.0393,
+      "step": 2566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763166199607159e-05,
+      "loss": 1.0295,
+      "step": 2567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7629648232583334e-05,
+      "loss": 1.1886,
+      "step": 2568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762763372842513e-05,
+      "loss": 1.2466,
+      "step": 2569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7625618483792543e-05,
+      "loss": 1.0563,
+      "step": 2570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7623602498881217e-05,
+      "loss": 1.1687,
+      "step": 2571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7621585773886844e-05,
+      "loss": 1.1987,
+      "step": 2572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619568309005214e-05,
+      "loss": 1.0271,
+      "step": 2573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7617550104432175e-05,
+      "loss": 1.0811,
+      "step": 2574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761553116036365e-05,
+      "loss": 0.9854,
+      "step": 2575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7613511476995637e-05,
+      "loss": 1.0369,
+      "step": 2576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76114910545242e-05,
+      "loss": 1.1343,
+      "step": 2577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7609469893145475e-05,
+      "loss": 1.0317,
+      "step": 2578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607447993055676e-05,
+      "loss": 1.0808,
+      "step": 2579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7605425354451085e-05,
+      "loss": 1.083,
+      "step": 2580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760340197752805e-05,
+      "loss": 1.115,
+      "step": 2581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7601377862483e-05,
+      "loss": 0.9875,
+      "step": 2582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7599353009512432e-05,
+      "loss": 1.0806,
+      "step": 2583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7597327418812912e-05,
+      "loss": 1.092,
+      "step": 2584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595301090581083e-05,
+      "loss": 1.032,
+      "step": 2585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7593274025013657e-05,
+      "loss": 1.0951,
+      "step": 2586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7591246222307413e-05,
+      "loss": 1.0872,
+      "step": 2587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7589217682659208e-05,
+      "loss": 1.0756,
+      "step": 2588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758718840626597e-05,
+      "loss": 1.0636,
+      "step": 2589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7585158393324696e-05,
+      "loss": 1.1912,
+      "step": 2590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7583127644032453e-05,
+      "loss": 1.0012,
+      "step": 2591
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7581096158586386e-05,
+      "loss": 1.0867,
+      "step": 2592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7579063937183702e-05,
+      "loss": 1.1562,
+      "step": 2593
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757703098002169e-05,
+      "loss": 1.1227,
+      "step": 2594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75749972872977e-05,
+      "loss": 1.0027,
+      "step": 2595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7572962859209163e-05,
+      "loss": 1.1084,
+      "step": 2596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7570927695953574e-05,
+      "loss": 1.147,
+      "step": 2597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7568891797728504e-05,
+      "loss": 1.2007,
+      "step": 2598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7566855164731594e-05,
+      "loss": 1.2473,
+      "step": 2599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7564817797160553e-05,
+      "loss": 1.2075,
+      "step": 2600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7562779695213167e-05,
+      "loss": 1.1448,
+      "step": 2601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7560740859087292e-05,
+      "loss": 0.9698,
+      "step": 2602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755870128898085e-05,
+      "loss": 0.9198,
+      "step": 2603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755666098509184e-05,
+      "loss": 1.2566,
+      "step": 2604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7554619947618328e-05,
+      "loss": 1.1562,
+      "step": 2605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755257817675846e-05,
+      "loss": 1.1467,
+      "step": 2606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755053567271044e-05,
+      "loss": 1.114,
+      "step": 2607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7548492435672553e-05,
+      "loss": 1.094,
+      "step": 2608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754644846584315e-05,
+      "loss": 1.1199,
+      "step": 2609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7544403763420657e-05,
+      "loss": 1.1008,
+      "step": 2610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542358328603574e-05,
+      "loss": 1.0793,
+      "step": 2611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754031216159046e-05,
+      "loss": 1.2332,
+      "step": 2612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753826526257995e-05,
+      "loss": 1.0098,
+      "step": 2613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7536217631770762e-05,
+      "loss": 1.1724,
+      "step": 2614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7534169269361672e-05,
+      "loss": 1.0206,
+      "step": 2615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7532120175551528e-05,
+      "loss": 1.1278,
+      "step": 2616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7530070350539254e-05,
+      "loss": 1.1248,
+      "step": 2617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528019794523844e-05,
+      "loss": 1.2507,
+      "step": 2618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7525968507704356e-05,
+      "loss": 1.1985,
+      "step": 2619
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752391649027993e-05,
+      "loss": 1.1378,
+      "step": 2620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7521863742449774e-05,
+      "loss": 1.1997,
+      "step": 2621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7519810264413154e-05,
+      "loss": 1.1642,
+      "step": 2622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751775605636943e-05,
+      "loss": 1.156,
+      "step": 2623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7515701118518004e-05,
+      "loss": 1.1802,
+      "step": 2624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751364545105838e-05,
+      "loss": 1.032,
+      "step": 2625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7511589054190112e-05,
+      "loss": 1.1019,
+      "step": 2626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7509531928112832e-05,
+      "loss": 1.1799,
+      "step": 2627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750747407302624e-05,
+      "loss": 1.0662,
+      "step": 2628
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750541548913011e-05,
+      "loss": 1.0994,
+      "step": 2629
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750335617662428e-05,
+      "loss": 1.1167,
+      "step": 2630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7501296135708668e-05,
+      "loss": 1.052,
+      "step": 2631
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7499235366583258e-05,
+      "loss": 1.2112,
+      "step": 2632
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7497173869448104e-05,
+      "loss": 1.1052,
+      "step": 2633
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7495111644503336e-05,
+      "loss": 1.0658,
+      "step": 2634
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749304869194914e-05,
+      "loss": 0.9666,
+      "step": 2635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7490985011985795e-05,
+      "loss": 1.1792,
+      "step": 2636
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7488920604813633e-05,
+      "loss": 1.1995,
+      "step": 2637
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7486855470633058e-05,
+      "loss": 1.0153,
+      "step": 2638
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484789609644556e-05,
+      "loss": 1.0686,
+      "step": 2639
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7482723022048675e-05,
+      "loss": 1.1033,
+      "step": 2640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7480655708046032e-05,
+      "loss": 1.2197,
+      "step": 2641
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747858766783732e-05,
+      "loss": 1.1506,
+      "step": 2642
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74765189016233e-05,
+      "loss": 1.0295,
+      "step": 2643
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7474449409604802e-05,
+      "loss": 1.2166,
+      "step": 2644
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747237919198273e-05,
+      "loss": 1.1265,
+      "step": 2645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747030824895805e-05,
+      "loss": 1.064,
+      "step": 2646
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746823658073181e-05,
+      "loss": 1.1696,
+      "step": 2647
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7466164187505126e-05,
+      "loss": 1.2278,
+      "step": 2648
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7464091069479175e-05,
+      "loss": 1.0559,
+      "step": 2649
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7462017226855213e-05,
+      "loss": 1.1467,
+      "step": 2650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7459942659834566e-05,
+      "loss": 1.1296,
+      "step": 2651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745786736861863e-05,
+      "loss": 1.1758,
+      "step": 2652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7455791353408866e-05,
+      "loss": 1.217,
+      "step": 2653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745371461440681e-05,
+      "loss": 1.0325,
+      "step": 2654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745163715181407e-05,
+      "loss": 1.011,
+      "step": 2655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7449558965832315e-05,
+      "loss": 1.1072,
+      "step": 2656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7447480056663296e-05,
+      "loss": 1.1371,
+      "step": 2657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744540042450883e-05,
+      "loss": 1.1011,
+      "step": 2658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7443320069570805e-05,
+      "loss": 1.1609,
+      "step": 2659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7441238992051167e-05,
+      "loss": 1.1924,
+      "step": 2660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7439157192151958e-05,
+      "loss": 1.1414,
+      "step": 2661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743707467007526e-05,
+      "loss": 1.1675,
+      "step": 2662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7434991426023248e-05,
+      "loss": 1.1492,
+      "step": 2663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7432907460198157e-05,
+      "loss": 1.1288,
+      "step": 2664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7430822772802293e-05,
+      "loss": 1.261,
+      "step": 2665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428737364038036e-05,
+      "loss": 1.1248,
+      "step": 2666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742665123410783e-05,
+      "loss": 1.2546,
+      "step": 2667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742456438321419e-05,
+      "loss": 1.0702,
+      "step": 2668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742247681155971e-05,
+      "loss": 1.1753,
+      "step": 2669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742038851934704e-05,
+      "loss": 1.0461,
+      "step": 2670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7418299506778914e-05,
+      "loss": 1.1399,
+      "step": 2671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7416209774058124e-05,
+      "loss": 1.0176,
+      "step": 2672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7414119321387538e-05,
+      "loss": 1.1213,
+      "step": 2673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741202814897009e-05,
+      "loss": 1.0413,
+      "step": 2674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740993625700879e-05,
+      "loss": 1.1921,
+      "step": 2675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740784364570671e-05,
+      "loss": 1.1711,
+      "step": 2676
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7405750315267003e-05,
+      "loss": 0.9497,
+      "step": 2677
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7403656265892882e-05,
+      "loss": 1.0709,
+      "step": 2678
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401561497787632e-05,
+      "loss": 1.1494,
+      "step": 2679
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399466011154603e-05,
+      "loss": 1.1216,
+      "step": 2680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7397369806197234e-05,
+      "loss": 1.1956,
+      "step": 2681
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739527288311901e-05,
+      "loss": 1.1523,
+      "step": 2682
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73931752421235e-05,
+      "loss": 1.1057,
+      "step": 2683
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739107688341433e-05,
+      "loss": 1.0227,
+      "step": 2684
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738897780719522e-05,
+      "loss": 1.0889,
+      "step": 2685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738687801366993e-05,
+      "loss": 1.26,
+      "step": 2686
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738477750304231e-05,
+      "loss": 1.1506,
+      "step": 2687
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7382676275516267e-05,
+      "loss": 1.1426,
+      "step": 2688
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7380574331295792e-05,
+      "loss": 1.0793,
+      "step": 2689
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7378471670584933e-05,
+      "loss": 1.1262,
+      "step": 2690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737636829358781e-05,
+      "loss": 1.1298,
+      "step": 2691
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7374264200508618e-05,
+      "loss": 1.1299,
+      "step": 2692
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7372159391551616e-05,
+      "loss": 1.2061,
+      "step": 2693
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7370053866921135e-05,
+      "loss": 1.168,
+      "step": 2694
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7367947626821574e-05,
+      "loss": 1.0791,
+      "step": 2695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7365840671457405e-05,
+      "loss": 1.0923,
+      "step": 2696
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736373300103316e-05,
+      "loss": 0.9774,
+      "step": 2697
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736162461575346e-05,
+      "loss": 1.1904,
+      "step": 2698
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7359515515822963e-05,
+      "loss": 1.0399,
+      "step": 2699
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7357405701446433e-05,
+      "loss": 1.2002,
+      "step": 2700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735529517282868e-05,
+      "loss": 1.1514,
+      "step": 2701
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735318393017459e-05,
+      "loss": 1.1965,
+      "step": 2702
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7351071973689118e-05,
+      "loss": 1.0618,
+      "step": 2703
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7348959303577287e-05,
+      "loss": 0.9915,
+      "step": 2704
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734684592004419e-05,
+      "loss": 1.116,
+      "step": 2705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7344731823294996e-05,
+      "loss": 1.1716,
+      "step": 2706
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734261701353493e-05,
+      "loss": 1.1821,
+      "step": 2707
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340501490969296e-05,
+      "loss": 1.1417,
+      "step": 2708
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733838525580346e-05,
+      "loss": 1.124,
+      "step": 2709
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733626830824287e-05,
+      "loss": 0.9874,
+      "step": 2710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733415064849303e-05,
+      "loss": 1.1731,
+      "step": 2711
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7332032276759515e-05,
+      "loss": 1.0933,
+      "step": 2712
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732991319324798e-05,
+      "loss": 1.2771,
+      "step": 2713
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732779339816413e-05,
+      "loss": 1.0376,
+      "step": 2714
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732567289171376e-05,
+      "loss": 1.0059,
+      "step": 2715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7323551674102725e-05,
+      "loss": 1.0818,
+      "step": 2716
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732142974553694e-05,
+      "loss": 1.0725,
+      "step": 2717
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7319307106222402e-05,
+      "loss": 1.1733,
+      "step": 2718
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731718375636517e-05,
+      "loss": 1.0406,
+      "step": 2719
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7315059696171376e-05,
+      "loss": 1.187,
+      "step": 2720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731293492584722e-05,
+      "loss": 1.2336,
+      "step": 2721
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731080944559897e-05,
+      "loss": 1.1907,
+      "step": 2722
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730868325563296e-05,
+      "loss": 1.1124,
+      "step": 2723
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73065563561556e-05,
+      "loss": 1.1528,
+      "step": 2724
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730442874737336e-05,
+      "loss": 1.0955,
+      "step": 2725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730230042949279e-05,
+      "loss": 1.1843,
+      "step": 2726
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7300171402720498e-05,
+      "loss": 1.2439,
+      "step": 2727
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729804166726317e-05,
+      "loss": 1.1958,
+      "step": 2728
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729591122332755e-05,
+      "loss": 1.0847,
+      "step": 2729
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729378007112046e-05,
+      "loss": 1.0691,
+      "step": 2730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7291648210848785e-05,
+      "loss": 1.0891,
+      "step": 2731
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728951564271949e-05,
+      "loss": 1.2527,
+      "step": 2732
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7287382366939586e-05,
+      "loss": 1.1971,
+      "step": 2733
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728524838371618e-05,
+      "loss": 1.1655,
+      "step": 2734
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283113693256433e-05,
+      "loss": 1.0889,
+      "step": 2735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7280978295767567e-05,
+      "loss": 1.0483,
+      "step": 2736
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727884219145689e-05,
+      "loss": 1.1572,
+      "step": 2737
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7276705380531768e-05,
+      "loss": 1.0627,
+      "step": 2738
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7274567863199637e-05,
+      "loss": 1.0999,
+      "step": 2739
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7272429639668006e-05,
+      "loss": 1.1079,
+      "step": 2740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7270290710144443e-05,
+      "loss": 1.2173,
+      "step": 2741
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7268151074836595e-05,
+      "loss": 1.1267,
+      "step": 2742
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7266010733952176e-05,
+      "loss": 1.0977,
+      "step": 2743
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7263869687698963e-05,
+      "loss": 1.1731,
+      "step": 2744
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72617279362848e-05,
+      "loss": 1.1892,
+      "step": 2745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7259585479917608e-05,
+      "loss": 1.1365,
+      "step": 2746
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7257442318805373e-05,
+      "loss": 1.1921,
+      "step": 2747
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7255298453156147e-05,
+      "loss": 1.1709,
+      "step": 2748
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725315388317805e-05,
+      "loss": 1.1497,
+      "step": 2749
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7251008609079275e-05,
+      "loss": 1.1584,
+      "step": 2750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724886263106808e-05,
+      "loss": 1.1204,
+      "step": 2751
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724671594935279e-05,
+      "loss": 1.1721,
+      "step": 2752
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7244568564141805e-05,
+      "loss": 1.1797,
+      "step": 2753
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7242420475643583e-05,
+      "loss": 1.0376,
+      "step": 2754
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.724027168406666e-05,
+      "loss": 1.1594,
+      "step": 2755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238122189619632e-05,
+      "loss": 1.1155,
+      "step": 2756
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723597199251117e-05,
+      "loss": 0.9983,
+      "step": 2757
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7233821092950013e-05,
+      "loss": 1.1753,
+      "step": 2758
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723166949114496e-05,
+      "loss": 1.1296,
+      "step": 2759
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722951718730489e-05,
+      "loss": 1.1321,
+      "step": 2760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227364181638737e-05,
+      "loss": 1.0375,
+      "step": 2761
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7225210474355516e-05,
+      "loss": 1.1152,
+      "step": 2762
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7223056065664306e-05,
+      "loss": 1.1443,
+      "step": 2763
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7220900955774244e-05,
+      "loss": 1.3306,
+      "step": 2764
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7218745144894552e-05,
+      "loss": 1.1375,
+      "step": 2765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7216588633234507e-05,
+      "loss": 1.1245,
+      "step": 2766
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7214431421003456e-05,
+      "loss": 1.1758,
+      "step": 2767
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212273508410822e-05,
+      "loss": 1.111,
+      "step": 2768
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721011489566609e-05,
+      "loss": 1.304,
+      "step": 2769
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7207955582978814e-05,
+      "loss": 1.063,
+      "step": 2770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720579557055861e-05,
+      "loss": 1.0745,
+      "step": 2771
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720363485861517e-05,
+      "loss": 1.167,
+      "step": 2772
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7201473447358253e-05,
+      "loss": 1.1353,
+      "step": 2773
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199311336997682e-05,
+      "loss": 1.2253,
+      "step": 2774
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197148527743356e-05,
+      "loss": 1.0887,
+      "step": 2775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7194985019805226e-05,
+      "loss": 0.9749,
+      "step": 2776
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719282081339333e-05,
+      "loss": 1.2009,
+      "step": 2777
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190655908717757e-05,
+      "loss": 1.1743,
+      "step": 2778
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7188490305988676e-05,
+      "loss": 0.9668,
+      "step": 2779
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718632400541632e-05,
+      "loss": 1.0763,
+      "step": 2780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7184157007210984e-05,
+      "loss": 1.2876,
+      "step": 2781
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718198931158304e-05,
+      "loss": 1.1316,
+      "step": 2782
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717982091874292e-05,
+      "loss": 1.1415,
+      "step": 2783
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7177651828901136e-05,
+      "loss": 0.9622,
+      "step": 2784
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717548204226824e-05,
+      "loss": 1.1096,
+      "step": 2785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173311559054888e-05,
+      "loss": 0.9869,
+      "step": 2786
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717114037947178e-05,
+      "loss": 1.1472,
+      "step": 2787
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7168968503729684e-05,
+      "loss": 1.1436,
+      "step": 2788
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716679593203945e-05,
+      "loss": 1.0674,
+      "step": 2789
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716462266461198e-05,
+      "loss": 1.1349,
+      "step": 2790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7162448701658253e-05,
+      "loss": 1.0935,
+      "step": 2791
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160274043389315e-05,
+      "loss": 1.2253,
+      "step": 2792
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715809869001627e-05,
+      "loss": 1.1238,
+      "step": 2793
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71559226417503e-05,
+      "loss": 1.1187,
+      "step": 2794
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7153745898802657e-05,
+      "loss": 1.1055,
+      "step": 2795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715156846138465e-05,
+      "loss": 1.0532,
+      "step": 2796
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7149390329707654e-05,
+      "loss": 1.0679,
+      "step": 2797
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714721150398313e-05,
+      "loss": 1.0919,
+      "step": 2798
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145031984422583e-05,
+      "loss": 1.1858,
+      "step": 2799
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71428517712376e-05,
+      "loss": 1.0823,
+      "step": 2800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7140670864639833e-05,
+      "loss": 1.0898,
+      "step": 2801
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7138489264840997e-05,
+      "loss": 1.0667,
+      "step": 2802
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7136306972052876e-05,
+      "loss": 1.1643,
+      "step": 2803
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713412398648733e-05,
+      "loss": 1.1816,
+      "step": 2804
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7131940308356267e-05,
+      "loss": 1.113,
+      "step": 2805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129755937871685e-05,
+      "loss": 1.1267,
+      "step": 2806
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712757087524563e-05,
+      "loss": 1.0662,
+      "step": 2807
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125385120690227e-05,
+      "loss": 1.0537,
+      "step": 2808
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712319867441766e-05,
+      "loss": 1.1975,
+      "step": 2809
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7121011536640195e-05,
+      "loss": 1.1162,
+      "step": 2810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118823707570144e-05,
+      "loss": 1.0327,
+      "step": 2811
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7116635187419898e-05,
+      "loss": 1.0994,
+      "step": 2812
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711444597640192e-05,
+      "loss": 0.9905,
+      "step": 2813
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7112256074728728e-05,
+      "loss": 1.1531,
+      "step": 2814
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7110065482612917e-05,
+      "loss": 1.1007,
+      "step": 2815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7107874200267142e-05,
+      "loss": 1.0527,
+      "step": 2816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7105682227904127e-05,
+      "loss": 1.1306,
+      "step": 2817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710348956573667e-05,
+      "loss": 0.984,
+      "step": 2818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7101296213977627e-05,
+      "loss": 1.2607,
+      "step": 2819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7099102172839918e-05,
+      "loss": 1.1738,
+      "step": 2820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7096907442536546e-05,
+      "loss": 1.1343,
+      "step": 2821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094712023280562e-05,
+      "loss": 1.1145,
+      "step": 2822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70925159152851e-05,
+      "loss": 1.1782,
+      "step": 2823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090319118763347e-05,
+      "loss": 1.1169,
+      "step": 2824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7088121633928566e-05,
+      "loss": 1.1155,
+      "step": 2825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708592346099408e-05,
+      "loss": 1.0948,
+      "step": 2826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7083724600173295e-05,
+      "loss": 1.178,
+      "step": 2827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708152505167966e-05,
+      "loss": 1.0063,
+      "step": 2828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079324815726708e-05,
+      "loss": 1.1819,
+      "step": 2829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7077123892528028e-05,
+      "loss": 1.1958,
+      "step": 2830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707492228229729e-05,
+      "loss": 1.0952,
+      "step": 2831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707271998524821e-05,
+      "loss": 1.1255,
+      "step": 2832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707051700159459e-05,
+      "loss": 1.0991,
+      "step": 2833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706831333155029e-05,
+      "loss": 1.1128,
+      "step": 2834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7066108975329237e-05,
+      "loss": 1.1077,
+      "step": 2835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7063903933145426e-05,
+      "loss": 1.0732,
+      "step": 2836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7061698205212916e-05,
+      "loss": 1.0488,
+      "step": 2837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7059491791745833e-05,
+      "loss": 1.1624,
+      "step": 2838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7057284692958376e-05,
+      "loss": 1.1682,
+      "step": 2839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7055076909064802e-05,
+      "loss": 1.1379,
+      "step": 2840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7052868440279435e-05,
+      "loss": 1.0129,
+      "step": 2841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7050659286816676e-05,
+      "loss": 1.1425,
+      "step": 2842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7048449448890977e-05,
+      "loss": 1.1975,
+      "step": 2843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704623892671687e-05,
+      "loss": 1.0623,
+      "step": 2844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7044027720508948e-05,
+      "loss": 1.1719,
+      "step": 2845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7041815830481867e-05,
+      "loss": 1.0989,
+      "step": 2846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703960325685035e-05,
+      "loss": 1.1008,
+      "step": 2847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037389999829197e-05,
+      "loss": 1.1919,
+      "step": 2848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703517605963326e-05,
+      "loss": 1.2583,
+      "step": 2849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7032961436477468e-05,
+      "loss": 1.1077,
+      "step": 2850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7030746130576805e-05,
+      "loss": 0.9806,
+      "step": 2851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7028530142146335e-05,
+      "loss": 1.1038,
+      "step": 2852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702631347140118e-05,
+      "loss": 1.2446,
+      "step": 2853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7024096118556523e-05,
+      "loss": 1.1279,
+      "step": 2854
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702187808382763e-05,
+      "loss": 1.0691,
+      "step": 2855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7019659367429815e-05,
+      "loss": 1.0496,
+      "step": 2856
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7017439969578473e-05,
+      "loss": 1.1633,
+      "step": 2857
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7015219890489052e-05,
+      "loss": 1.2012,
+      "step": 2858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701299913037708e-05,
+      "loss": 1.0667,
+      "step": 2859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7010777689458134e-05,
+      "loss": 1.1008,
+      "step": 2860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700855556794788e-05,
+      "loss": 1.1072,
+      "step": 2861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700633276606202e-05,
+      "loss": 1.0994,
+      "step": 2862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700410928401635e-05,
+      "loss": 1.078,
+      "step": 2863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7001885122026724e-05,
+      "loss": 1.0925,
+      "step": 2864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6999660280309047e-05,
+      "loss": 1.0725,
+      "step": 2865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699743475907931e-05,
+      "loss": 0.9375,
+      "step": 2866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699520855855356e-05,
+      "loss": 1.1899,
+      "step": 2867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699298167894791e-05,
+      "loss": 1.1401,
+      "step": 2868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699075412047855e-05,
+      "loss": 1.2842,
+      "step": 2869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6988525883361713e-05,
+      "loss": 1.2043,
+      "step": 2870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698629696781372e-05,
+      "loss": 1.0449,
+      "step": 2871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6984067374050944e-05,
+      "loss": 1.134,
+      "step": 2872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6981837102289835e-05,
+      "loss": 1.1613,
+      "step": 2873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69796061527469e-05,
+      "loss": 1.1274,
+      "step": 2874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6977374525638713e-05,
+      "loss": 1.2026,
+      "step": 2875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975142221181923e-05,
+      "loss": 1.0767,
+      "step": 2876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697290923959323e-05,
+      "loss": 1.1118,
+      "step": 2877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6970675581089408e-05,
+      "loss": 1.0153,
+      "step": 2878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69684412458873e-05,
+      "loss": 1.146,
+      "step": 2879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69662062342038e-05,
+      "loss": 1.2983,
+      "step": 2880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696397054625589e-05,
+      "loss": 1.0353,
+      "step": 2881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6961734182260603e-05,
+      "loss": 1.0854,
+      "step": 2882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695949714243504e-05,
+      "loss": 1.2056,
+      "step": 2883
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6957259426996362e-05,
+      "loss": 1.1809,
+      "step": 2884
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695502103616181e-05,
+      "loss": 1.0496,
+      "step": 2885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6952781970148677e-05,
+      "loss": 1.146,
+      "step": 2886
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695054222917433e-05,
+      "loss": 1.074,
+      "step": 2887
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948301813456196e-05,
+      "loss": 1.1135,
+      "step": 2888
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694606072321177e-05,
+      "loss": 1.1663,
+      "step": 2889
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6943818958658616e-05,
+      "loss": 1.1514,
+      "step": 2890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941576520014355e-05,
+      "loss": 1.207,
+      "step": 2891
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6939333407496678e-05,
+      "loss": 1.1223,
+      "step": 2892
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693708962132334e-05,
+      "loss": 1.0168,
+      "step": 2893
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693484516171217e-05,
+      "loss": 1.1365,
+      "step": 2894
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6932600028881053e-05,
+      "loss": 1.2178,
+      "step": 2895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6930354223047936e-05,
+      "loss": 1.0461,
+      "step": 2896
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6928107744430847e-05,
+      "loss": 1.1567,
+      "step": 2897
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692586059324786e-05,
+      "loss": 1.1573,
+      "step": 2898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692361276971713e-05,
+      "loss": 1.0583,
+      "step": 2899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692136427405686e-05,
+      "loss": 1.0535,
+      "step": 2900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6919115106485346e-05,
+      "loss": 1.054,
+      "step": 2901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6916865267220922e-05,
+      "loss": 1.2217,
+      "step": 2902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6914614756481995e-05,
+      "loss": 1.13,
+      "step": 2903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691236357448705e-05,
+      "loss": 1.1191,
+      "step": 2904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6910111721454617e-05,
+      "loss": 1.0291,
+      "step": 2905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6907859197603306e-05,
+      "loss": 1.1797,
+      "step": 2906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6905606003151785e-05,
+      "loss": 1.0284,
+      "step": 2907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6903352138318793e-05,
+      "loss": 1.1091,
+      "step": 2908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6901097603323125e-05,
+      "loss": 1.0566,
+      "step": 2909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689884239838365e-05,
+      "loss": 1.2168,
+      "step": 2910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68965865237193e-05,
+      "loss": 1.1196,
+      "step": 2911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894329979549068e-05,
+      "loss": 1.2847,
+      "step": 2912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892072766092014e-05,
+      "loss": 1.1707,
+      "step": 2913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6889814883567264e-05,
+      "loss": 1.1301,
+      "step": 2914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688755633219401e-05,
+      "loss": 1.0354,
+      "step": 2915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6885297112191504e-05,
+      "loss": 1.0974,
+      "step": 2916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688303722377907e-05,
+      "loss": 1.1687,
+      "step": 2917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688077666717609e-05,
+      "loss": 1.0983,
+      "step": 2918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687851544260202e-05,
+      "loss": 0.9137,
+      "step": 2919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687625355027636e-05,
+      "loss": 0.9937,
+      "step": 2920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687399099041871e-05,
+      "loss": 1.0747,
+      "step": 2921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6871727763248697e-05,
+      "loss": 1.0796,
+      "step": 2922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6869463868986046e-05,
+      "loss": 1.0321,
+      "step": 2923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6867199307850514e-05,
+      "loss": 1.0793,
+      "step": 2924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864934080061955e-05,
+      "loss": 1.0664,
+      "step": 2925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686266818584026e-05,
+      "loss": 0.936,
+      "step": 2926
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686040162540541e-05,
+      "loss": 0.9548,
+      "step": 2927
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6858134398977428e-05,
+      "loss": 1.0559,
+      "step": 2928
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6855866506776413e-05,
+      "loss": 1.0294,
+      "step": 2929
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685359794902253e-05,
+      "loss": 1.1122,
+      "step": 2930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6851328725936003e-05,
+      "loss": 1.1829,
+      "step": 2931
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6849058837737126e-05,
+      "loss": 1.1541,
+      "step": 2932
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6846788284646252e-05,
+      "loss": 1.166,
+      "step": 2933
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6844517066883806e-05,
+      "loss": 1.1667,
+      "step": 2934
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6842245184670268e-05,
+      "loss": 1.1589,
+      "step": 2935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6839972638226188e-05,
+      "loss": 1.0977,
+      "step": 2936
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683769942777218e-05,
+      "loss": 1.1287,
+      "step": 2937
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683542555352893e-05,
+      "loss": 1.1194,
+      "step": 2938
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683315101571717e-05,
+      "loss": 1.1094,
+      "step": 2939
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683087581455771e-05,
+      "loss": 1.2231,
+      "step": 2940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828599950271428e-05,
+      "loss": 0.9922,
+      "step": 2941
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6826323423079257e-05,
+      "loss": 1.0625,
+      "step": 2942
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682404623320219e-05,
+      "loss": 1.1638,
+      "step": 2943
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6821768380861302e-05,
+      "loss": 1.0996,
+      "step": 2944
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6819489866277716e-05,
+      "loss": 1.1692,
+      "step": 2945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817210689672626e-05,
+      "loss": 0.9762,
+      "step": 2946
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6814930851267295e-05,
+      "loss": 1.1187,
+      "step": 2947
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6812650351283038e-05,
+      "loss": 1.0459,
+      "step": 2948
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810369189941242e-05,
+      "loss": 1.1633,
+      "step": 2949
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680808736746336e-05,
+      "loss": 1.0674,
+      "step": 2950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6805804884070905e-05,
+      "loss": 1.1605,
+      "step": 2951
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803521739985455e-05,
+      "loss": 0.9783,
+      "step": 2952
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6801237935428657e-05,
+      "loss": 1.0837,
+      "step": 2953
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6798953470622213e-05,
+      "loss": 1.0837,
+      "step": 2954
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6796668345787896e-05,
+      "loss": 1.1008,
+      "step": 2955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6794382561147537e-05,
+      "loss": 1.0962,
+      "step": 2956
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6792096116923043e-05,
+      "loss": 1.0142,
+      "step": 2957
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678980901333637e-05,
+      "loss": 1.1194,
+      "step": 2958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678752125060955e-05,
+      "loss": 1.0874,
+      "step": 2959
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678523282896467e-05,
+      "loss": 1.0352,
+      "step": 2960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6782943748623886e-05,
+      "loss": 1.0858,
+      "step": 2961
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6780654009809422e-05,
+      "loss": 1.0345,
+      "step": 2962
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6778363612743553e-05,
+      "loss": 1.1311,
+      "step": 2963
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776072557648634e-05,
+      "loss": 1.1519,
+      "step": 2964
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677378084474707e-05,
+      "loss": 1.0659,
+      "step": 2965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6771488474261334e-05,
+      "loss": 1.2495,
+      "step": 2966
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676919544641397e-05,
+      "loss": 1.1724,
+      "step": 2967
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6766901761427583e-05,
+      "loss": 1.0651,
+      "step": 2968
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6764607419524825e-05,
+      "loss": 1.1449,
+      "step": 2969
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6762312420928436e-05,
+      "loss": 1.0944,
+      "step": 2970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6760016765861213e-05,
+      "loss": 1.1289,
+      "step": 2971
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675772045454601e-05,
+      "loss": 1.0537,
+      "step": 2972
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675542348720574e-05,
+      "loss": 1.0054,
+      "step": 2973
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6753125864063397e-05,
+      "loss": 1.1555,
+      "step": 2974
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675082758534203e-05,
+      "loss": 1.1122,
+      "step": 2975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674852865126474e-05,
+      "loss": 1.1387,
+      "step": 2976
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746229062054716e-05,
+      "loss": 1.1311,
+      "step": 2977
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6743928817935194e-05,
+      "loss": 1.0281,
+      "step": 2978
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6741627919129467e-05,
+      "loss": 1.0615,
+      "step": 2979
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673932636586091e-05,
+      "loss": 1.2234,
+      "step": 2980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737024158352957e-05,
+      "loss": 1.1592,
+      "step": 2981
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673472129682909e-05,
+      "loss": 1.1267,
+      "step": 2982
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6732417781512874e-05,
+      "loss": 1.2686,
+      "step": 2983
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673011361262793e-05,
+      "loss": 1.0601,
+      "step": 2984
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6727808790397934e-05,
+      "loss": 1.0333,
+      "step": 2985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672550331504664e-05,
+      "loss": 1.0786,
+      "step": 2986
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6723197186797854e-05,
+      "loss": 1.2134,
+      "step": 2987
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6720890405875458e-05,
+      "loss": 1.04,
+      "step": 2988
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6718582972503382e-05,
+      "loss": 1.1328,
+      "step": 2989
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6716274886905624e-05,
+      "loss": 1.1143,
+      "step": 2990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713966149306257e-05,
+      "loss": 1.1289,
+      "step": 2991
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6711656759929403e-05,
+      "loss": 1.173,
+      "step": 2992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6709346718999254e-05,
+      "loss": 0.9945,
+      "step": 2993
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670703602674006e-05,
+      "loss": 1.2476,
+      "step": 2994
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6704724683376148e-05,
+      "loss": 1.1843,
+      "step": 2995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6702412689131885e-05,
+      "loss": 1.0626,
+      "step": 2996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6700100044231728e-05,
+      "loss": 1.1333,
+      "step": 2997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697786748900172e-05,
+      "loss": 1.1519,
+      "step": 2998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6695472803361795e-05,
+      "loss": 1.2197,
+      "step": 2999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6693158207841224e-05,
+      "loss": 1.0979,
+      "step": 3000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669084296256316e-05,
+      "loss": 1.1731,
+      "step": 3001
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6688527067752355e-05,
+      "loss": 1.1112,
+      "step": 3002
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668621052363364e-05,
+      "loss": 1.0215,
+      "step": 3003
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6683893330431895e-05,
+      "loss": 1.0735,
+      "step": 3004
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6681575488372067e-05,
+      "loss": 1.0388,
+      "step": 3005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667925699767917e-05,
+      "loss": 1.0824,
+      "step": 3006
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6676937858578278e-05,
+      "loss": 1.1089,
+      "step": 3007
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667461807129453e-05,
+      "loss": 1.1567,
+      "step": 3008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6672297636053118e-05,
+      "loss": 1.0754,
+      "step": 3009
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6669976553079313e-05,
+      "loss": 1.1477,
+      "step": 3010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666765482259844e-05,
+      "loss": 1.0244,
+      "step": 3011
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666533244483588e-05,
+      "loss": 1.1987,
+      "step": 3012
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6663009420017097e-05,
+      "loss": 1.0991,
+      "step": 3013
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6660685748367592e-05,
+      "loss": 1.1509,
+      "step": 3014
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6658361430112956e-05,
+      "loss": 1.1772,
+      "step": 3015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6656036465478814e-05,
+      "loss": 1.1416,
+      "step": 3016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6653710854690877e-05,
+      "loss": 1.2405,
+      "step": 3017
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665138459797491e-05,
+      "loss": 1.1519,
+      "step": 3018
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6649057695556746e-05,
+      "loss": 1.09,
+      "step": 3019
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664673014766226e-05,
+      "loss": 1.105,
+      "step": 3020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644401954517426e-05,
+      "loss": 1.1833,
+      "step": 3021
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6642073116348243e-05,
+      "loss": 1.1333,
+      "step": 3022
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6639743633380796e-05,
+      "loss": 1.1362,
+      "step": 3023
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637413505841225e-05,
+      "loss": 1.071,
+      "step": 3024
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663508273395574e-05,
+      "loss": 1.1042,
+      "step": 3025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66327513179506e-05,
+      "loss": 1.2136,
+      "step": 3026
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6630419258052136e-05,
+      "loss": 1.1667,
+      "step": 3027
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6628086554486737e-05,
+      "loss": 1.1215,
+      "step": 3028
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6625753207480863e-05,
+      "loss": 1.1924,
+      "step": 3029
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623419217261026e-05,
+      "loss": 1.2073,
+      "step": 3030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6621084584053807e-05,
+      "loss": 1.1292,
+      "step": 3031
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6618749308085847e-05,
+      "loss": 1.0829,
+      "step": 3032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616413389583844e-05,
+      "loss": 1.0345,
+      "step": 3033
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661407682877457e-05,
+      "loss": 1.0337,
+      "step": 3034
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6611739625884854e-05,
+      "loss": 1.0143,
+      "step": 3035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6609401781141583e-05,
+      "loss": 0.9402,
+      "step": 3036
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6607063294771712e-05,
+      "loss": 1.2336,
+      "step": 3037
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6604724167002255e-05,
+      "loss": 0.9875,
+      "step": 3038
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660238439806029e-05,
+      "loss": 1.2456,
+      "step": 3039
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6600043988172954e-05,
+      "loss": 1.0464,
+      "step": 3040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6597702937567458e-05,
+      "loss": 1.1072,
+      "step": 3041
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6595361246471057e-05,
+      "loss": 1.1548,
+      "step": 3042
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659301891511108e-05,
+      "loss": 1.031,
+      "step": 3043
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659067594371492e-05,
+      "loss": 1.027,
+      "step": 3044
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658833233251002e-05,
+      "loss": 1.1172,
+      "step": 3045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6585988081723898e-05,
+      "loss": 1.0691,
+      "step": 3046
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6583643191584124e-05,
+      "loss": 1.157,
+      "step": 3047
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6581297662318343e-05,
+      "loss": 1.1299,
+      "step": 3048
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6578951494154245e-05,
+      "loss": 1.0596,
+      "step": 3049
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6576604687319597e-05,
+      "loss": 1.1917,
+      "step": 3050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657425724204222e-05,
+      "loss": 1.1606,
+      "step": 3051
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657190915855e-05,
+      "loss": 1.0828,
+      "step": 3052
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569560437070883e-05,
+      "loss": 1.2126,
+      "step": 3053
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6567211077832878e-05,
+      "loss": 1.1133,
+      "step": 3054
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6564861081064053e-05,
+      "loss": 0.9862,
+      "step": 3055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6562510446992548e-05,
+      "loss": 1.0281,
+      "step": 3056
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656015917584655e-05,
+      "loss": 1.0288,
+      "step": 3057
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655780726785432e-05,
+      "loss": 1.1575,
+      "step": 3058
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555454723244176e-05,
+      "loss": 0.9766,
+      "step": 3059
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655310154224449e-05,
+      "loss": 1.1267,
+      "step": 3060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6550747725083716e-05,
+      "loss": 1.2263,
+      "step": 3061
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654839327199035e-05,
+      "loss": 0.9598,
+      "step": 3062
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654603818319296e-05,
+      "loss": 1.1184,
+      "step": 3063
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654368245892017e-05,
+      "loss": 0.9807,
+      "step": 3064
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6541326099400673e-05,
+      "loss": 1.1785,
+      "step": 3065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6538969104863215e-05,
+      "loss": 1.1024,
+      "step": 3066
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6536611475536613e-05,
+      "loss": 0.9873,
+      "step": 3067
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6534253211649735e-05,
+      "loss": 1.2273,
+      "step": 3068
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6531894313431522e-05,
+      "loss": 1.0535,
+      "step": 3069
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6529534781110966e-05,
+      "loss": 1.1008,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6527174614917126e-05,
+      "loss": 1.1514,
+      "step": 3071
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6524813815079124e-05,
+      "loss": 1.0509,
+      "step": 3072
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652245238182614e-05,
+      "loss": 1.1418,
+      "step": 3073
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652009031538742e-05,
+      "loss": 0.9971,
+      "step": 3074
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517727615992263e-05,
+      "loss": 1.2334,
+      "step": 3075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6515364283870037e-05,
+      "loss": 1.2034,
+      "step": 3076
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651300031925017e-05,
+      "loss": 1.131,
+      "step": 3077
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6510635722362152e-05,
+      "loss": 1.116,
+      "step": 3078
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650827049343553e-05,
+      "loss": 1.1921,
+      "step": 3079
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6505904632699915e-05,
+      "loss": 1.0742,
+      "step": 3080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6503538140384984e-05,
+      "loss": 1.1748,
+      "step": 3081
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650117101672047e-05,
+      "loss": 1.118,
+      "step": 3082
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6498803261936164e-05,
+      "loss": 1.2427,
+      "step": 3083
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6496434876261926e-05,
+      "loss": 1.0439,
+      "step": 3084
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649406585992767e-05,
+      "loss": 1.1194,
+      "step": 3085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6491696213163386e-05,
+      "loss": 1.2449,
+      "step": 3086
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64893259361991e-05,
+      "loss": 1.0879,
+      "step": 3087
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648695502926492e-05,
+      "loss": 1.2305,
+      "step": 3088
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6484583492591013e-05,
+      "loss": 1.0112,
+      "step": 3089
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482211326407595e-05,
+      "loss": 1.0688,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6479838530944952e-05,
+      "loss": 1.1323,
+      "step": 3091
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6477465106433434e-05,
+      "loss": 1.0737,
+      "step": 3092
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6475091053103446e-05,
+      "loss": 1.1677,
+      "step": 3093
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6472716371185456e-05,
+      "loss": 1.0469,
+      "step": 3094
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6470341060909993e-05,
+      "loss": 1.167,
+      "step": 3095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646796512250765e-05,
+      "loss": 1.1489,
+      "step": 3096
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6465588556209068e-05,
+      "loss": 1.0664,
+      "step": 3097
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646321136224497e-05,
+      "loss": 1.063,
+      "step": 3098
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6460833540846123e-05,
+      "loss": 1.1284,
+      "step": 3099
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6458455092243367e-05,
+      "loss": 1.2102,
+      "step": 3100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456076016667587e-05,
+      "loss": 1.0864,
+      "step": 3101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453696314349746e-05,
+      "loss": 1.1685,
+      "step": 3102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645131598552086e-05,
+      "loss": 1.0018,
+      "step": 3103
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6448935030412003e-05,
+      "loss": 0.9949,
+      "step": 3104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6446553449254316e-05,
+      "loss": 1.0713,
+      "step": 3105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6444171242278994e-05,
+      "loss": 1.0348,
+      "step": 3106
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6441788409717304e-05,
+      "loss": 1.1541,
+      "step": 3107
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439404951800557e-05,
+      "loss": 1.0458,
+      "step": 3108
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643702086876014e-05,
+      "loss": 1.0554,
+      "step": 3109
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643463616082749e-05,
+      "loss": 1.0571,
+      "step": 3110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6432250828234117e-05,
+      "loss": 1.0532,
+      "step": 3111
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6429864871211575e-05,
+      "loss": 1.1318,
+      "step": 3112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6427478289991495e-05,
+      "loss": 1.067,
+      "step": 3113
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6425091084805557e-05,
+      "loss": 1.0388,
+      "step": 3114
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642270325588551e-05,
+      "loss": 1.0774,
+      "step": 3115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6420314803463152e-05,
+      "loss": 1.0942,
+      "step": 3116
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6417925727770355e-05,
+      "loss": 0.946,
+      "step": 3117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415536029039045e-05,
+      "loss": 1.23,
+      "step": 3118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64131457075012e-05,
+      "loss": 1.2146,
+      "step": 3119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6410754763388884e-05,
+      "loss": 1.038,
+      "step": 3120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640836319693419e-05,
+      "loss": 1.1052,
+      "step": 3121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640597100836929e-05,
+      "loss": 1.146,
+      "step": 3122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640357819792642e-05,
+      "loss": 1.1526,
+      "step": 3123
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640118476583786e-05,
+      "loss": 1.1465,
+      "step": 3124
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6398790712335964e-05,
+      "loss": 1.0962,
+      "step": 3125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639639603765314e-05,
+      "loss": 1.0566,
+      "step": 3126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6394000742021854e-05,
+      "loss": 1.1804,
+      "step": 3127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6391604825674644e-05,
+      "loss": 1.1548,
+      "step": 3128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6389208288844094e-05,
+      "loss": 1.0138,
+      "step": 3129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638681113176286e-05,
+      "loss": 1.1475,
+      "step": 3130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6384413354663643e-05,
+      "loss": 1.0957,
+      "step": 3131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6382014957779227e-05,
+      "loss": 1.0757,
+      "step": 3132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637961594134244e-05,
+      "loss": 1.2651,
+      "step": 3133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6377216305586167e-05,
+      "loss": 1.0874,
+      "step": 3134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6374816050743357e-05,
+      "loss": 1.1594,
+      "step": 3135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6372415177047037e-05,
+      "loss": 1.1986,
+      "step": 3136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6370013684730267e-05,
+      "loss": 1.0826,
+      "step": 3137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636761157402618e-05,
+      "loss": 1.0713,
+      "step": 3138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636520884516797e-05,
+      "loss": 1.1484,
+      "step": 3139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6362805498388886e-05,
+      "loss": 1.0518,
+      "step": 3140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6360401533922244e-05,
+      "loss": 1.0271,
+      "step": 3141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6357996952001414e-05,
+      "loss": 1.0962,
+      "step": 3142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6355591752859826e-05,
+      "loss": 1.0851,
+      "step": 3143
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353185936730972e-05,
+      "loss": 1.2344,
+      "step": 3144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6350779503848403e-05,
+      "loss": 1.1782,
+      "step": 3145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348372454445734e-05,
+      "loss": 1.3098,
+      "step": 3146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634596478875663e-05,
+      "loss": 1.0679,
+      "step": 3147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6343556507014827e-05,
+      "loss": 1.2122,
+      "step": 3148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6341147609454114e-05,
+      "loss": 1.0967,
+      "step": 3149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6338738096308345e-05,
+      "loss": 1.1199,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6336327967811428e-05,
+      "loss": 1.0632,
+      "step": 3151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6333917224197332e-05,
+      "loss": 1.1199,
+      "step": 3152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6331505865700084e-05,
+      "loss": 1.0752,
+      "step": 3153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632909389255378e-05,
+      "loss": 1.0439,
+      "step": 3154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6326681304992565e-05,
+      "loss": 1.1965,
+      "step": 3155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6324268103250648e-05,
+      "loss": 1.0126,
+      "step": 3156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63218542875623e-05,
+      "loss": 1.0365,
+      "step": 3157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631943985816185e-05,
+      "loss": 1.0684,
+      "step": 3158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631702481528368e-05,
+      "loss": 1.2207,
+      "step": 3159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631460915916224e-05,
+      "loss": 1.0413,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631219289003204e-05,
+      "loss": 1.0723,
+      "step": 3161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6309776008127642e-05,
+      "loss": 1.1243,
+      "step": 3162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630735851368367e-05,
+      "loss": 1.0344,
+      "step": 3163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6304940406934818e-05,
+      "loss": 1.1489,
+      "step": 3164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6302521688115823e-05,
+      "loss": 1.1543,
+      "step": 3165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6300102357461492e-05,
+      "loss": 1.1145,
+      "step": 3166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6297682415206683e-05,
+      "loss": 1.0631,
+      "step": 3167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6295261861586328e-05,
+      "loss": 1.1726,
+      "step": 3168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6292840696835405e-05,
+      "loss": 0.9968,
+      "step": 3169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6290418921188952e-05,
+      "loss": 1.2712,
+      "step": 3170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628799653488207e-05,
+      "loss": 1.1597,
+      "step": 3171
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6285573538149926e-05,
+      "loss": 1.0306,
+      "step": 3172
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628314993122774e-05,
+      "loss": 1.0525,
+      "step": 3173
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628072571435078e-05,
+      "loss": 1.2062,
+      "step": 3174
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6278300887754388e-05,
+      "loss": 1.1865,
+      "step": 3175
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6275875451673962e-05,
+      "loss": 1.1123,
+      "step": 3176
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6273449406344964e-05,
+      "loss": 1.1501,
+      "step": 3177
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6271022752002902e-05,
+      "loss": 1.079,
+      "step": 3178
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6268595488883353e-05,
+      "loss": 0.9547,
+      "step": 3179
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6266167617221948e-05,
+      "loss": 1.025,
+      "step": 3180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6263739137254382e-05,
+      "loss": 1.1614,
+      "step": 3181
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626131004921641e-05,
+      "loss": 1.0994,
+      "step": 3182
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6258880353343832e-05,
+      "loss": 0.9725,
+      "step": 3183
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625645004987253e-05,
+      "loss": 1.1262,
+      "step": 3184
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6254019139038425e-05,
+      "loss": 1.1069,
+      "step": 3185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6251587621077506e-05,
+      "loss": 1.0884,
+      "step": 3186
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6249155496225823e-05,
+      "loss": 1.0372,
+      "step": 3187
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624672276471948e-05,
+      "loss": 1.209,
+      "step": 3188
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6244289426794636e-05,
+      "loss": 1.0583,
+      "step": 3189
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624185548268752e-05,
+      "loss": 1.103,
+      "step": 3190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6239420932634412e-05,
+      "loss": 1.0176,
+      "step": 3191
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6236985776871656e-05,
+      "loss": 1.0023,
+      "step": 3192
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234550015635647e-05,
+      "loss": 1.1104,
+      "step": 3193
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6232113649162848e-05,
+      "loss": 1.0281,
+      "step": 3194
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6229676677689773e-05,
+      "loss": 1.1724,
+      "step": 3195
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6227239101452998e-05,
+      "loss": 1.0667,
+      "step": 3196
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622480092068916e-05,
+      "loss": 1.0596,
+      "step": 3197
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622236213563495e-05,
+      "loss": 1.2156,
+      "step": 3198
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6219922746527125e-05,
+      "loss": 1.1608,
+      "step": 3199
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621748275360249e-05,
+      "loss": 1.1931,
+      "step": 3200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621504215709791e-05,
+      "loss": 1.0554,
+      "step": 3201
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6212600957250322e-05,
+      "loss": 1.0413,
+      "step": 3202
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210159154296715e-05,
+      "loss": 1.1467,
+      "step": 3203
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6207716748474122e-05,
+      "loss": 1.0359,
+      "step": 3204
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6205273740019657e-05,
+      "loss": 0.9841,
+      "step": 3205
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6202830129170477e-05,
+      "loss": 1.0569,
+      "step": 3206
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200385916163803e-05,
+      "loss": 1.0549,
+      "step": 3207
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6197941101236912e-05,
+      "loss": 1.1328,
+      "step": 3208
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6195495684627147e-05,
+      "loss": 1.1433,
+      "step": 3209
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61930496665719e-05,
+      "loss": 1.0371,
+      "step": 3210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6190603047308625e-05,
+      "loss": 1.1316,
+      "step": 3211
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6188155827074837e-05,
+      "loss": 1.0529,
+      "step": 3212
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618570800610811e-05,
+      "loss": 1.1565,
+      "step": 3213
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618325958464606e-05,
+      "loss": 1.1414,
+      "step": 3214
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6180810562926387e-05,
+      "loss": 1.1433,
+      "step": 3215
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6178360941186834e-05,
+      "loss": 1.0012,
+      "step": 3216
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6175910719665205e-05,
+      "loss": 1.0796,
+      "step": 3217
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617345989859936e-05,
+      "loss": 1.1826,
+      "step": 3218
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6171008478227216e-05,
+      "loss": 1.1567,
+      "step": 3219
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6168556458786762e-05,
+      "loss": 1.176,
+      "step": 3220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6166103840516027e-05,
+      "loss": 1.0284,
+      "step": 3221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616365062365311e-05,
+      "loss": 1.0781,
+      "step": 3222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6161196808436162e-05,
+      "loss": 1.2102,
+      "step": 3223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615874239510339e-05,
+      "loss": 1.1184,
+      "step": 3224
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6156287383893073e-05,
+      "loss": 1.1882,
+      "step": 3225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615383177504353e-05,
+      "loss": 1.0078,
+      "step": 3226
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6151375568793152e-05,
+      "loss": 1.0037,
+      "step": 3227
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148918765380377e-05,
+      "loss": 1.0732,
+      "step": 3228
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6146461365043708e-05,
+      "loss": 1.2061,
+      "step": 3229
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6144003368021706e-05,
+      "loss": 1.0918,
+      "step": 3230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6141544774552987e-05,
+      "loss": 1.0229,
+      "step": 3231
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6139085584876225e-05,
+      "loss": 1.175,
+      "step": 3232
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613662579923015e-05,
+      "loss": 1.0879,
+      "step": 3233
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6134165417853563e-05,
+      "loss": 1.1174,
+      "step": 3234
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6131704440985303e-05,
+      "loss": 1.085,
+      "step": 3235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6129242868864283e-05,
+      "loss": 1.1041,
+      "step": 3236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6126780701729458e-05,
+      "loss": 1.2168,
+      "step": 3237
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6124317939819854e-05,
+      "loss": 1.1958,
+      "step": 3238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612185458337456e-05,
+      "loss": 1.0735,
+      "step": 3239
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61193906326327e-05,
+      "loss": 1.1257,
+      "step": 3240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6116926087833477e-05,
+      "loss": 1.0878,
+      "step": 3241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6114460949216136e-05,
+      "loss": 1.1143,
+      "step": 3242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611199521702e-05,
+      "loss": 1.2133,
+      "step": 3243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6109528891484425e-05,
+      "loss": 1.1646,
+      "step": 3244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6107061972848843e-05,
+      "loss": 1.1824,
+      "step": 3245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6104594461352733e-05,
+      "loss": 1.031,
+      "step": 3246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6102126357235645e-05,
+      "loss": 1.1018,
+      "step": 3247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6099657660737165e-05,
+      "loss": 1.0957,
+      "step": 3248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6097188372096955e-05,
+      "loss": 1.0715,
+      "step": 3249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6094718491554727e-05,
+      "loss": 1.1245,
+      "step": 3250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092248019350254e-05,
+      "loss": 1.0962,
+      "step": 3251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6089776955723362e-05,
+      "loss": 1.2234,
+      "step": 3252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6087305300913934e-05,
+      "loss": 1.1306,
+      "step": 3253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6084833055161922e-05,
+      "loss": 1.1771,
+      "step": 3254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6082360218707316e-05,
+      "loss": 1.1758,
+      "step": 3255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6079886791790177e-05,
+      "loss": 1.1467,
+      "step": 3256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6077412774650623e-05,
+      "loss": 1.1345,
+      "step": 3257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6074938167528827e-05,
+      "loss": 1.0228,
+      "step": 3258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6072462970665014e-05,
+      "loss": 1.0062,
+      "step": 3259
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6069987184299473e-05,
+      "loss": 1.0447,
+      "step": 3260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067510808672545e-05,
+      "loss": 1.157,
+      "step": 3261
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6065033844024635e-05,
+      "loss": 1.1147,
+      "step": 3262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6062556290596202e-05,
+      "loss": 1.1514,
+      "step": 3263
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606007814862776e-05,
+      "loss": 1.1804,
+      "step": 3264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605759941835988e-05,
+      "loss": 1.1028,
+      "step": 3265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6055120100033195e-05,
+      "loss": 1.1641,
+      "step": 3266
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605264019388839e-05,
+      "loss": 1.1396,
+      "step": 3267
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6050159700166208e-05,
+      "loss": 1.1182,
+      "step": 3268
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6047678619107456e-05,
+      "loss": 1.0896,
+      "step": 3269
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6045196950952983e-05,
+      "loss": 1.0686,
+      "step": 3270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604271469594371e-05,
+      "loss": 1.0974,
+      "step": 3271
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604023185432061e-05,
+      "loss": 1.1907,
+      "step": 3272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6037748426324705e-05,
+      "loss": 1.0647,
+      "step": 3273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6035264412197088e-05,
+      "loss": 1.095,
+      "step": 3274
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.60327798121789e-05,
+      "loss": 1.0945,
+      "step": 3275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603029462651134e-05,
+      "loss": 1.1517,
+      "step": 3276
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602780885543566e-05,
+      "loss": 1.0559,
+      "step": 3277
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6025322499193186e-05,
+      "loss": 1.1519,
+      "step": 3278
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6022835558025273e-05,
+      "loss": 1.0991,
+      "step": 3279
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6020348032173362e-05,
+      "loss": 1.0759,
+      "step": 3280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6017859921878922e-05,
+      "loss": 1.1895,
+      "step": 3281
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6015371227383504e-05,
+      "loss": 0.9882,
+      "step": 3282
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6012881948928702e-05,
+      "loss": 1.0474,
+      "step": 3283
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601039208675617e-05,
+      "loss": 1.0271,
+      "step": 3284
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600790164110762e-05,
+      "loss": 1.1724,
+      "step": 3285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6005410612224814e-05,
+      "loss": 1.199,
+      "step": 3286
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600291900034958e-05,
+      "loss": 1.1172,
+      "step": 3287
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6000426805723802e-05,
+      "loss": 1.0906,
+      "step": 3288
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599793402858941e-05,
+      "loss": 1.1099,
+      "step": 3289
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5995440669188397e-05,
+      "loss": 1.2141,
+      "step": 3290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599294672776282e-05,
+      "loss": 1.1001,
+      "step": 3291
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599045220455478e-05,
+      "loss": 1.1099,
+      "step": 3292
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5987957099806444e-05,
+      "loss": 1.1313,
+      "step": 3293
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5985461413760028e-05,
+      "loss": 1.1841,
+      "step": 3294
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598296514665781e-05,
+      "loss": 1.2021,
+      "step": 3295
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5980468298742117e-05,
+      "loss": 1.0984,
+      "step": 3296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5977970870255345e-05,
+      "loss": 1.1373,
+      "step": 3297
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5975472861439937e-05,
+      "loss": 1.0635,
+      "step": 3298
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597297427253839e-05,
+      "loss": 1.0393,
+      "step": 3299
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5970475103793266e-05,
+      "loss": 1.0503,
+      "step": 3300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596797535544718e-05,
+      "loss": 1.1006,
+      "step": 3301
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59654750277428e-05,
+      "loss": 1.1013,
+      "step": 3302
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596297412092285e-05,
+      "loss": 1.0117,
+      "step": 3303
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5960472635230117e-05,
+      "loss": 1.0162,
+      "step": 3304
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5957970570907437e-05,
+      "loss": 0.9783,
+      "step": 3305
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5955467928197704e-05,
+      "loss": 1.1174,
+      "step": 3306
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5952964707343875e-05,
+      "loss": 1.1523,
+      "step": 3307
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5950460908588952e-05,
+      "loss": 1.1648,
+      "step": 3308
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5947956532176e-05,
+      "loss": 1.0896,
+      "step": 3309
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594545157834814e-05,
+      "loss": 1.0957,
+      "step": 3310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5942946047348547e-05,
+      "loss": 1.0535,
+      "step": 3311
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940439939420448e-05,
+      "loss": 1.1547,
+      "step": 3312
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593793325480714e-05,
+      "loss": 1.0104,
+      "step": 3313
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5935425993751957e-05,
+      "loss": 1.2148,
+      "step": 3314
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5932918156498306e-05,
+      "loss": 1.0435,
+      "step": 3315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5930409743289636e-05,
+      "loss": 1.1619,
+      "step": 3316
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592790075436946e-05,
+      "loss": 1.1409,
+      "step": 3317
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592539118998135e-05,
+      "loss": 1.1526,
+      "step": 3318
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5922881050368927e-05,
+      "loss": 1.0979,
+      "step": 3319
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5920370335775867e-05,
+      "loss": 1.1895,
+      "step": 3320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5917859046445906e-05,
+      "loss": 1.1313,
+      "step": 3321
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5915347182622833e-05,
+      "loss": 1.0492,
+      "step": 3322
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59128347445505e-05,
+      "loss": 1.049,
+      "step": 3323
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5910321732472805e-05,
+      "loss": 1.0303,
+      "step": 3324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5907808146633708e-05,
+      "loss": 1.0333,
+      "step": 3325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5905293987277218e-05,
+      "loss": 1.0231,
+      "step": 3326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590277925464741e-05,
+      "loss": 1.2112,
+      "step": 3327
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590026394898841e-05,
+      "loss": 1.1375,
+      "step": 3328
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5897748070544387e-05,
+      "loss": 1.1458,
+      "step": 3329
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895231619559588e-05,
+      "loss": 1.1467,
+      "step": 3330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5892714596278302e-05,
+      "loss": 1.1128,
+      "step": 3331
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5890197000944873e-05,
+      "loss": 1.198,
+      "step": 3332
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588767883380371e-05,
+      "loss": 0.9286,
+      "step": 3333
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5885160095099264e-05,
+      "loss": 1.123,
+      "step": 3334
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588264078507606e-05,
+      "loss": 1.1421,
+      "step": 3335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5880120903978652e-05,
+      "loss": 1.1963,
+      "step": 3336
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5877600452051677e-05,
+      "loss": 1.114,
+      "step": 3337
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587507942953981e-05,
+      "loss": 1.1135,
+      "step": 3338
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587255783668779e-05,
+      "loss": 1.1292,
+      "step": 3339
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5870035673740396e-05,
+      "loss": 1.0916,
+      "step": 3340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586751294094249e-05,
+      "loss": 1.0302,
+      "step": 3341
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5864989638538967e-05,
+      "loss": 1.0349,
+      "step": 3342
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586246576677478e-05,
+      "loss": 1.0745,
+      "step": 3343
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5859941325894946e-05,
+      "loss": 1.1055,
+      "step": 3344
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585741631614453e-05,
+      "loss": 1.1914,
+      "step": 3345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5854890737768657e-05,
+      "loss": 1.0481,
+      "step": 3346
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58523645910125e-05,
+      "loss": 1.2913,
+      "step": 3347
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5849837876121295e-05,
+      "loss": 1.0747,
+      "step": 3348
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5847310593340336e-05,
+      "loss": 1.0836,
+      "step": 3349
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5844782742914958e-05,
+      "loss": 1.2131,
+      "step": 3350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5842254325090557e-05,
+      "loss": 1.0859,
+      "step": 3351
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5839725340112594e-05,
+      "loss": 1.147,
+      "step": 3352
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5837195788226578e-05,
+      "loss": 1.1401,
+      "step": 3353
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834665669678065e-05,
+      "loss": 1.1111,
+      "step": 3354
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5832134984712678e-05,
+      "loss": 1.0106,
+      "step": 3355
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582960373357609e-05,
+      "loss": 1.0957,
+      "step": 3356
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582707191651403e-05,
+      "loss": 1.2034,
+      "step": 3357
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582453953377228e-05,
+      "loss": 1.0911,
+      "step": 3358
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582200658559668e-05,
+      "loss": 1.1248,
+      "step": 3359
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5819473072233123e-05,
+      "loss": 1.1133,
+      "step": 3360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5816938993927554e-05,
+      "loss": 1.2571,
+      "step": 3361
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5814404350925984e-05,
+      "loss": 1.0106,
+      "step": 3362
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5811869143474458e-05,
+      "loss": 1.2363,
+      "step": 3363
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58093333718191e-05,
+      "loss": 1.2617,
+      "step": 3364
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5806797036206072e-05,
+      "loss": 1.1382,
+      "step": 3365
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5804260136881596e-05,
+      "loss": 1.022,
+      "step": 3366
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580172267409195e-05,
+      "loss": 1.0304,
+      "step": 3367
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5799184648083464e-05,
+      "loss": 1.192,
+      "step": 3368
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5796646059102527e-05,
+      "loss": 1.1959,
+      "step": 3369
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5794106907395576e-05,
+      "loss": 1.095,
+      "step": 3370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579156719320911e-05,
+      "loss": 1.1492,
+      "step": 3371
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5789026916789677e-05,
+      "loss": 1.0308,
+      "step": 3372
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5786486078383884e-05,
+      "loss": 1.1305,
+      "step": 3373
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5783944678238387e-05,
+      "loss": 1.0719,
+      "step": 3374
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5781402716599902e-05,
+      "loss": 1.0129,
+      "step": 3375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5778860193715195e-05,
+      "loss": 1.0547,
+      "step": 3376
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5776317109831086e-05,
+      "loss": 1.0427,
+      "step": 3377
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577377346519446e-05,
+      "loss": 1.0955,
+      "step": 3378
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577122926005225e-05,
+      "loss": 1.1782,
+      "step": 3379
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576868449465143e-05,
+      "loss": 1.1326,
+      "step": 3380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5766139169239044e-05,
+      "loss": 1.0693,
+      "step": 3381
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5763593284062192e-05,
+      "loss": 1.09,
+      "step": 3382
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5761046839368024e-05,
+      "loss": 1.0808,
+      "step": 3383
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575849983540374e-05,
+      "loss": 1.0469,
+      "step": 3384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5755952272416595e-05,
+      "loss": 1.0259,
+      "step": 3385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575340415065391e-05,
+      "loss": 1.1274,
+      "step": 3386
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575085547036304e-05,
+      "loss": 1.0847,
+      "step": 3387
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574830623179141e-05,
+      "loss": 1.1226,
+      "step": 3388
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5745756435186494e-05,
+      "loss": 1.0774,
+      "step": 3389
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5743206080795825e-05,
+      "loss": 1.1106,
+      "step": 3390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5740655168866982e-05,
+      "loss": 1.2322,
+      "step": 3391
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57381036996476e-05,
+      "loss": 1.1119,
+      "step": 3392
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573555167338538e-05,
+      "loss": 1.0466,
+      "step": 3393
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5732999090328055e-05,
+      "loss": 1.041,
+      "step": 3394
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5730445950723435e-05,
+      "loss": 0.9941,
+      "step": 3395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727892254819363e-05,
+      "loss": 1.1584,
+      "step": 3396
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5725338002863755e-05,
+      "loss": 0.9574,
+      "step": 3397
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5722783195104568e-05,
+      "loss": 1.0059,
+      "step": 3398
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5720227831789816e-05,
+      "loss": 1.1093,
+      "step": 3399
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5717671913167573e-05,
+      "loss": 1.2535,
+      "step": 3400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5715115439485958e-05,
+      "loss": 0.9751,
+      "step": 3401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5712558410993147e-05,
+      "loss": 1.0515,
+      "step": 3402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5710000827937376e-05,
+      "loss": 1.1144,
+      "step": 3403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707442690566924e-05,
+      "loss": 1.0739,
+      "step": 3404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5704883999130137e-05,
+      "loss": 1.1079,
+      "step": 3405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57023247538754e-05,
+      "loss": 1.1833,
+      "step": 3406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5699764955051163e-05,
+      "loss": 1.1414,
+      "step": 3407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5697204602905926e-05,
+      "loss": 1.0503,
+      "step": 3408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5694643697688236e-05,
+      "loss": 1.126,
+      "step": 3409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569208223964671e-05,
+      "loss": 1.1791,
+      "step": 3410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5689520229030002e-05,
+      "loss": 1.1792,
+      "step": 3411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568695766608683e-05,
+      "loss": 1.2051,
+      "step": 3412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5684394551065956e-05,
+      "loss": 1.0737,
+      "step": 3413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5681830884216215e-05,
+      "loss": 1.0935,
+      "step": 3414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567926666578647e-05,
+      "loss": 1.1982,
+      "step": 3415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5676701896025656e-05,
+      "loss": 1.0715,
+      "step": 3416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567413657518275e-05,
+      "loss": 1.0979,
+      "step": 3417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5671570703506788e-05,
+      "loss": 1.2446,
+      "step": 3418
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5669004281246868e-05,
+      "loss": 1.1025,
+      "step": 3419
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5666437308652122e-05,
+      "loss": 1.2314,
+      "step": 3420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5663869785971754e-05,
+      "loss": 1.1694,
+      "step": 3421
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5661301713455013e-05,
+      "loss": 1.1528,
+      "step": 3422
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5658733091351195e-05,
+      "loss": 1.1455,
+      "step": 3423
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5656163919909666e-05,
+      "loss": 1.1504,
+      "step": 3424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565359419937983e-05,
+      "loss": 1.1398,
+      "step": 3425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565102393001115e-05,
+      "loss": 1.116,
+      "step": 3426
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564845311205314e-05,
+      "loss": 1.0688,
+      "step": 3427
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564588174575538e-05,
+      "loss": 1.0457,
+      "step": 3428
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5643309831367477e-05,
+      "loss": 0.9886,
+      "step": 3429
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5640737369139122e-05,
+      "loss": 1.0847,
+      "step": 3430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5638164359320033e-05,
+      "loss": 1.0984,
+      "step": 3431
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563559080216e-05,
+      "loss": 1.1619,
+      "step": 3432
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563301669790885e-05,
+      "loss": 1.1897,
+      "step": 3433
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5630442046816484e-05,
+      "loss": 1.0284,
+      "step": 3434
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5627866849132837e-05,
+      "loss": 1.1711,
+      "step": 3435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5625291105107895e-05,
+      "loss": 1.0565,
+      "step": 3436
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5622714814991717e-05,
+      "loss": 0.9717,
+      "step": 3437
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5620137979034403e-05,
+      "loss": 1.2339,
+      "step": 3438
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5617560597486102e-05,
+      "loss": 1.1729,
+      "step": 3439
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5614982670597024e-05,
+      "loss": 1.2402,
+      "step": 3440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561240419861743e-05,
+      "loss": 0.9896,
+      "step": 3441
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5609825181797627e-05,
+      "loss": 1.0837,
+      "step": 3442
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5607245620387983e-05,
+      "loss": 1.1439,
+      "step": 3443
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560466551463892e-05,
+      "loss": 1.0049,
+      "step": 3444
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5602084864800904e-05,
+      "loss": 1.1223,
+      "step": 3445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5599503671124462e-05,
+      "loss": 1.1699,
+      "step": 3446
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5596921933860167e-05,
+      "loss": 1.0713,
+      "step": 3447
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559433965325865e-05,
+      "loss": 1.3228,
+      "step": 3448
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5591756829570602e-05,
+      "loss": 1.219,
+      "step": 3449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5589173463046745e-05,
+      "loss": 1.1372,
+      "step": 3450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5586589553937874e-05,
+      "loss": 1.1586,
+      "step": 3451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5584005102494828e-05,
+      "loss": 1.1348,
+      "step": 3452
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55814201089685e-05,
+      "loss": 1.0327,
+      "step": 3453
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5578834573609835e-05,
+      "loss": 1.2649,
+      "step": 3454
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5576248496669827e-05,
+      "loss": 0.9811,
+      "step": 3455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557366187839954e-05,
+      "loss": 1.1558,
+      "step": 3456
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557107471905006e-05,
+      "loss": 0.869,
+      "step": 3457
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5568487018872558e-05,
+      "loss": 1.1115,
+      "step": 3458
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5565898778118236e-05,
+      "loss": 1.0916,
+      "step": 3459
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5563309997038354e-05,
+      "loss": 1.2419,
+      "step": 3460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5560720675884224e-05,
+      "loss": 1.1162,
+      "step": 3461
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5558130814907215e-05,
+      "loss": 1.0944,
+      "step": 3462
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5555540414358746e-05,
+      "loss": 0.9563,
+      "step": 3463
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5552949474490285e-05,
+      "loss": 0.9913,
+      "step": 3464
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5550357995553353e-05,
+      "loss": 1.1523,
+      "step": 3465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554776597779953e-05,
+      "loss": 1.0063,
+      "step": 3466
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5545173421480445e-05,
+      "loss": 1.127,
+      "step": 3467
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554258032684777e-05,
+      "loss": 0.9917,
+      "step": 3468
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5539986694153243e-05,
+      "loss": 1.0703,
+      "step": 3469
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5537392523648646e-05,
+      "loss": 1.1494,
+      "step": 3470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5534797815585813e-05,
+      "loss": 1.1448,
+      "step": 3471
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553220257021664e-05,
+      "loss": 1.0962,
+      "step": 3472
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5529606787793063e-05,
+      "loss": 1.135,
+      "step": 3473
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552701046856708e-05,
+      "loss": 1.1575,
+      "step": 3474
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5524413612790725e-05,
+      "loss": 1.092,
+      "step": 3475
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5521816220716106e-05,
+      "loss": 1.0228,
+      "step": 3476
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551921829259537e-05,
+      "loss": 1.1963,
+      "step": 3477
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5516619828680713e-05,
+      "loss": 1.1814,
+      "step": 3478
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5514020829224394e-05,
+      "loss": 1.137,
+      "step": 3479
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5511421294478715e-05,
+      "loss": 1.1514,
+      "step": 3480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5508821224696035e-05,
+      "loss": 1.0557,
+      "step": 3481
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5506220620128764e-05,
+      "loss": 1.1853,
+      "step": 3482
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550361948102936e-05,
+      "loss": 1.1379,
+      "step": 3483
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5501017807650343e-05,
+      "loss": 1.0281,
+      "step": 3484
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549841560024427e-05,
+      "loss": 1.0752,
+      "step": 3485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549581285906376e-05,
+      "loss": 1.0526,
+      "step": 3486
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549320958436148e-05,
+      "loss": 1.1001,
+      "step": 3487
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5490605776390156e-05,
+      "loss": 1.2949,
+      "step": 3488
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548800143540256e-05,
+      "loss": 1.0352,
+      "step": 3489
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5485396561651503e-05,
+      "loss": 1.1448,
+      "step": 3490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5482791155389877e-05,
+      "loss": 1.1003,
+      "step": 3491
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54801852168706e-05,
+      "loss": 1.2101,
+      "step": 3492
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5477578746346655e-05,
+      "loss": 1.2395,
+      "step": 3493
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547497174407107e-05,
+      "loss": 1.1021,
+      "step": 3494
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472364210296928e-05,
+      "loss": 1.1133,
+      "step": 3495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546975614527736e-05,
+      "loss": 1.2307,
+      "step": 3496
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5467147549265564e-05,
+      "loss": 1.1467,
+      "step": 3497
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5464538422514757e-05,
+      "loss": 1.1492,
+      "step": 3498
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5461928765278244e-05,
+      "loss": 1.1233,
+      "step": 3499
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5459318577809355e-05,
+      "loss": 1.0867,
+      "step": 3500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5456707860361485e-05,
+      "loss": 1.1384,
+      "step": 3501
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5454096613188076e-05,
+      "loss": 1.1582,
+      "step": 3502
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5451484836542626e-05,
+      "loss": 1.0483,
+      "step": 3503
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5448872530678678e-05,
+      "loss": 1.1135,
+      "step": 3504
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544625969584983e-05,
+      "loss": 1.2031,
+      "step": 3505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5443646332309726e-05,
+      "loss": 1.0648,
+      "step": 3506
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544103244031207e-05,
+      "loss": 1.1035,
+      "step": 3507
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5438418020110616e-05,
+      "loss": 1.1404,
+      "step": 3508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543580307195916e-05,
+      "loss": 1.1536,
+      "step": 3509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5433187596111557e-05,
+      "loss": 1.0901,
+      "step": 3510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5430571592821712e-05,
+      "loss": 1.0879,
+      "step": 3511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5427955062343586e-05,
+      "loss": 1.1504,
+      "step": 3512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542533800493118e-05,
+      "loss": 1.1809,
+      "step": 3513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5422720420838554e-05,
+      "loss": 1.0851,
+      "step": 3514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5420102310319818e-05,
+      "loss": 0.9758,
+      "step": 3515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541748367362913e-05,
+      "loss": 1.2181,
+      "step": 3516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541486451102071e-05,
+      "loss": 1.0021,
+      "step": 3517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541224482274881e-05,
+      "loss": 1.1423,
+      "step": 3518
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540962460906775e-05,
+      "loss": 1.0229,
+      "step": 3519
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5407003870231894e-05,
+      "loss": 1.1165,
+      "step": 3520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5404382606495657e-05,
+      "loss": 1.1567,
+      "step": 3521
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5401760818113504e-05,
+      "loss": 1.1409,
+      "step": 3522
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5399138505339958e-05,
+      "loss": 1.1287,
+      "step": 3523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539651566842958e-05,
+      "loss": 1.1221,
+      "step": 3524
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5393892307636996e-05,
+      "loss": 1.1042,
+      "step": 3525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5391268423216873e-05,
+      "loss": 1.1481,
+      "step": 3526
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5388644015423932e-05,
+      "loss": 1.1731,
+      "step": 3527
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5386019084512946e-05,
+      "loss": 1.1829,
+      "step": 3528
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5383393630738737e-05,
+      "loss": 1.0962,
+      "step": 3529
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538076765435618e-05,
+      "loss": 1.1296,
+      "step": 3530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5378141155620196e-05,
+      "loss": 1.1584,
+      "step": 3531
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537551413478576e-05,
+      "loss": 1.0872,
+      "step": 3532
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5372886592107904e-05,
+      "loss": 1.0996,
+      "step": 3533
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370258527841696e-05,
+      "loss": 1.2075,
+      "step": 3534
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536762994224227e-05,
+      "loss": 1.1836,
+      "step": 3535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5365000835564798e-05,
+      "loss": 1.0881,
+      "step": 3536
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362371208064507e-05,
+      "loss": 1.1187,
+      "step": 3537
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5359741059996682e-05,
+      "loss": 1.1079,
+      "step": 3538
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5357110391616647e-05,
+      "loss": 1.0604,
+      "step": 3539
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5354479203179785e-05,
+      "loss": 1.1072,
+      "step": 3540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5351847494941524e-05,
+      "loss": 0.995,
+      "step": 3541
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5349215267157346e-05,
+      "loss": 1.1135,
+      "step": 3542
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5346582520082777e-05,
+      "loss": 1.1768,
+      "step": 3543
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5343949253973408e-05,
+      "loss": 1.1987,
+      "step": 3544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534131546908486e-05,
+      "loss": 1.2122,
+      "step": 3545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5338681165672825e-05,
+      "loss": 1.026,
+      "step": 3546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533604634399303e-05,
+      "loss": 1.1721,
+      "step": 3547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533341100430126e-05,
+      "loss": 1.1301,
+      "step": 3548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5330775146853345e-05,
+      "loss": 1.1077,
+      "step": 3549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532813877190517e-05,
+      "loss": 1.0437,
+      "step": 3550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532550187971267e-05,
+      "loss": 0.9614,
+      "step": 3551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5322864470531834e-05,
+      "loss": 1.0597,
+      "step": 3552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5320226544618682e-05,
+      "loss": 1.0195,
+      "step": 3553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5317588102229312e-05,
+      "loss": 1.092,
+      "step": 3554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5314949143619853e-05,
+      "loss": 1.0239,
+      "step": 3555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5312309669046487e-05,
+      "loss": 1.0239,
+      "step": 3556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530966967876545e-05,
+      "loss": 1.1077,
+      "step": 3557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5307029173033026e-05,
+      "loss": 1.0457,
+      "step": 3558
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5304388152105554e-05,
+      "loss": 0.9764,
+      "step": 3559
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5301746616239416e-05,
+      "loss": 1.165,
+      "step": 3560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299104565691047e-05,
+      "loss": 1.1853,
+      "step": 3561
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5296462000716926e-05,
+      "loss": 1.0889,
+      "step": 3562
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5293818921573597e-05,
+      "loss": 1.0806,
+      "step": 3563
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5291175328517638e-05,
+      "loss": 1.2356,
+      "step": 3564
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5288531221805687e-05,
+      "loss": 1.1593,
+      "step": 3565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528588660169443e-05,
+      "loss": 1.0713,
+      "step": 3566
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528324146844059e-05,
+      "loss": 1.1189,
+      "step": 3567
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528059582230096e-05,
+      "loss": 1.0818,
+      "step": 3568
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5277949663532376e-05,
+      "loss": 1.0856,
+      "step": 3569
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5275302992391717e-05,
+      "loss": 1.1414,
+      "step": 3570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5272655809135914e-05,
+      "loss": 1.0326,
+      "step": 3571
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5270008114021954e-05,
+      "loss": 1.0493,
+      "step": 3572
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526735990730687e-05,
+      "loss": 1.0085,
+      "step": 3573
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5264711189247743e-05,
+      "loss": 1.1147,
+      "step": 3574
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5262061960101702e-05,
+      "loss": 1.1317,
+      "step": 3575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525941222012593e-05,
+      "loss": 1.1123,
+      "step": 3576
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5256761969577665e-05,
+      "loss": 1.0106,
+      "step": 3577
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5254111208714175e-05,
+      "loss": 1.1895,
+      "step": 3578
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52514599377928e-05,
+      "loss": 1.0933,
+      "step": 3579
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5248808157070916e-05,
+      "loss": 1.1816,
+      "step": 3580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5246155866805954e-05,
+      "loss": 1.0981,
+      "step": 3581
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5243503067255389e-05,
+      "loss": 1.2378,
+      "step": 3582
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5240849758676754e-05,
+      "loss": 1.2039,
+      "step": 3583
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5238195941327622e-05,
+      "loss": 1.1071,
+      "step": 3584
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5235541615465628e-05,
+      "loss": 1.1252,
+      "step": 3585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5232886781348435e-05,
+      "loss": 1.0496,
+      "step": 3586
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5230231439233778e-05,
+      "loss": 1.0972,
+      "step": 3587
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5227575589379433e-05,
+      "loss": 1.2036,
+      "step": 3588
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522491923204322e-05,
+      "loss": 1.0168,
+      "step": 3589
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5222262367483012e-05,
+      "loss": 1.0487,
+      "step": 3590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5219604995956732e-05,
+      "loss": 1.012,
+      "step": 3591
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5216947117722355e-05,
+      "loss": 1.1274,
+      "step": 3592
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214288733037903e-05,
+      "loss": 0.9733,
+      "step": 3593
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211629842161442e-05,
+      "loss": 1.051,
+      "step": 3594
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520897044535109e-05,
+      "loss": 1.1956,
+      "step": 3595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5206310542865023e-05,
+      "loss": 1.0344,
+      "step": 3596
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5203650134961454e-05,
+      "loss": 1.0598,
+      "step": 3597
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520098922189865e-05,
+      "loss": 1.0557,
+      "step": 3598
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198327803934925e-05,
+      "loss": 1.1562,
+      "step": 3599
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5195665881328646e-05,
+      "loss": 1.3301,
+      "step": 3600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5193003454338229e-05,
+      "loss": 1.1064,
+      "step": 3601
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5190340523222132e-05,
+      "loss": 1.2498,
+      "step": 3602
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5187677088238869e-05,
+      "loss": 1.2141,
+      "step": 3603
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5185013149646998e-05,
+      "loss": 1.0708,
+      "step": 3604
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5182348707705135e-05,
+      "loss": 1.1324,
+      "step": 3605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5179683762671933e-05,
+      "loss": 1.0355,
+      "step": 3606
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.51770183148061e-05,
+      "loss": 1.209,
+      "step": 3607
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5174352364366391e-05,
+      "loss": 0.9922,
+      "step": 3608
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5171685911611613e-05,
+      "loss": 1.1445,
+      "step": 3609
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5169018956800619e-05,
+      "loss": 1.2642,
+      "step": 3610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166351500192309e-05,
+      "loss": 1.0603,
+      "step": 3611
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516368354204564e-05,
+      "loss": 1.0947,
+      "step": 3612
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5161015082619606e-05,
+      "loss": 1.1008,
+      "step": 3613
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5158346122173253e-05,
+      "loss": 1.0659,
+      "step": 3614
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155676660965686e-05,
+      "loss": 1.0593,
+      "step": 3615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5153006699256046e-05,
+      "loss": 1.0754,
+      "step": 3616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5150336237303527e-05,
+      "loss": 1.1292,
+      "step": 3617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5147665275367374e-05,
+      "loss": 1.0747,
+      "step": 3618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5144993813706874e-05,
+      "loss": 1.1143,
+      "step": 3619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514232185258137e-05,
+      "loss": 1.0771,
+      "step": 3620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513964939225025e-05,
+      "loss": 1.0298,
+      "step": 3621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136976432972951e-05,
+      "loss": 1.187,
+      "step": 3622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5134302975008957e-05,
+      "loss": 1.2139,
+      "step": 3623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5131629018617803e-05,
+      "loss": 1.2241,
+      "step": 3624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128954564059068e-05,
+      "loss": 1.1013,
+      "step": 3625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5126279611592388e-05,
+      "loss": 1.1211,
+      "step": 3626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5123604161477436e-05,
+      "loss": 1.105,
+      "step": 3627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5120928213973942e-05,
+      "loss": 1.0916,
+      "step": 3628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511825176934168e-05,
+      "loss": 1.155,
+      "step": 3629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5115574827840475e-05,
+      "loss": 1.017,
+      "step": 3630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112897389730197e-05,
+      "loss": 1.1138,
+      "step": 3631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511021945527077e-05,
+      "loss": 1.0144,
+      "step": 3632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5107541024722152e-05,
+      "loss": 1.1467,
+      "step": 3633
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510486209834437e-05,
+      "loss": 1.2766,
+      "step": 3634
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5102182676397484e-05,
+      "loss": 0.9958,
+      "step": 3635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5099502759141609e-05,
+      "loss": 1.1472,
+      "step": 3636
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5096822346836901e-05,
+      "loss": 1.0404,
+      "step": 3637
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5094141439743572e-05,
+      "loss": 1.0686,
+      "step": 3638
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5091460038121878e-05,
+      "loss": 1.2095,
+      "step": 3639
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508877814223213e-05,
+      "loss": 1.1453,
+      "step": 3640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086095752334666e-05,
+      "loss": 1.1921,
+      "step": 3641
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5083412868689899e-05,
+      "loss": 1.0642,
+      "step": 3642
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5080729491558277e-05,
+      "loss": 1.0551,
+      "step": 3643
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507804562120029e-05,
+      "loss": 1.1055,
+      "step": 3644
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5075361257876484e-05,
+      "loss": 0.9957,
+      "step": 3645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5072676401847456e-05,
+      "loss": 1.1182,
+      "step": 3646
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5069991053373847e-05,
+      "loss": 1.0828,
+      "step": 3647
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5067305212716332e-05,
+      "loss": 1.1411,
+      "step": 3648
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064618880135661e-05,
+      "loss": 1.1335,
+      "step": 3649
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5061932055892614e-05,
+      "loss": 1.0697,
+      "step": 3650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505924474024802e-05,
+      "loss": 1.1289,
+      "step": 3651
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5056556933462756e-05,
+      "loss": 1.0632,
+      "step": 3652
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505386863579775e-05,
+      "loss": 1.1013,
+      "step": 3653
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5051179847513979e-05,
+      "loss": 1.146,
+      "step": 3654
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5048490568872465e-05,
+      "loss": 1.1577,
+      "step": 3655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045800800134269e-05,
+      "loss": 1.1448,
+      "step": 3656
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043110541560519e-05,
+      "loss": 1.0024,
+      "step": 3657
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5040419793412375e-05,
+      "loss": 1.1138,
+      "step": 3658
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5037728555951047e-05,
+      "loss": 0.942,
+      "step": 3659
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5035036829437797e-05,
+      "loss": 1.033,
+      "step": 3660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503234461413393e-05,
+      "loss": 0.9226,
+      "step": 3661
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5029651910300806e-05,
+      "loss": 0.9556,
+      "step": 3662
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5026958718199823e-05,
+      "loss": 1.1343,
+      "step": 3663
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5024265038092422e-05,
+      "loss": 1.1819,
+      "step": 3664
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5021570870240117e-05,
+      "loss": 1.1035,
+      "step": 3665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5018876214904441e-05,
+      "loss": 1.0764,
+      "step": 3666
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5016181072346986e-05,
+      "loss": 1.1494,
+      "step": 3667
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013485442829389e-05,
+      "loss": 1.127,
+      "step": 3668
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5010789326613347e-05,
+      "loss": 1.0662,
+      "step": 3669
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5008092723960579e-05,
+      "loss": 0.9656,
+      "step": 3670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005395635132873e-05,
+      "loss": 1.1028,
+      "step": 3671
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5002698060392056e-05,
+      "loss": 1.1079,
+      "step": 3672
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.176,
+      "step": 3673
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4997301454218634e-05,
+      "loss": 1.0415,
+      "step": 3674
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4994602423309919e-05,
+      "loss": 1.217,
+      "step": 3675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4991902907535872e-05,
+      "loss": 1.1748,
+      "step": 3676
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4989202907158562e-05,
+      "loss": 1.2129,
+      "step": 3677
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4986502422440095e-05,
+      "loss": 0.9613,
+      "step": 3678
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4983801453642628e-05,
+      "loss": 1.1663,
+      "step": 3679
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981100001028365e-05,
+      "loss": 1.1533,
+      "step": 3680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497839806485956e-05,
+      "loss": 1.0852,
+      "step": 3681
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4975695645398512e-05,
+      "loss": 0.8873,
+      "step": 3682
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497299274290756e-05,
+      "loss": 1.1548,
+      "step": 3683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4970289357649104e-05,
+      "loss": 1.1218,
+      "step": 3684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967585489885574e-05,
+      "loss": 0.9968,
+      "step": 3685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4964881139879464e-05,
+      "loss": 1.1494,
+      "step": 3686
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.49621763078933e-05,
+      "loss": 0.9501,
+      "step": 3687
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4959470994189666e-05,
+      "loss": 1.2432,
+      "step": 3688
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4956765199031189e-05,
+      "loss": 1.2012,
+      "step": 3689
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4954058922680535e-05,
+      "loss": 1.1001,
+      "step": 3690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4951352165400428e-05,
+      "loss": 1.1025,
+      "step": 3691
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4948644927453635e-05,
+      "loss": 1.1819,
+      "step": 3692
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494593720910297e-05,
+      "loss": 1.041,
+      "step": 3693
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4943229010611287e-05,
+      "loss": 1.0854,
+      "step": 3694
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4940520332241494e-05,
+      "loss": 1.1162,
+      "step": 3695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4937811174256549e-05,
+      "loss": 1.1831,
+      "step": 3696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4935101536919445e-05,
+      "loss": 1.1973,
+      "step": 3697
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493239142049323e-05,
+      "loss": 1.1626,
+      "step": 3698
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4929680825241e-05,
+      "loss": 1.1868,
+      "step": 3699
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4926969751425885e-05,
+      "loss": 1.0955,
+      "step": 3700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492425819931108e-05,
+      "loss": 1.1389,
+      "step": 3701
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4921546169159808e-05,
+      "loss": 0.9713,
+      "step": 3702
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4918833661235354e-05,
+      "loss": 1.1919,
+      "step": 3703
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491612067580104e-05,
+      "loss": 1.1318,
+      "step": 3704
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913407213120235e-05,
+      "loss": 1.1113,
+      "step": 3705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4910693273456358e-05,
+      "loss": 1.2416,
+      "step": 3706
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4907978857072873e-05,
+      "loss": 1.0935,
+      "step": 3707
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490526396423329e-05,
+      "loss": 1.1852,
+      "step": 3708
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4902548595201163e-05,
+      "loss": 1.1899,
+      "step": 3709
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4899832750240094e-05,
+      "loss": 1.0681,
+      "step": 3710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4897116429613734e-05,
+      "loss": 1.1794,
+      "step": 3711
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4894399633585774e-05,
+      "loss": 1.0681,
+      "step": 3712
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489168236241996e-05,
+      "loss": 1.1892,
+      "step": 3713
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4888964616380073e-05,
+      "loss": 1.1716,
+      "step": 3714
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4886246395729952e-05,
+      "loss": 1.1364,
+      "step": 3715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4883527700733469e-05,
+      "loss": 1.0527,
+      "step": 3716
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4880808531654554e-05,
+      "loss": 1.1387,
+      "step": 3717
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4878088888757178e-05,
+      "loss": 1.1238,
+      "step": 3718
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4875368772305358e-05,
+      "loss": 1.1155,
+      "step": 3719
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872648182563154e-05,
+      "loss": 1.0581,
+      "step": 3720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486992711979468e-05,
+      "loss": 1.1511,
+      "step": 3721
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4867205584264087e-05,
+      "loss": 1.0648,
+      "step": 3722
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4864483576235577e-05,
+      "loss": 1.1248,
+      "step": 3723
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861761095973399e-05,
+      "loss": 1.1699,
+      "step": 3724
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485903814374184e-05,
+      "loss": 1.187,
+      "step": 3725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4856314719805243e-05,
+      "loss": 1.0862,
+      "step": 3726
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485359082442799e-05,
+      "loss": 0.9492,
+      "step": 3727
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4850866457874516e-05,
+      "loss": 1.1523,
+      "step": 3728
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4848141620409289e-05,
+      "loss": 1.0447,
+      "step": 3729
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4845416312296833e-05,
+      "loss": 1.1794,
+      "step": 3730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842690533801717e-05,
+      "loss": 1.0515,
+      "step": 3731
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4839964285188556e-05,
+      "loss": 1.1084,
+      "step": 3732
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4837237566722e-05,
+      "loss": 0.9951,
+      "step": 3733
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4834510378666762e-05,
+      "loss": 1.1299,
+      "step": 3734
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483178272128759e-05,
+      "loss": 1.0988,
+      "step": 3735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4829054594849273e-05,
+      "loss": 1.1472,
+      "step": 3736
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826325999616655e-05,
+      "loss": 1.0364,
+      "step": 3737
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4823596935854625e-05,
+      "loss": 1.0554,
+      "step": 3738
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4820867403828115e-05,
+      "loss": 1.1056,
+      "step": 3739
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4818137403802096e-05,
+      "loss": 1.0776,
+      "step": 3740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4815406936041592e-05,
+      "loss": 1.1196,
+      "step": 3741
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4812676000811677e-05,
+      "loss": 1.1611,
+      "step": 3742
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809944598377461e-05,
+      "loss": 1.1545,
+      "step": 3743
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48072127290041e-05,
+      "loss": 1.1672,
+      "step": 3744
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48044803929568e-05,
+      "loss": 1.1611,
+      "step": 3745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4801747590500814e-05,
+      "loss": 1.0455,
+      "step": 3746
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4799014321901431e-05,
+      "loss": 1.1938,
+      "step": 3747
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4796280587423991e-05,
+      "loss": 1.1931,
+      "step": 3748
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793546387333883e-05,
+      "loss": 1.0808,
+      "step": 3749
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790811721896539e-05,
+      "loss": 1.1633,
+      "step": 3750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4788076591377427e-05,
+      "loss": 1.0938,
+      "step": 3751
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4785340996042072e-05,
+      "loss": 1.0475,
+      "step": 3752
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478260493615604e-05,
+      "loss": 1.0586,
+      "step": 3753
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4779868411984943e-05,
+      "loss": 1.0066,
+      "step": 3754
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4777131423794435e-05,
+      "loss": 1.1814,
+      "step": 3755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4774393971850214e-05,
+      "loss": 1.1096,
+      "step": 3756
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4771656056418031e-05,
+      "loss": 1.1173,
+      "step": 3757
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4768917677763677e-05,
+      "loss": 1.1477,
+      "step": 3758
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4766178836152983e-05,
+      "loss": 1.2095,
+      "step": 3759
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4763439531851833e-05,
+      "loss": 1.1571,
+      "step": 3760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760699765126157e-05,
+      "loss": 1.0483,
+      "step": 3761
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4757959536241917e-05,
+      "loss": 1.1511,
+      "step": 3762
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755218845465133e-05,
+      "loss": 1.2109,
+      "step": 3763
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4752477693061865e-05,
+      "loss": 1.1951,
+      "step": 3764
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4749736079298222e-05,
+      "loss": 1.1172,
+      "step": 3765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474699400444035e-05,
+      "loss": 1.1589,
+      "step": 3766
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4744251468754443e-05,
+      "loss": 1.1237,
+      "step": 3767
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4741508472506743e-05,
+      "loss": 1.084,
+      "step": 3768
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738765015963531e-05,
+      "loss": 1.1382,
+      "step": 3769
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473602109939114e-05,
+      "loss": 1.1558,
+      "step": 3770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733276723055941e-05,
+      "loss": 1.0432,
+      "step": 3771
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4730531887224353e-05,
+      "loss": 1.1848,
+      "step": 3772
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4727786592162839e-05,
+      "loss": 1.0374,
+      "step": 3773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725040838137905e-05,
+      "loss": 1.1592,
+      "step": 3774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722294625416104e-05,
+      "loss": 1.1992,
+      "step": 3775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719547954264032e-05,
+      "loss": 1.0404,
+      "step": 3776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4716800824948331e-05,
+      "loss": 1.0649,
+      "step": 3777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4714053237735685e-05,
+      "loss": 1.0898,
+      "step": 3778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4711305192892826e-05,
+      "loss": 1.0784,
+      "step": 3779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4708556690686526e-05,
+      "loss": 1.0315,
+      "step": 3780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705807731383605e-05,
+      "loss": 1.0928,
+      "step": 3781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4703058315250926e-05,
+      "loss": 1.1523,
+      "step": 3782
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4700308442555395e-05,
+      "loss": 1.0071,
+      "step": 3783
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4697558113563965e-05,
+      "loss": 1.166,
+      "step": 3784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4694807328543634e-05,
+      "loss": 1.1543,
+      "step": 3785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692056087761439e-05,
+      "loss": 1.0177,
+      "step": 3786
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4689304391484467e-05,
+      "loss": 1.208,
+      "step": 3787
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4686552239979846e-05,
+      "loss": 1.0784,
+      "step": 3788
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4683799633514748e-05,
+      "loss": 1.1091,
+      "step": 3789
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.468104657235639e-05,
+      "loss": 1.0536,
+      "step": 3790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4678293056772035e-05,
+      "loss": 1.0909,
+      "step": 3791
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4675539087028987e-05,
+      "loss": 1.2532,
+      "step": 3792
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4672784663394597e-05,
+      "loss": 1.1003,
+      "step": 3793
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4670029786136257e-05,
+      "loss": 1.083,
+      "step": 3794
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4667274455521405e-05,
+      "loss": 1.0753,
+      "step": 3795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4664518671817525e-05,
+      "loss": 1.1191,
+      "step": 3796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4661762435292138e-05,
+      "loss": 1.0217,
+      "step": 3797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4659005746212816e-05,
+      "loss": 1.2576,
+      "step": 3798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4656248604847174e-05,
+      "loss": 1.251,
+      "step": 3799
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4653491011462866e-05,
+      "loss": 1.0393,
+      "step": 3800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4650732966327598e-05,
+      "loss": 1.2249,
+      "step": 3801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464797446970911e-05,
+      "loss": 1.1162,
+      "step": 3802
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4645215521875194e-05,
+      "loss": 1.0593,
+      "step": 3803
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464245612309368e-05,
+      "loss": 1.1941,
+      "step": 3804
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463969627363245e-05,
+      "loss": 0.9213,
+      "step": 3805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4636935973759418e-05,
+      "loss": 1.0503,
+      "step": 3806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4634175223742553e-05,
+      "loss": 1.1807,
+      "step": 3807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463141402384986e-05,
+      "loss": 1.1208,
+      "step": 3808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462865237434939e-05,
+      "loss": 1.0654,
+      "step": 3809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4625890275509237e-05,
+      "loss": 1.1492,
+      "step": 3810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4623127727597545e-05,
+      "loss": 1.0398,
+      "step": 3811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4620364730882492e-05,
+      "loss": 1.1233,
+      "step": 3812
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4617601285632308e-05,
+      "loss": 1.0034,
+      "step": 3813
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4614837392115254e-05,
+      "loss": 1.1255,
+      "step": 3814
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4612073050599649e-05,
+      "loss": 1.0608,
+      "step": 3815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460930826135385e-05,
+      "loss": 1.1128,
+      "step": 3816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460654302464625e-05,
+      "loss": 1.0131,
+      "step": 3817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4603777340745305e-05,
+      "loss": 1.0568,
+      "step": 3818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4601011209919489e-05,
+      "loss": 1.0564,
+      "step": 3819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4598244632437339e-05,
+      "loss": 1.1451,
+      "step": 3820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4595477608567426e-05,
+      "loss": 1.1633,
+      "step": 3821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4592710138578365e-05,
+      "loss": 1.0188,
+      "step": 3822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4589942222738822e-05,
+      "loss": 1.1772,
+      "step": 3823
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4587173861317498e-05,
+      "loss": 1.0431,
+      "step": 3824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4584405054583133e-05,
+      "loss": 1.0986,
+      "step": 3825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4581635802804526e-05,
+      "loss": 0.8491,
+      "step": 3826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578866106250507e-05,
+      "loss": 1.0303,
+      "step": 3827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457609596518995e-05,
+      "loss": 1.1266,
+      "step": 3828
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4573325379891775e-05,
+      "loss": 1.2278,
+      "step": 3829
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4570554350624943e-05,
+      "loss": 1.1113,
+      "step": 3830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4567782877658467e-05,
+      "loss": 1.1187,
+      "step": 3831
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456501096126139e-05,
+      "loss": 1.2371,
+      "step": 3832
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45622386017028e-05,
+      "loss": 1.1104,
+      "step": 3833
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455946579925184e-05,
+      "loss": 1.1616,
+      "step": 3834
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4556692554177684e-05,
+      "loss": 1.1135,
+      "step": 3835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4553918866749552e-05,
+      "loss": 1.1182,
+      "step": 3836
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4551144737236707e-05,
+      "loss": 0.9581,
+      "step": 3837
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454837016590846e-05,
+      "loss": 1.097,
+      "step": 3838
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4545595153034154e-05,
+      "loss": 0.9863,
+      "step": 3839
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542819698883184e-05,
+      "loss": 1.1064,
+      "step": 3840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4540043803724983e-05,
+      "loss": 1.2874,
+      "step": 3841
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453726746782904e-05,
+      "loss": 1.1404,
+      "step": 3842
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4534490691464862e-05,
+      "loss": 0.9934,
+      "step": 3843
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531713474902018e-05,
+      "loss": 1.1726,
+      "step": 3844
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4528935818410113e-05,
+      "loss": 1.0132,
+      "step": 3845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45261577222588e-05,
+      "loss": 1.0671,
+      "step": 3846
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4523379186717767e-05,
+      "loss": 1.1753,
+      "step": 3847
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.452060021205675e-05,
+      "loss": 1.0981,
+      "step": 3848
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4517820798545524e-05,
+      "loss": 1.1763,
+      "step": 3849
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4515040946453907e-05,
+      "loss": 1.103,
+      "step": 3850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4512260656051767e-05,
+      "loss": 1.0581,
+      "step": 3851
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4509479927609001e-05,
+      "loss": 1.0864,
+      "step": 3852
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4506698761395566e-05,
+      "loss": 1.1222,
+      "step": 3853
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4503917157681444e-05,
+      "loss": 1.1648,
+      "step": 3854
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450113511673667e-05,
+      "loss": 1.0061,
+      "step": 3855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4498352638831312e-05,
+      "loss": 0.9812,
+      "step": 3856
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44955697242355e-05,
+      "loss": 1.0859,
+      "step": 3857
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4492786373219382e-05,
+      "loss": 1.0494,
+      "step": 3858
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490002586053167e-05,
+      "loss": 1.092,
+      "step": 3859
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4487218363007096e-05,
+      "loss": 1.059,
+      "step": 3860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484433704351453e-05,
+      "loss": 1.1541,
+      "step": 3861
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481648610356572e-05,
+      "loss": 1.1162,
+      "step": 3862
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478863081292818e-05,
+      "loss": 1.1248,
+      "step": 3863
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4476077117430609e-05,
+      "loss": 1.2327,
+      "step": 3864
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44732907190404e-05,
+      "loss": 1.0184,
+      "step": 3865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4470503886392686e-05,
+      "loss": 1.0188,
+      "step": 3866
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446771661975801e-05,
+      "loss": 1.0657,
+      "step": 3867
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464928919406948e-05,
+      "loss": 1.0513,
+      "step": 3868
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4462140785610132e-05,
+      "loss": 1.0837,
+      "step": 3869
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4459352218638226e-05,
+      "loss": 1.176,
+      "step": 3870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4456563218761929e-05,
+      "loss": 1.1599,
+      "step": 3871
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4453773786252006e-05,
+      "loss": 1.1548,
+      "step": 3872
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4450983921379238e-05,
+      "loss": 1.2563,
+      "step": 3873
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448193624414461e-05,
+      "loss": 1.2678,
+      "step": 3874
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4445402895628555e-05,
+      "loss": 1.2522,
+      "step": 3875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4442611735292435e-05,
+      "loss": 0.9779,
+      "step": 3876
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4439820143677064e-05,
+      "loss": 1.1191,
+      "step": 3877
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443702812105344e-05,
+      "loss": 1.1221,
+      "step": 3878
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4434235667692607e-05,
+      "loss": 1.0901,
+      "step": 3879
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4431442783865652e-05,
+      "loss": 1.0671,
+      "step": 3880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44286494698437e-05,
+      "loss": 1.2119,
+      "step": 3881
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4425855725897924e-05,
+      "loss": 1.1277,
+      "step": 3882
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4423061552299531e-05,
+      "loss": 1.2495,
+      "step": 3883
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4420266949319777e-05,
+      "loss": 1.1414,
+      "step": 3884
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441747191722995e-05,
+      "loss": 1.1711,
+      "step": 3885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441467645630139e-05,
+      "loss": 1.1343,
+      "step": 3886
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4411880566805477e-05,
+      "loss": 1.1202,
+      "step": 3887
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409084249013623e-05,
+      "loss": 1.1255,
+      "step": 3888
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4406287503197294e-05,
+      "loss": 1.0466,
+      "step": 3889
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4403490329627992e-05,
+      "loss": 1.1292,
+      "step": 3890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4400692728577257e-05,
+      "loss": 1.0736,
+      "step": 3891
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4397894700316676e-05,
+      "loss": 1.1604,
+      "step": 3892
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4395096245117878e-05,
+      "loss": 1.124,
+      "step": 3893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4392297363252525e-05,
+      "loss": 1.0671,
+      "step": 3894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4389498054992333e-05,
+      "loss": 1.1067,
+      "step": 3895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4386698320609052e-05,
+      "loss": 1.1714,
+      "step": 3896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438389816037447e-05,
+      "loss": 1.0603,
+      "step": 3897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4381097574560423e-05,
+      "loss": 1.1873,
+      "step": 3898
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4378296563438785e-05,
+      "loss": 1.0143,
+      "step": 3899
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4375495127281474e-05,
+      "loss": 1.0901,
+      "step": 3900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4372693266360448e-05,
+      "loss": 1.1179,
+      "step": 3901
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4369890980947703e-05,
+      "loss": 1.0846,
+      "step": 3902
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4367088271315279e-05,
+      "loss": 1.2876,
+      "step": 3903
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364285137735263e-05,
+      "loss": 1.0896,
+      "step": 3904
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4361481580479767e-05,
+      "loss": 1.2008,
+      "step": 3905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4358677599820961e-05,
+      "loss": 1.1606,
+      "step": 3906
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4355873196031048e-05,
+      "loss": 1.0825,
+      "step": 3907
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4353068369382275e-05,
+      "loss": 1.1147,
+      "step": 3908
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4350263120146925e-05,
+      "loss": 1.0863,
+      "step": 3909
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4347457448597328e-05,
+      "loss": 1.1831,
+      "step": 3910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4344651355005855e-05,
+      "loss": 1.1338,
+      "step": 3911
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4341844839644913e-05,
+      "loss": 1.0408,
+      "step": 3912
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4339037902786952e-05,
+      "loss": 1.0159,
+      "step": 3913
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433623054470446e-05,
+      "loss": 1.1577,
+      "step": 3914
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433342276566998e-05,
+      "loss": 1.0586,
+      "step": 3915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4330614565956073e-05,
+      "loss": 1.0559,
+      "step": 3916
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432780594583536e-05,
+      "loss": 0.9518,
+      "step": 3917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324996905580493e-05,
+      "loss": 1.03,
+      "step": 3918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4322187445464172e-05,
+      "loss": 1.1643,
+      "step": 3919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4319377565759129e-05,
+      "loss": 1.2527,
+      "step": 3920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316567266738142e-05,
+      "loss": 1.1201,
+      "step": 3921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431375654867403e-05,
+      "loss": 1.1597,
+      "step": 3922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4310945411839652e-05,
+      "loss": 1.1614,
+      "step": 3923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4308133856507905e-05,
+      "loss": 1.1501,
+      "step": 3924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4305321882951726e-05,
+      "loss": 1.0337,
+      "step": 3925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4302509491444103e-05,
+      "loss": 1.0247,
+      "step": 3926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4299696682258053e-05,
+      "loss": 1.0492,
+      "step": 3927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4296883455666636e-05,
+      "loss": 1.0594,
+      "step": 3928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294069811942955e-05,
+      "loss": 1.2512,
+      "step": 3929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4291255751360158e-05,
+      "loss": 1.2156,
+      "step": 3930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4288441274191418e-05,
+      "loss": 1.1427,
+      "step": 3931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4285626380709965e-05,
+      "loss": 1.082,
+      "step": 3932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428281107118906e-05,
+      "loss": 1.087,
+      "step": 3933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4279995345902013e-05,
+      "loss": 1.0514,
+      "step": 3934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4277179205122163e-05,
+      "loss": 0.9839,
+      "step": 3935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4274362649122892e-05,
+      "loss": 1.1431,
+      "step": 3936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4271545678177634e-05,
+      "loss": 1.271,
+      "step": 3937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4268728292559846e-05,
+      "loss": 1.0762,
+      "step": 3938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4265910492543041e-05,
+      "loss": 1.0061,
+      "step": 3939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4263092278400758e-05,
+      "loss": 1.1377,
+      "step": 3940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426027365040659e-05,
+      "loss": 1.0583,
+      "step": 3941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425745460883416e-05,
+      "loss": 1.0232,
+      "step": 3942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254635153957132e-05,
+      "loss": 1.2283,
+      "step": 3943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425181528604922e-05,
+      "loss": 1.1326,
+      "step": 3944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4248995005384163e-05,
+      "loss": 1.1052,
+      "step": 3945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4246174312235751e-05,
+      "loss": 1.2117,
+      "step": 3946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424335320687781e-05,
+      "loss": 1.1357,
+      "step": 3947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424053168958421e-05,
+      "loss": 1.1621,
+      "step": 3948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4237709760628855e-05,
+      "loss": 1.127,
+      "step": 3949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4234887420285693e-05,
+      "loss": 1.0562,
+      "step": 3950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423206466882871e-05,
+      "loss": 1.1235,
+      "step": 3951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229241506531933e-05,
+      "loss": 1.0872,
+      "step": 3952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4226417933669427e-05,
+      "loss": 1.0151,
+      "step": 3953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42235939505153e-05,
+      "loss": 1.1592,
+      "step": 3954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42207695573437e-05,
+      "loss": 1.1486,
+      "step": 3955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4217944754428809e-05,
+      "loss": 1.1375,
+      "step": 3956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4215119542044858e-05,
+      "loss": 0.9832,
+      "step": 3957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4212293920466106e-05,
+      "loss": 1.1582,
+      "step": 3958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4209467889966863e-05,
+      "loss": 1.1045,
+      "step": 3959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4206641450821474e-05,
+      "loss": 1.0264,
+      "step": 3960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4203814603304322e-05,
+      "loss": 1.1743,
+      "step": 3961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4200987347689835e-05,
+      "loss": 1.2004,
+      "step": 3962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4198159684252468e-05,
+      "loss": 1.2512,
+      "step": 3963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4195331613266734e-05,
+      "loss": 1.1523,
+      "step": 3964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4192503135007172e-05,
+      "loss": 1.1484,
+      "step": 3965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4189674249748364e-05,
+      "loss": 1.1228,
+      "step": 3966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186844957764934e-05,
+      "loss": 1.1567,
+      "step": 3967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4184015259331544e-05,
+      "loss": 1.1636,
+      "step": 3968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4181185154722892e-05,
+      "loss": 1.1226,
+      "step": 3969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417835464421372e-05,
+      "loss": 1.0303,
+      "step": 3970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4175523728078811e-05,
+      "loss": 1.0342,
+      "step": 3971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4172692406592981e-05,
+      "loss": 1.0977,
+      "step": 3972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4169860680031086e-05,
+      "loss": 1.1062,
+      "step": 3973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4167028548668031e-05,
+      "loss": 1.0098,
+      "step": 3974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4164196012778745e-05,
+      "loss": 1.0476,
+      "step": 3975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4161363072638215e-05,
+      "loss": 1.1538,
+      "step": 3976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4158529728521447e-05,
+      "loss": 1.2178,
+      "step": 3977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4155695980703502e-05,
+      "loss": 1.0074,
+      "step": 3978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4152861829459468e-05,
+      "loss": 1.103,
+      "step": 3979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4150027275064484e-05,
+      "loss": 1.0955,
+      "step": 3980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4147192317793721e-05,
+      "loss": 1.1067,
+      "step": 3981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4144356957922394e-05,
+      "loss": 1.1223,
+      "step": 3982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141521195725745e-05,
+      "loss": 1.002,
+      "step": 3983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.413868503147907e-05,
+      "loss": 1.1318,
+      "step": 3984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41358484654577e-05,
+      "loss": 1.0327,
+      "step": 3985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133011497936995e-05,
+      "loss": 1.2158,
+      "step": 3986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4130174129192367e-05,
+      "loss": 1.1343,
+      "step": 3987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412733635949926e-05,
+      "loss": 1.1738,
+      "step": 3988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4124498189133163e-05,
+      "loss": 1.1326,
+      "step": 3989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4121659618369591e-05,
+      "loss": 1.0653,
+      "step": 3990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118820647484115e-05,
+      "loss": 1.1272,
+      "step": 3991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4115981276752332e-05,
+      "loss": 1.2073,
+      "step": 3992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411314150644988e-05,
+      "loss": 1.035,
+      "step": 3993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411030133685244e-05,
+      "loss": 1.0889,
+      "step": 3994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4107460768235734e-05,
+      "loss": 1.0197,
+      "step": 3995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4104619800875514e-05,
+      "loss": 1.0747,
+      "step": 3996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4101778435047574e-05,
+      "loss": 1.2261,
+      "step": 3997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4098936671027747e-05,
+      "loss": 1.1381,
+      "step": 3998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096094509091911e-05,
+      "loss": 1.1284,
+      "step": 3999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4093251949515972e-05,
+      "loss": 1.073,
+      "step": 4000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4090408992575881e-05,
+      "loss": 1.0854,
+      "step": 4001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4087565638547625e-05,
+      "loss": 1.1113,
+      "step": 4002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4084721887707234e-05,
+      "loss": 1.1013,
+      "step": 4003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4081877740330772e-05,
+      "loss": 1.1206,
+      "step": 4004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4079033196694339e-05,
+      "loss": 1.1165,
+      "step": 4005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4076188257074079e-05,
+      "loss": 1.1592,
+      "step": 4006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4073342921746177e-05,
+      "loss": 0.9884,
+      "step": 4007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407049719098685e-05,
+      "loss": 1.1018,
+      "step": 4008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4067651065072353e-05,
+      "loss": 1.0618,
+      "step": 4009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4064804544278978e-05,
+      "loss": 1.1306,
+      "step": 4010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061957628883072e-05,
+      "loss": 1.1538,
+      "step": 4011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4059110319160998e-05,
+      "loss": 1.1523,
+      "step": 4012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4056262615389168e-05,
+      "loss": 1.0071,
+      "step": 4013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4053414517844028e-05,
+      "loss": 1.1978,
+      "step": 4014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4050566026802073e-05,
+      "loss": 1.0596,
+      "step": 4015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4047717142539823e-05,
+      "loss": 1.1267,
+      "step": 4016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4044867865333843e-05,
+      "loss": 1.1311,
+      "step": 4017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042018195460733e-05,
+      "loss": 1.1243,
+      "step": 4018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039168133197136e-05,
+      "loss": 1.1836,
+      "step": 4019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4036317678819729e-05,
+      "loss": 1.125,
+      "step": 4020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4033466832605224e-05,
+      "loss": 1.1423,
+      "step": 4021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4030615594830383e-05,
+      "loss": 1.1699,
+      "step": 4022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4027763965771993e-05,
+      "loss": 0.9822,
+      "step": 4023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402491194570688e-05,
+      "loss": 1.0673,
+      "step": 4024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4022059534911918e-05,
+      "loss": 1.1362,
+      "step": 4025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4019206733664016e-05,
+      "loss": 1.1033,
+      "step": 4026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016353542240111e-05,
+      "loss": 1.1263,
+      "step": 4027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4013499960917187e-05,
+      "loss": 1.1108,
+      "step": 4028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4010645989972262e-05,
+      "loss": 1.0381,
+      "step": 4029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4007791629682396e-05,
+      "loss": 1.134,
+      "step": 4030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004936880324685e-05,
+      "loss": 1.1387,
+      "step": 4031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400208174217626e-05,
+      "loss": 1.0835,
+      "step": 4032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3999226215514288e-05,
+      "loss": 1.1042,
+      "step": 4033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3996370300615984e-05,
+      "loss": 1.1353,
+      "step": 4034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3993513997758592e-05,
+      "loss": 1.2065,
+      "step": 4035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3990657307219396e-05,
+      "loss": 1.1869,
+      "step": 4036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3987800229275713e-05,
+      "loss": 1.0267,
+      "step": 4037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398494276420491e-05,
+      "loss": 1.0664,
+      "step": 4038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982084912284377e-05,
+      "loss": 1.127,
+      "step": 4039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3979226673791551e-05,
+      "loss": 0.9525,
+      "step": 4040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3976368049003903e-05,
+      "loss": 1.1016,
+      "step": 4041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3973509038198942e-05,
+      "loss": 1.0427,
+      "step": 4042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3970649641654217e-05,
+      "loss": 1.2407,
+      "step": 4043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967789859647308e-05,
+      "loss": 1.0231,
+      "step": 4044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396492969245584e-05,
+      "loss": 1.1328,
+      "step": 4045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3962069140357467e-05,
+      "loss": 1.1184,
+      "step": 4046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3959208203629892e-05,
+      "loss": 1.1704,
+      "step": 4047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3956346882550846e-05,
+      "loss": 1.0492,
+      "step": 4048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3953485177398098e-05,
+      "loss": 1.0173,
+      "step": 4049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.395062308844946e-05,
+      "loss": 1.0913,
+      "step": 4050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394776061598277e-05,
+      "loss": 1.1658,
+      "step": 4051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3944897760275922e-05,
+      "loss": 1.0759,
+      "step": 4052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3942034521606824e-05,
+      "loss": 1.2004,
+      "step": 4053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3939170900253442e-05,
+      "loss": 0.8815,
+      "step": 4054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936306896493768e-05,
+      "loss": 1.2069,
+      "step": 4055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393344251060583e-05,
+      "loss": 1.0886,
+      "step": 4056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3930577742867699e-05,
+      "loss": 1.0875,
+      "step": 4057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392771259355748e-05,
+      "loss": 1.0981,
+      "step": 4058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3924847062953318e-05,
+      "loss": 1.114,
+      "step": 4059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392198115133339e-05,
+      "loss": 1.2144,
+      "step": 4060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919114858975912e-05,
+      "loss": 1.0789,
+      "step": 4061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3916248186159143e-05,
+      "loss": 1.1321,
+      "step": 4062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3913381133161363e-05,
+      "loss": 1.1091,
+      "step": 4063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910513700260908e-05,
+      "loss": 1.1487,
+      "step": 4064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3907645887736138e-05,
+      "loss": 1.1458,
+      "step": 4065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390477769586546e-05,
+      "loss": 1.0551,
+      "step": 4066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3901909124927305e-05,
+      "loss": 1.0835,
+      "step": 4067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899040175200152e-05,
+      "loss": 1.0168,
+      "step": 4068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3896170846962508e-05,
+      "loss": 1.1028,
+      "step": 4069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3893301140492929e-05,
+      "loss": 1.2151,
+      "step": 4070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3890431056069991e-05,
+      "loss": 1.2212,
+      "step": 4071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3887560593972323e-05,
+      "loss": 1.0835,
+      "step": 4072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388468975447858e-05,
+      "loss": 1.0894,
+      "step": 4073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3881818537867455e-05,
+      "loss": 1.2725,
+      "step": 4074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3878946944417682e-05,
+      "loss": 1.1221,
+      "step": 4075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3876074974408029e-05,
+      "loss": 1.0826,
+      "step": 4076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873202628117306e-05,
+      "loss": 1.1438,
+      "step": 4077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3870329905824345e-05,
+      "loss": 1.1279,
+      "step": 4078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3867456807808024e-05,
+      "loss": 1.166,
+      "step": 4079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3864583334347263e-05,
+      "loss": 1.1388,
+      "step": 4080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861709485721016e-05,
+      "loss": 0.9849,
+      "step": 4081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3858835262208259e-05,
+      "loss": 0.8567,
+      "step": 4082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855960664088022e-05,
+      "loss": 1.1089,
+      "step": 4083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3853085691639364e-05,
+      "loss": 1.1169,
+      "step": 4084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3850210345141383e-05,
+      "loss": 1.0925,
+      "step": 4085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3847334624873205e-05,
+      "loss": 1.1458,
+      "step": 4086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3844458531114006e-05,
+      "loss": 1.124,
+      "step": 4087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3841582064142993e-05,
+      "loss": 1.1038,
+      "step": 4088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3838705224239397e-05,
+      "loss": 1.0886,
+      "step": 4089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38358280116825e-05,
+      "loss": 1.1002,
+      "step": 4090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.383295042675162e-05,
+      "loss": 1.1287,
+      "step": 4091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3830072469726105e-05,
+      "loss": 1.1356,
+      "step": 4092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382719414088534e-05,
+      "loss": 1.0698,
+      "step": 4093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3824315440508742e-05,
+      "loss": 1.1516,
+      "step": 4094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3821436368875779e-05,
+      "loss": 1.1084,
+      "step": 4095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3818556926265935e-05,
+      "loss": 1.2046,
+      "step": 4096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3815677112958748e-05,
+      "loss": 1.1028,
+      "step": 4097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3812796929233776e-05,
+      "loss": 1.1293,
+      "step": 4098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3809916375370633e-05,
+      "loss": 1.0933,
+      "step": 4099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3807035451648948e-05,
+      "loss": 1.1843,
+      "step": 4100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3804154158348395e-05,
+      "loss": 1.0155,
+      "step": 4101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3801272495748683e-05,
+      "loss": 1.1075,
+      "step": 4102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3798390464129568e-05,
+      "loss": 1.0994,
+      "step": 4103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795508063770819e-05,
+      "loss": 0.9963,
+      "step": 4104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.379262529495226e-05,
+      "loss": 1.0546,
+      "step": 4105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3789742157953741e-05,
+      "loss": 1.1443,
+      "step": 4106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3786858653055154e-05,
+      "loss": 1.0138,
+      "step": 4107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3783974780536418e-05,
+      "loss": 1.2473,
+      "step": 4108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37810905406775e-05,
+      "loss": 1.0564,
+      "step": 4109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377820593375839e-05,
+      "loss": 1.0714,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3775320960059123e-05,
+      "loss": 1.1755,
+      "step": 4111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3772435619859766e-05,
+      "loss": 1.2036,
+      "step": 4112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769549913440422e-05,
+      "loss": 1.1768,
+      "step": 4113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3766663841081225e-05,
+      "loss": 1.1025,
+      "step": 4114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376377740306235e-05,
+      "loss": 1.0066,
+      "step": 4115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3760890599664012e-05,
+      "loss": 1.1868,
+      "step": 4116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375800343116645e-05,
+      "loss": 1.1025,
+      "step": 4117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3755115897849946e-05,
+      "loss": 1.0767,
+      "step": 4118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3752227999994816e-05,
+      "loss": 1.2021,
+      "step": 4119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374933973788141e-05,
+      "loss": 1.2173,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3746451111790113e-05,
+      "loss": 1.105,
+      "step": 4121
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374356212200135e-05,
+      "loss": 1.1169,
+      "step": 4122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3740672768795576e-05,
+      "loss": 1.1792,
+      "step": 4123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3737783052453284e-05,
+      "loss": 1.1426,
+      "step": 4124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3734892973254998e-05,
+      "loss": 0.991,
+      "step": 4125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732002531481287e-05,
+      "loss": 1.0769,
+      "step": 4126
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3729111727412744e-05,
+      "loss": 1.0325,
+      "step": 4127
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726220561330004e-05,
+      "loss": 1.0576,
+      "step": 4128
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3723329033513735e-05,
+      "loss": 1.0562,
+      "step": 4129
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372043714424464e-05,
+      "loss": 1.1633,
+      "step": 4130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3717544893803458e-05,
+      "loss": 1.1245,
+      "step": 4131
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714652282470961e-05,
+      "loss": 1.1005,
+      "step": 4132
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3711759310527961e-05,
+      "loss": 1.1018,
+      "step": 4133
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3708865978255298e-05,
+      "loss": 1.0874,
+      "step": 4134
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3705972285933851e-05,
+      "loss": 1.0011,
+      "step": 4135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3703078233844534e-05,
+      "loss": 1.0377,
+      "step": 4136
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3700183822268296e-05,
+      "loss": 1.1046,
+      "step": 4137
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369728905148612e-05,
+      "loss": 1.1001,
+      "step": 4138
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3694393921779024e-05,
+      "loss": 1.1675,
+      "step": 4139
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369149843342806e-05,
+      "loss": 1.1433,
+      "step": 4140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3688602586714319e-05,
+      "loss": 1.1141,
+      "step": 4141
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368570638191892e-05,
+      "loss": 1.0718,
+      "step": 4142
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3682809819323023e-05,
+      "loss": 1.1101,
+      "step": 4143
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3679912899207817e-05,
+      "loss": 1.0808,
+      "step": 4144
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3677015621854533e-05,
+      "loss": 1.2556,
+      "step": 4145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3674117987544429e-05,
+      "loss": 1.1328,
+      "step": 4146
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3671219996558805e-05,
+      "loss": 1.0922,
+      "step": 4147
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3668321649178983e-05,
+      "loss": 1.0787,
+      "step": 4148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3665422945686339e-05,
+      "loss": 1.1932,
+      "step": 4149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3662523886362267e-05,
+      "loss": 1.1226,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3659624471488206e-05,
+      "loss": 1.075,
+      "step": 4151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3656724701345616e-05,
+      "loss": 1.1016,
+      "step": 4152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3653824576216011e-05,
+      "loss": 1.0708,
+      "step": 4153
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3650924096380927e-05,
+      "loss": 0.9995,
+      "step": 4154
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3648023262121925e-05,
+      "loss": 1.1482,
+      "step": 4155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3645122073720628e-05,
+      "loss": 1.1487,
+      "step": 4156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3642220531458667e-05,
+      "loss": 1.0664,
+      "step": 4157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3639318635617722e-05,
+      "loss": 1.1013,
+      "step": 4158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3636416386479498e-05,
+      "loss": 1.1049,
+      "step": 4159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3633513784325744e-05,
+      "loss": 1.0273,
+      "step": 4160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3630610829438235e-05,
+      "loss": 1.1289,
+      "step": 4161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627707522098792e-05,
+      "loss": 1.1042,
+      "step": 4162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3624803862589246e-05,
+      "loss": 1.0959,
+      "step": 4163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3621899851191494e-05,
+      "loss": 1.0211,
+      "step": 4164
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3618995488187443e-05,
+      "loss": 1.2073,
+      "step": 4165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361609077385904e-05,
+      "loss": 1.1042,
+      "step": 4166
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3613185708488275e-05,
+      "loss": 1.0677,
+      "step": 4167
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3610280292357163e-05,
+      "loss": 1.1277,
+      "step": 4168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3607374525747758e-05,
+      "loss": 1.1731,
+      "step": 4169
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360446840894214e-05,
+      "loss": 1.1086,
+      "step": 4170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360156194222243e-05,
+      "loss": 1.1283,
+      "step": 4171
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3598655125870783e-05,
+      "loss": 1.1216,
+      "step": 4172
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3595747960169393e-05,
+      "loss": 1.0701,
+      "step": 4173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359284044540047e-05,
+      "loss": 1.1115,
+      "step": 4174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3589932581846275e-05,
+      "loss": 1.1215,
+      "step": 4175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3587024369789097e-05,
+      "loss": 1.1604,
+      "step": 4176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358411580951126e-05,
+      "loss": 0.9368,
+      "step": 4177
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3581206901295119e-05,
+      "loss": 1.1792,
+      "step": 4178
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3578297645423062e-05,
+      "loss": 1.1819,
+      "step": 4179
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357538804217752e-05,
+      "loss": 1.0779,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3572478091840946e-05,
+      "loss": 1.1382,
+      "step": 4181
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3569567794695835e-05,
+      "loss": 1.1565,
+      "step": 4182
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3566657151024706e-05,
+      "loss": 1.1268,
+      "step": 4183
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356374616111013e-05,
+      "loss": 1.1267,
+      "step": 4184
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3560834825234689e-05,
+      "loss": 1.1833,
+      "step": 4185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355792314368101e-05,
+      "loss": 1.1184,
+      "step": 4186
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3555011116731758e-05,
+      "loss": 1.0145,
+      "step": 4187
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3552098744669624e-05,
+      "loss": 1.1289,
+      "step": 4188
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3549186027777335e-05,
+      "loss": 1.2424,
+      "step": 4189
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546272966337647e-05,
+      "loss": 1.0945,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3543359560633365e-05,
+      "loss": 1.1648,
+      "step": 4191
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3540445810947305e-05,
+      "loss": 1.0667,
+      "step": 4192
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3537531717562332e-05,
+      "loss": 1.0336,
+      "step": 4193
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3534617280761334e-05,
+      "loss": 1.0537,
+      "step": 4194
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3531702500827253e-05,
+      "loss": 1.0416,
+      "step": 4195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3528787378043038e-05,
+      "loss": 1.1956,
+      "step": 4196
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3525871912691683e-05,
+      "loss": 1.0684,
+      "step": 4197
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.352295610505622e-05,
+      "loss": 1.0769,
+      "step": 4198
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520039955419706e-05,
+      "loss": 1.1833,
+      "step": 4199
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517123464065234e-05,
+      "loss": 1.0774,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3514206631275934e-05,
+      "loss": 1.1958,
+      "step": 4201
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3511289457334964e-05,
+      "loss": 1.1028,
+      "step": 4202
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3508371942525516e-05,
+      "loss": 1.1104,
+      "step": 4203
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3505454087130819e-05,
+      "loss": 1.1038,
+      "step": 4204
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3502535891434127e-05,
+      "loss": 1.1267,
+      "step": 4205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3499617355718739e-05,
+      "loss": 1.1553,
+      "step": 4206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3496698480267976e-05,
+      "loss": 1.0837,
+      "step": 4207
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3493779265365197e-05,
+      "loss": 1.0712,
+      "step": 4208
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3490859711293794e-05,
+      "loss": 1.0646,
+      "step": 4209
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3487939818337185e-05,
+      "loss": 0.9551,
+      "step": 4210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3485019586778838e-05,
+      "loss": 1.038,
+      "step": 4211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3482099016902234e-05,
+      "loss": 1.1572,
+      "step": 4212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34791781089909e-05,
+      "loss": 1.0741,
+      "step": 4213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3476256863328392e-05,
+      "loss": 1.0399,
+      "step": 4214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3473335280198295e-05,
+      "loss": 1.2273,
+      "step": 4215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3470413359884231e-05,
+      "loss": 1.1182,
+      "step": 4216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467491102669854e-05,
+      "loss": 1.1057,
+      "step": 4217
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3464568508838852e-05,
+      "loss": 1.071,
+      "step": 4218
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3461645578674942e-05,
+      "loss": 1.0416,
+      "step": 4219
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3458722312461876e-05,
+      "loss": 1.1353,
+      "step": 4220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345579871048344e-05,
+      "loss": 1.2231,
+      "step": 4221
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3452874773023452e-05,
+      "loss": 1.1707,
+      "step": 4222
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3449950500365759e-05,
+      "loss": 1.1267,
+      "step": 4223
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3447025892794243e-05,
+      "loss": 1.1948,
+      "step": 4224
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344410095059282e-05,
+      "loss": 1.0648,
+      "step": 4225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3441175674045437e-05,
+      "loss": 1.1021,
+      "step": 4226
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438250063436074e-05,
+      "loss": 1.1968,
+      "step": 4227
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343532411904874e-05,
+      "loss": 1.1101,
+      "step": 4228
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3432397841167483e-05,
+      "loss": 1.0111,
+      "step": 4229
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3429471230076381e-05,
+      "loss": 1.1594,
+      "step": 4230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342654428605954e-05,
+      "loss": 0.9968,
+      "step": 4231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34236170094011e-05,
+      "loss": 1.158,
+      "step": 4232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3420689400385239e-05,
+      "loss": 1.1855,
+      "step": 4233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3417761459296159e-05,
+      "loss": 1.1758,
+      "step": 4234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3414833186418104e-05,
+      "loss": 0.9895,
+      "step": 4235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411904582035336e-05,
+      "loss": 1.1892,
+      "step": 4236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3408975646432164e-05,
+      "loss": 1.0547,
+      "step": 4237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340604637989292e-05,
+      "loss": 1.0802,
+      "step": 4238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3403116782701977e-05,
+      "loss": 1.1963,
+      "step": 4239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3400186855143726e-05,
+      "loss": 1.1013,
+      "step": 4240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3397256597502602e-05,
+      "loss": 1.0894,
+      "step": 4241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3394326010063067e-05,
+      "loss": 1.1765,
+      "step": 4242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3391395093109619e-05,
+      "loss": 1.071,
+      "step": 4243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388463846926778e-05,
+      "loss": 1.1589,
+      "step": 4244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338553227179911e-05,
+      "loss": 1.0436,
+      "step": 4245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3382600368011207e-05,
+      "loss": 1.0251,
+      "step": 4246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3379668135847687e-05,
+      "loss": 1.2617,
+      "step": 4247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3376735575593202e-05,
+      "loss": 1.2178,
+      "step": 4248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3373802687532448e-05,
+      "loss": 1.1316,
+      "step": 4249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3370869471950142e-05,
+      "loss": 0.9681,
+      "step": 4250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3367935929131028e-05,
+      "loss": 1.0933,
+      "step": 4251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3365002059359888e-05,
+      "loss": 1.1184,
+      "step": 4252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3362067862921546e-05,
+      "loss": 1.1387,
+      "step": 4253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3359133340100836e-05,
+      "loss": 1.1484,
+      "step": 4254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335619849118264e-05,
+      "loss": 1.0916,
+      "step": 4255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3353263316451865e-05,
+      "loss": 1.1123,
+      "step": 4256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3350327816193457e-05,
+      "loss": 1.0615,
+      "step": 4257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347391990692384e-05,
+      "loss": 1.0471,
+      "step": 4258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3344455840233647e-05,
+      "loss": 1.0488,
+      "step": 4259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3341519365102286e-05,
+      "loss": 1.1016,
+      "step": 4260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333858256558337e-05,
+      "loss": 1.0845,
+      "step": 4261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335645441961986e-05,
+      "loss": 1.0146,
+      "step": 4262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3332707994523273e-05,
+      "loss": 1.0908,
+      "step": 4263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329770223552392e-05,
+      "loss": 1.0525,
+      "step": 4264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3326832129334533e-05,
+      "loss": 1.0356,
+      "step": 4265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3323893712154918e-05,
+      "loss": 1.3489,
+      "step": 4266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3320954972298806e-05,
+      "loss": 1.113,
+      "step": 4267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3318015910051483e-05,
+      "loss": 1.051,
+      "step": 4268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3315076525698264e-05,
+      "loss": 1.0997,
+      "step": 4269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3312136819524503e-05,
+      "loss": 1.0333,
+      "step": 4270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3309196791815574e-05,
+      "loss": 1.1118,
+      "step": 4271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3306256442856897e-05,
+      "loss": 1.2739,
+      "step": 4272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3303315772933907e-05,
+      "loss": 1.0466,
+      "step": 4273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3300374782332081e-05,
+      "loss": 1.0671,
+      "step": 4274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3297433471336923e-05,
+      "loss": 1.1228,
+      "step": 4275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3294491840233975e-05,
+      "loss": 1.1597,
+      "step": 4276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3291549889308794e-05,
+      "loss": 1.1442,
+      "step": 4277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3288607618846987e-05,
+      "loss": 1.0818,
+      "step": 4278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3285665029134177e-05,
+      "loss": 1.0717,
+      "step": 4279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3282722120456031e-05,
+      "loss": 0.985,
+      "step": 4280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3279778893098237e-05,
+      "loss": 1.1455,
+      "step": 4281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3276835347346513e-05,
+      "loss": 1.0747,
+      "step": 4282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327389148348662e-05,
+      "loss": 1.0679,
+      "step": 4283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3270947301804336e-05,
+      "loss": 0.9788,
+      "step": 4284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3268002802585477e-05,
+      "loss": 1.1194,
+      "step": 4285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3265057986115893e-05,
+      "loss": 1.1355,
+      "step": 4286
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3262112852681455e-05,
+      "loss": 1.0762,
+      "step": 4287
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259167402568075e-05,
+      "loss": 1.1714,
+      "step": 4288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3256221636061683e-05,
+      "loss": 0.9458,
+      "step": 4289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3253275553448257e-05,
+      "loss": 1.1107,
+      "step": 4290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3250329155013792e-05,
+      "loss": 1.0013,
+      "step": 4291
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3247382441044318e-05,
+      "loss": 1.0514,
+      "step": 4292
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3244435411825898e-05,
+      "loss": 1.2117,
+      "step": 4293
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324148806764462e-05,
+      "loss": 0.9963,
+      "step": 4294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323854040878661e-05,
+      "loss": 1.1206,
+      "step": 4295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3235592435538013e-05,
+      "loss": 1.0012,
+      "step": 4296
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3232644148185019e-05,
+      "loss": 1.2729,
+      "step": 4297
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3229695547013841e-05,
+      "loss": 1.1167,
+      "step": 4298
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3226746632310721e-05,
+      "loss": 1.1173,
+      "step": 4299
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.322379740436193e-05,
+      "loss": 1.2134,
+      "step": 4300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3220847863453781e-05,
+      "loss": 1.0879,
+      "step": 4301
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3217898009872601e-05,
+      "loss": 1.1636,
+      "step": 4302
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3214947843904761e-05,
+      "loss": 1.0935,
+      "step": 4303
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3211997365836652e-05,
+      "loss": 1.1689,
+      "step": 4304
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3209046575954707e-05,
+      "loss": 1.1593,
+      "step": 4305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3206095474545376e-05,
+      "loss": 1.1287,
+      "step": 4306
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3203144061895146e-05,
+      "loss": 0.9636,
+      "step": 4307
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3200192338290538e-05,
+      "loss": 1.151,
+      "step": 4308
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3197240304018094e-05,
+      "loss": 1.1167,
+      "step": 4309
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3194287959364396e-05,
+      "loss": 1.1365,
+      "step": 4310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3191335304616049e-05,
+      "loss": 1.1841,
+      "step": 4311
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3188382340059693e-05,
+      "loss": 1.1082,
+      "step": 4312
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3185429065981988e-05,
+      "loss": 1.1323,
+      "step": 4313
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3182475482669639e-05,
+      "loss": 1.1909,
+      "step": 4314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317952159040937e-05,
+      "loss": 1.1208,
+      "step": 4315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3176567389487941e-05,
+      "loss": 1.2014,
+      "step": 4316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173612880192137e-05,
+      "loss": 0.9679,
+      "step": 4317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170658062808778e-05,
+      "loss": 1.1323,
+      "step": 4318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316770293762471e-05,
+      "loss": 1.0676,
+      "step": 4319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316474750492681e-05,
+      "loss": 1.1157,
+      "step": 4320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3161791765001982e-05,
+      "loss": 1.1091,
+      "step": 4321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3158835718137171e-05,
+      "loss": 1.0388,
+      "step": 4322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.315587936461934e-05,
+      "loss": 1.0869,
+      "step": 4323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3152922704735477e-05,
+      "loss": 1.0657,
+      "step": 4324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3149965738772621e-05,
+      "loss": 1.0468,
+      "step": 4325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3147008467017824e-05,
+      "loss": 1.0991,
+      "step": 4326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3144050889758168e-05,
+      "loss": 1.2656,
+      "step": 4327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314109300728077e-05,
+      "loss": 1.132,
+      "step": 4328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3138134819872775e-05,
+      "loss": 1.0686,
+      "step": 4329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135176327821358e-05,
+      "loss": 0.9951,
+      "step": 4330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3132217531413725e-05,
+      "loss": 1.1958,
+      "step": 4331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3129258430937104e-05,
+      "loss": 1.1,
+      "step": 4332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3126299026678764e-05,
+      "loss": 1.0112,
+      "step": 4333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3123339318925997e-05,
+      "loss": 1.1487,
+      "step": 4334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120379307966124e-05,
+      "loss": 1.1737,
+      "step": 4335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311741899408649e-05,
+      "loss": 1.0176,
+      "step": 4336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3114458377574489e-05,
+      "loss": 1.0962,
+      "step": 4337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3111497458717521e-05,
+      "loss": 1.1104,
+      "step": 4338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310853623780303e-05,
+      "loss": 1.092,
+      "step": 4339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105574715118481e-05,
+      "loss": 1.1116,
+      "step": 4340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102612890951379e-05,
+      "loss": 1.0903,
+      "step": 4341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3099650765589246e-05,
+      "loss": 1.0938,
+      "step": 4342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3096688339319644e-05,
+      "loss": 1.084,
+      "step": 4343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309372561243015e-05,
+      "loss": 1.0459,
+      "step": 4344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309076258520839e-05,
+      "loss": 1.0752,
+      "step": 4345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087799257942e-05,
+      "loss": 1.0496,
+      "step": 4346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3084835630918657e-05,
+      "loss": 1.157,
+      "step": 4347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3081871704426063e-05,
+      "loss": 1.0513,
+      "step": 4348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3078907478751951e-05,
+      "loss": 1.0653,
+      "step": 4349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3075942954184082e-05,
+      "loss": 1.0681,
+      "step": 4350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3072978131010244e-05,
+      "loss": 1.1018,
+      "step": 4351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307001300951825e-05,
+      "loss": 1.0299,
+      "step": 4352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306704758999596e-05,
+      "loss": 1.1675,
+      "step": 4353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3064081872731245e-05,
+      "loss": 1.2004,
+      "step": 4354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3061115858012006e-05,
+      "loss": 1.0114,
+      "step": 4355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3058149546126182e-05,
+      "loss": 1.0775,
+      "step": 4356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3055182937361737e-05,
+      "loss": 1.1318,
+      "step": 4357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.305221603200666e-05,
+      "loss": 1.1514,
+      "step": 4358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304924883034897e-05,
+      "loss": 1.0623,
+      "step": 4359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3046281332676727e-05,
+      "loss": 1.1411,
+      "step": 4360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3043313539278e-05,
+      "loss": 1.0872,
+      "step": 4361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040345450440897e-05,
+      "loss": 1.0527,
+      "step": 4362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3037377066453552e-05,
+      "loss": 1.064,
+      "step": 4363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034408387604135e-05,
+      "loss": 1.1753,
+      "step": 4364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3031439414180838e-05,
+      "loss": 1.1738,
+      "step": 4365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3028470146471878e-05,
+      "loss": 1.1245,
+      "step": 4366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302550058476551e-05,
+      "loss": 1.0569,
+      "step": 4367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302253072935001e-05,
+      "loss": 1.0936,
+      "step": 4368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3019560580513684e-05,
+      "loss": 0.9915,
+      "step": 4369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016590138544874e-05,
+      "loss": 1.0811,
+      "step": 4370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3013619403731936e-05,
+      "loss": 1.0996,
+      "step": 4371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3010648376363266e-05,
+      "loss": 0.9961,
+      "step": 4372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3007677056727289e-05,
+      "loss": 1.1178,
+      "step": 4373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3004705445112449e-05,
+      "loss": 1.0649,
+      "step": 4374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3001733541807225e-05,
+      "loss": 1.1062,
+      "step": 4375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2998761347100124e-05,
+      "loss": 1.031,
+      "step": 4376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299578886127968e-05,
+      "loss": 1.0831,
+      "step": 4377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2992816084634454e-05,
+      "loss": 1.0594,
+      "step": 4378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298984301745304e-05,
+      "loss": 1.0045,
+      "step": 4379
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2986869660024052e-05,
+      "loss": 1.0703,
+      "step": 4380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2983896012636143e-05,
+      "loss": 1.0647,
+      "step": 4381
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2980922075577987e-05,
+      "loss": 0.988,
+      "step": 4382
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977947849138283e-05,
+      "loss": 1.0261,
+      "step": 4383
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2974973333605767e-05,
+      "loss": 1.1099,
+      "step": 4384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2971998529269198e-05,
+      "loss": 1.0063,
+      "step": 4385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969023436417363e-05,
+      "loss": 1.0394,
+      "step": 4386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2966048055339077e-05,
+      "loss": 1.1069,
+      "step": 4387
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2963072386323185e-05,
+      "loss": 1.1758,
+      "step": 4388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2960096429658557e-05,
+      "loss": 1.1021,
+      "step": 4389
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2957120185634094e-05,
+      "loss": 1.0879,
+      "step": 4390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2954143654538723e-05,
+      "loss": 0.9502,
+      "step": 4391
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2951166836661397e-05,
+      "loss": 1.1821,
+      "step": 4392
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2948189732291104e-05,
+      "loss": 1.1697,
+      "step": 4393
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2945212341716849e-05,
+      "loss": 1.0657,
+      "step": 4394
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2942234665227675e-05,
+      "loss": 1.0747,
+      "step": 4395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2939256703112649e-05,
+      "loss": 1.0684,
+      "step": 4396
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2936278455660862e-05,
+      "loss": 1.1149,
+      "step": 4397
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2933299923161437e-05,
+      "loss": 1.1199,
+      "step": 4398
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2930321105903524e-05,
+      "loss": 1.0204,
+      "step": 4399
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2927342004176305e-05,
+      "loss": 1.2407,
+      "step": 4400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2924362618268971e-05,
+      "loss": 1.1665,
+      "step": 4401
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2921382948470769e-05,
+      "loss": 1.0771,
+      "step": 4402
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2918402995070954e-05,
+      "loss": 1.1235,
+      "step": 4403
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2915422758358811e-05,
+      "loss": 1.0791,
+      "step": 4404
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2912442238623655e-05,
+      "loss": 1.0242,
+      "step": 4405
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2909461436154833e-05,
+      "loss": 1.2515,
+      "step": 4406
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2906480351241713e-05,
+      "loss": 1.0474,
+      "step": 4407
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2903498984173692e-05,
+      "loss": 1.0989,
+      "step": 4408
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290051733524019e-05,
+      "loss": 1.1567,
+      "step": 4409
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2897535404730667e-05,
+      "loss": 1.2153,
+      "step": 4410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.28945531929346e-05,
+      "loss": 1.0962,
+      "step": 4411
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2891570700141496e-05,
+      "loss": 1.1367,
+      "step": 4412
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2888587926640881e-05,
+      "loss": 1.0354,
+      "step": 4413
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2885604872722328e-05,
+      "loss": 1.1401,
+      "step": 4414
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2882621538675422e-05,
+      "loss": 1.1985,
+      "step": 4415
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2879637924789775e-05,
+      "loss": 1.1572,
+      "step": 4416
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.287665403135503e-05,
+      "loss": 1.1465,
+      "step": 4417
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2873669858660858e-05,
+      "loss": 1.2397,
+      "step": 4418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2870685406996964e-05,
+      "loss": 1.1699,
+      "step": 4419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2867700676653062e-05,
+      "loss": 1.1593,
+      "step": 4420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2864715667918901e-05,
+      "loss": 0.9923,
+      "step": 4421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2861730381084275e-05,
+      "loss": 1.1129,
+      "step": 4422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2858744816438972e-05,
+      "loss": 0.9962,
+      "step": 4423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2855758974272832e-05,
+      "loss": 0.9523,
+      "step": 4424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2852772854875715e-05,
+      "loss": 1.1665,
+      "step": 4425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2849786458537507e-05,
+      "loss": 1.1194,
+      "step": 4426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284679978554812e-05,
+      "loss": 1.0159,
+      "step": 4427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2843812836197488e-05,
+      "loss": 1.1328,
+      "step": 4428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2840825610775586e-05,
+      "loss": 1.062,
+      "step": 4429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2837838109572407e-05,
+      "loss": 1.0886,
+      "step": 4430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2834850332877968e-05,
+      "loss": 1.1191,
+      "step": 4431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2831862280982314e-05,
+      "loss": 1.1012,
+      "step": 4432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2828873954175523e-05,
+      "loss": 1.0691,
+      "step": 4433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825885352747695e-05,
+      "loss": 1.0811,
+      "step": 4434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2822896476988957e-05,
+      "loss": 1.0737,
+      "step": 4435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2819907327189457e-05,
+      "loss": 0.9824,
+      "step": 4436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2816917903639386e-05,
+      "loss": 1.155,
+      "step": 4437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281392820662894e-05,
+      "loss": 1.104,
+      "step": 4438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2810938236448358e-05,
+      "loss": 1.0259,
+      "step": 4439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2807947993387898e-05,
+      "loss": 1.1296,
+      "step": 4440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2804957477737849e-05,
+      "loss": 1.0854,
+      "step": 4441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2801966689788523e-05,
+      "loss": 1.1169,
+      "step": 4442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2798975629830258e-05,
+      "loss": 1.0481,
+      "step": 4443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2795984298153416e-05,
+      "loss": 1.1282,
+      "step": 4444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27929926950484e-05,
+      "loss": 1.0066,
+      "step": 4445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2790000820805618e-05,
+      "loss": 1.126,
+      "step": 4446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2787008675715523e-05,
+      "loss": 1.2161,
+      "step": 4447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2784016260068577e-05,
+      "loss": 1.134,
+      "step": 4448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2781023574155283e-05,
+      "loss": 1.0977,
+      "step": 4449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2778030618266164e-05,
+      "loss": 1.0601,
+      "step": 4450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2775037392691769e-05,
+      "loss": 1.0671,
+      "step": 4451
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2772043897722672e-05,
+      "loss": 1.0271,
+      "step": 4452
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.276905013364948e-05,
+      "loss": 1.0991,
+      "step": 4453
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2766056100762818e-05,
+      "loss": 1.1582,
+      "step": 4454
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2763061799353342e-05,
+      "loss": 1.1538,
+      "step": 4455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.276006722971173e-05,
+      "loss": 1.1179,
+      "step": 4456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2757072392128689e-05,
+      "loss": 1.1055,
+      "step": 4457
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2754077286894953e-05,
+      "loss": 0.9972,
+      "step": 4458
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2751081914301277e-05,
+      "loss": 1.1213,
+      "step": 4459
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748086274638452e-05,
+      "loss": 1.0574,
+      "step": 4460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274509036819728e-05,
+      "loss": 1.0801,
+      "step": 4461
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2742094195268604e-05,
+      "loss": 1.02,
+      "step": 4462
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2739097756143285e-05,
+      "loss": 1.033,
+      "step": 4463
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2736101051112207e-05,
+      "loss": 1.1404,
+      "step": 4464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2733104080466285e-05,
+      "loss": 1.0708,
+      "step": 4465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730106844496461e-05,
+      "loss": 1.1377,
+      "step": 4466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2727109343493698e-05,
+      "loss": 1.0847,
+      "step": 4467
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2724111577748987e-05,
+      "loss": 1.1538,
+      "step": 4468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2721113547553347e-05,
+      "loss": 1.1018,
+      "step": 4469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271811525319782e-05,
+      "loss": 1.0911,
+      "step": 4470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271511669497347e-05,
+      "loss": 1.0135,
+      "step": 4471
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2712117873171394e-05,
+      "loss": 1.1038,
+      "step": 4472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2709118788082713e-05,
+      "loss": 1.0928,
+      "step": 4473
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2706119439998568e-05,
+      "loss": 1.1348,
+      "step": 4474
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2703119829210131e-05,
+      "loss": 1.0854,
+      "step": 4475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2700119956008597e-05,
+      "loss": 1.0979,
+      "step": 4476
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697119820685189e-05,
+      "loss": 0.9458,
+      "step": 4477
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269411942353115e-05,
+      "loss": 1.093,
+      "step": 4478
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2691118764837758e-05,
+      "loss": 0.921,
+      "step": 4479
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2688117844896306e-05,
+      "loss": 1.2568,
+      "step": 4480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2685116663998118e-05,
+      "loss": 1.2126,
+      "step": 4481
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2682115222434541e-05,
+      "loss": 1.0,
+      "step": 4482
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2679113520496951e-05,
+      "loss": 1.1002,
+      "step": 4483
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2676111558476744e-05,
+      "loss": 1.0613,
+      "step": 4484
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673109336665353e-05,
+      "loss": 1.0615,
+      "step": 4485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2670106855354211e-05,
+      "loss": 0.9694,
+      "step": 4486
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2667104114834805e-05,
+      "loss": 1.0889,
+      "step": 4487
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2664101115398633e-05,
+      "loss": 1.05,
+      "step": 4488
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2661097857337216e-05,
+      "loss": 0.9351,
+      "step": 4489
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2658094340942105e-05,
+      "loss": 1.0745,
+      "step": 4490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2655090566504877e-05,
+      "loss": 1.1492,
+      "step": 4491
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2652086534317132e-05,
+      "loss": 1.095,
+      "step": 4492
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2649082244670493e-05,
+      "loss": 1.0759,
+      "step": 4493
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264607769785661e-05,
+      "loss": 1.0909,
+      "step": 4494
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2643072894167157e-05,
+      "loss": 1.2266,
+      "step": 4495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2640067833893839e-05,
+      "loss": 1.0138,
+      "step": 4496
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2637062517328373e-05,
+      "loss": 1.0938,
+      "step": 4497
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2634056944762519e-05,
+      "loss": 1.0641,
+      "step": 4498
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2631051116488043e-05,
+      "loss": 1.0085,
+      "step": 4499
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262804503279675e-05,
+      "loss": 1.198,
+      "step": 4500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2625038693980454e-05,
+      "loss": 1.1248,
+      "step": 4501
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2622032100331017e-05,
+      "loss": 1.1497,
+      "step": 4502
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261902525214031e-05,
+      "loss": 1.0367,
+      "step": 4503
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2616018149700222e-05,
+      "loss": 1.1345,
+      "step": 4504
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2613010793302683e-05,
+      "loss": 1.0508,
+      "step": 4505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2610003183239643e-05,
+      "loss": 1.0376,
+      "step": 4506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2606995319803072e-05,
+      "loss": 1.1453,
+      "step": 4507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2603987203284966e-05,
+      "loss": 1.0654,
+      "step": 4508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2600978833977344e-05,
+      "loss": 1.1326,
+      "step": 4509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2597970212172256e-05,
+      "loss": 0.9796,
+      "step": 4510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2594961338161777e-05,
+      "loss": 1.1091,
+      "step": 4511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2591952212237992e-05,
+      "loss": 0.9895,
+      "step": 4512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2588942834693023e-05,
+      "loss": 1.0586,
+      "step": 4513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2585933205819023e-05,
+      "loss": 1.0864,
+      "step": 4514
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258292332590815e-05,
+      "loss": 1.1204,
+      "step": 4515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.25799131952526e-05,
+      "loss": 1.1497,
+      "step": 4516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2576902814144588e-05,
+      "loss": 1.1531,
+      "step": 4517
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2573892182876361e-05,
+      "loss": 1.1283,
+      "step": 4518
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2570881301740181e-05,
+      "loss": 1.0765,
+      "step": 4519
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2567870171028337e-05,
+      "loss": 1.105,
+      "step": 4520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2564858791033143e-05,
+      "loss": 1.1287,
+      "step": 4521
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2561847162046943e-05,
+      "loss": 1.0872,
+      "step": 4522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2558835284362093e-05,
+      "loss": 1.1384,
+      "step": 4523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2555823158270982e-05,
+      "loss": 1.0991,
+      "step": 4524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2552810784066019e-05,
+      "loss": 1.111,
+      "step": 4525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2549798162039642e-05,
+      "loss": 1.0864,
+      "step": 4526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2546785292484307e-05,
+      "loss": 1.0911,
+      "step": 4527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2543772175692497e-05,
+      "loss": 1.135,
+      "step": 4528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2540758811956725e-05,
+      "loss": 1.1345,
+      "step": 4529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537745201569516e-05,
+      "loss": 1.2559,
+      "step": 4530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2534731344823423e-05,
+      "loss": 1.1704,
+      "step": 4531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2531717242011027e-05,
+      "loss": 1.0416,
+      "step": 4532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2528702893424938e-05,
+      "loss": 1.1667,
+      "step": 4533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2525688299357773e-05,
+      "loss": 1.0781,
+      "step": 4534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2522673460102184e-05,
+      "loss": 1.1299,
+      "step": 4535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2519658375950845e-05,
+      "loss": 0.9705,
+      "step": 4536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2516643047196458e-05,
+      "loss": 1.0905,
+      "step": 4537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251362747413174e-05,
+      "loss": 1.0206,
+      "step": 4538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251061165704944e-05,
+      "loss": 1.0408,
+      "step": 4539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2507595596242326e-05,
+      "loss": 1.2079,
+      "step": 4540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2504579292003189e-05,
+      "loss": 1.1318,
+      "step": 4541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2501562744624846e-05,
+      "loss": 1.2439,
+      "step": 4542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249854595440014e-05,
+      "loss": 1.104,
+      "step": 4543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249552892162193e-05,
+      "loss": 1.1565,
+      "step": 4544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492511646583108e-05,
+      "loss": 1.1909,
+      "step": 4545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2489494129576578e-05,
+      "loss": 1.1318,
+      "step": 4546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.248647637089528e-05,
+      "loss": 1.0366,
+      "step": 4547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2483458370832171e-05,
+      "loss": 1.0865,
+      "step": 4548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2480440129680229e-05,
+      "loss": 1.1101,
+      "step": 4549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2477421647732459e-05,
+      "loss": 0.968,
+      "step": 4550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247440292528189e-05,
+      "loss": 1.1963,
+      "step": 4551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2471383962621575e-05,
+      "loss": 1.1224,
+      "step": 4552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2468364760044585e-05,
+      "loss": 1.0203,
+      "step": 4553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465345317844018e-05,
+      "loss": 0.9467,
+      "step": 4554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2462325636312996e-05,
+      "loss": 1.1877,
+      "step": 4555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2459305715744665e-05,
+      "loss": 1.0399,
+      "step": 4556
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2456285556432188e-05,
+      "loss": 1.1318,
+      "step": 4557
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2453265158668762e-05,
+      "loss": 1.1333,
+      "step": 4558
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2450244522747595e-05,
+      "loss": 1.0334,
+      "step": 4559
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2447223648961924e-05,
+      "loss": 1.1631,
+      "step": 4560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2444202537605012e-05,
+      "loss": 0.9922,
+      "step": 4561
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2441181188970144e-05,
+      "loss": 1.1194,
+      "step": 4562
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.243815960335062e-05,
+      "loss": 1.1868,
+      "step": 4563
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2435137781039772e-05,
+      "loss": 1.1038,
+      "step": 4564
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432115722330953e-05,
+      "loss": 1.2673,
+      "step": 4565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2429093427517537e-05,
+      "loss": 0.9961,
+      "step": 4566
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242607089689292e-05,
+      "loss": 1.0754,
+      "step": 4567
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242304813075053e-05,
+      "loss": 1.1393,
+      "step": 4568
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2420025129383803e-05,
+      "loss": 1.0797,
+      "step": 4569
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417001893086207e-05,
+      "loss": 0.9199,
+      "step": 4570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2413978422151231e-05,
+      "loss": 1.2195,
+      "step": 4571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2410954716872392e-05,
+      "loss": 1.1331,
+      "step": 4572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240793077754322e-05,
+      "loss": 1.0815,
+      "step": 4573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2404906604457272e-05,
+      "loss": 1.1274,
+      "step": 4574
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2401882197908133e-05,
+      "loss": 1.1477,
+      "step": 4575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2398857558189401e-05,
+      "loss": 1.0981,
+      "step": 4576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2395832685594707e-05,
+      "loss": 1.0442,
+      "step": 4577
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392807580417689e-05,
+      "loss": 1.0571,
+      "step": 4578
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2389782242952031e-05,
+      "loss": 1.1113,
+      "step": 4579
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2386756673491415e-05,
+      "loss": 0.9989,
+      "step": 4580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2383730872329566e-05,
+      "loss": 1.0288,
+      "step": 4581
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2380704839760216e-05,
+      "loss": 1.1223,
+      "step": 4582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2377678576077128e-05,
+      "loss": 0.9747,
+      "step": 4583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2374652081574086e-05,
+      "loss": 1.1426,
+      "step": 4584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2371625356544894e-05,
+      "loss": 1.2805,
+      "step": 4585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2368598401283378e-05,
+      "loss": 1.1208,
+      "step": 4586
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2365571216083393e-05,
+      "loss": 1.1348,
+      "step": 4587
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236254380123881e-05,
+      "loss": 1.2256,
+      "step": 4588
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2359516157043525e-05,
+      "loss": 1.0457,
+      "step": 4589
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356488283791448e-05,
+      "loss": 1.0474,
+      "step": 4590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2353460181776528e-05,
+      "loss": 1.1274,
+      "step": 4591
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2350431851292726e-05,
+      "loss": 1.1433,
+      "step": 4592
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2347403292634016e-05,
+      "loss": 1.0833,
+      "step": 4593
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2344374506094416e-05,
+      "loss": 1.1616,
+      "step": 4594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2341345491967948e-05,
+      "loss": 1.0616,
+      "step": 4595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2338316250548665e-05,
+      "loss": 1.2139,
+      "step": 4596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335286782130634e-05,
+      "loss": 1.0682,
+      "step": 4597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233225708700795e-05,
+      "loss": 1.0906,
+      "step": 4598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232922716547474e-05,
+      "loss": 1.0186,
+      "step": 4599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232619701782513e-05,
+      "loss": 1.0544,
+      "step": 4600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2323166644353282e-05,
+      "loss": 1.0942,
+      "step": 4601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2320136045353384e-05,
+      "loss": 1.0059,
+      "step": 4602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2317105221119638e-05,
+      "loss": 1.0212,
+      "step": 4603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2314074171946267e-05,
+      "loss": 0.9808,
+      "step": 4604
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2311042898127519e-05,
+      "loss": 1.0979,
+      "step": 4605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2308011399957668e-05,
+      "loss": 1.0125,
+      "step": 4606
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2304979677731001e-05,
+      "loss": 1.0391,
+      "step": 4607
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2301947731741834e-05,
+      "loss": 1.2117,
+      "step": 4608
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2298915562284497e-05,
+      "loss": 1.0686,
+      "step": 4609
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2295883169653356e-05,
+      "loss": 0.9985,
+      "step": 4610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229285055414278e-05,
+      "loss": 1.0992,
+      "step": 4611
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228981771604717e-05,
+      "loss": 1.0499,
+      "step": 4612
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228678465566095e-05,
+      "loss": 1.0562,
+      "step": 4613
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283751373278565e-05,
+      "loss": 1.0962,
+      "step": 4614
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2280717869194476e-05,
+      "loss": 1.053,
+      "step": 4615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227768414370317e-05,
+      "loss": 1.1711,
+      "step": 4616
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2274650197099154e-05,
+      "loss": 1.0981,
+      "step": 4617
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2271616029676958e-05,
+      "loss": 1.1599,
+      "step": 4618
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2268581641731134e-05,
+      "loss": 1.0076,
+      "step": 4619
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.226554703355625e-05,
+      "loss": 1.1301,
+      "step": 4620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2262512205446904e-05,
+      "loss": 1.031,
+      "step": 4621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2259477157697706e-05,
+      "loss": 1.1929,
+      "step": 4622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2256441890603294e-05,
+      "loss": 1.1357,
+      "step": 4623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2253406404458326e-05,
+      "loss": 1.1726,
+      "step": 4624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225037069955748e-05,
+      "loss": 0.9951,
+      "step": 4625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2247334776195457e-05,
+      "loss": 1.1067,
+      "step": 4626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2244298634666973e-05,
+      "loss": 1.0826,
+      "step": 4627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2241262275266777e-05,
+      "loss": 1.0096,
+      "step": 4628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2238225698289626e-05,
+      "loss": 1.1218,
+      "step": 4629
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2235188904030309e-05,
+      "loss": 1.1128,
+      "step": 4630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223215189278363e-05,
+      "loss": 1.0867,
+      "step": 4631
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2229114664844415e-05,
+      "loss": 1.0613,
+      "step": 4632
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2226077220507511e-05,
+      "loss": 1.0316,
+      "step": 4633
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2223039560067788e-05,
+      "loss": 1.0876,
+      "step": 4634
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2220001683820135e-05,
+      "loss": 1.2302,
+      "step": 4635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2216963592059466e-05,
+      "loss": 1.177,
+      "step": 4636
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2213925285080706e-05,
+      "loss": 1.0708,
+      "step": 4637
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2210886763178808e-05,
+      "loss": 1.1301,
+      "step": 4638
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2207848026648752e-05,
+      "loss": 1.098,
+      "step": 4639
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2204809075785527e-05,
+      "loss": 1.1562,
+      "step": 4640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2201769910884147e-05,
+      "loss": 1.1543,
+      "step": 4641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219873053223965e-05,
+      "loss": 1.0884,
+      "step": 4642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2195690940147093e-05,
+      "loss": 1.2302,
+      "step": 4643
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2192651134901549e-05,
+      "loss": 1.0928,
+      "step": 4644
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2189611116798121e-05,
+      "loss": 1.0422,
+      "step": 4645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2186570886131923e-05,
+      "loss": 1.04,
+      "step": 4646
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2183530443198097e-05,
+      "loss": 1.1421,
+      "step": 4647
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2180489788291803e-05,
+      "loss": 1.2737,
+      "step": 4648
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217744892170822e-05,
+      "loss": 1.0894,
+      "step": 4649
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217440784374255e-05,
+      "loss": 1.0167,
+      "step": 4650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2171366554690009e-05,
+      "loss": 1.1239,
+      "step": 4651
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2168325054845844e-05,
+      "loss": 1.0959,
+      "step": 4652
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2165283344505319e-05,
+      "loss": 1.1024,
+      "step": 4653
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2162241423963715e-05,
+      "loss": 1.0647,
+      "step": 4654
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2159199293516331e-05,
+      "loss": 0.998,
+      "step": 4655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2156156953458494e-05,
+      "loss": 1.0908,
+      "step": 4656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2153114404085549e-05,
+      "loss": 1.1838,
+      "step": 4657
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215007164569286e-05,
+      "loss": 1.1072,
+      "step": 4658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147028678575808e-05,
+      "loss": 1.1243,
+      "step": 4659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2143985503029801e-05,
+      "loss": 1.135,
+      "step": 4660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2140942119350263e-05,
+      "loss": 1.0322,
+      "step": 4661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2137898527832639e-05,
+      "loss": 1.1641,
+      "step": 4662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2134854728772395e-05,
+      "loss": 1.1514,
+      "step": 4663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2131810722465016e-05,
+      "loss": 1.1758,
+      "step": 4664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212876650920601e-05,
+      "loss": 0.9761,
+      "step": 4665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2125722089290895e-05,
+      "loss": 1.2058,
+      "step": 4666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2122677463015226e-05,
+      "loss": 1.1848,
+      "step": 4667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2119632630674565e-05,
+      "loss": 1.2036,
+      "step": 4668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21165875925645e-05,
+      "loss": 1.1809,
+      "step": 4669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2113542348980629e-05,
+      "loss": 1.0317,
+      "step": 4670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2110496900218588e-05,
+      "loss": 1.0071,
+      "step": 4671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2107451246574018e-05,
+      "loss": 1.2051,
+      "step": 4672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2104405388342584e-05,
+      "loss": 1.1458,
+      "step": 4673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2101359325819968e-05,
+      "loss": 1.228,
+      "step": 4674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2098313059301882e-05,
+      "loss": 1.1123,
+      "step": 4675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2095266589084052e-05,
+      "loss": 1.0527,
+      "step": 4676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2092219915462216e-05,
+      "loss": 0.9679,
+      "step": 4677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208917303873214e-05,
+      "loss": 1.1033,
+      "step": 4678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2086125959189613e-05,
+      "loss": 1.0934,
+      "step": 4679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2083078677130435e-05,
+      "loss": 1.1335,
+      "step": 4680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208003119285043e-05,
+      "loss": 1.0815,
+      "step": 4681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207698350664544e-05,
+      "loss": 1.1013,
+      "step": 4682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2073935618811335e-05,
+      "loss": 0.989,
+      "step": 4683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2070887529643989e-05,
+      "loss": 1.0767,
+      "step": 4684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067839239439306e-05,
+      "loss": 1.1307,
+      "step": 4685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2064790748493208e-05,
+      "loss": 1.0879,
+      "step": 4686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206174205710164e-05,
+      "loss": 1.1218,
+      "step": 4687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2058693165560558e-05,
+      "loss": 1.0879,
+      "step": 4688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2055644074165941e-05,
+      "loss": 1.0232,
+      "step": 4689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205259478321379e-05,
+      "loss": 1.0999,
+      "step": 4690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2049545293000127e-05,
+      "loss": 1.1426,
+      "step": 4691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2046495603820984e-05,
+      "loss": 1.1216,
+      "step": 4692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2043445715972424e-05,
+      "loss": 1.1292,
+      "step": 4693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2040395629750518e-05,
+      "loss": 1.0793,
+      "step": 4694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2037345345451368e-05,
+      "loss": 1.0615,
+      "step": 4695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2034294863371082e-05,
+      "loss": 1.0098,
+      "step": 4696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2031244183805797e-05,
+      "loss": 1.0291,
+      "step": 4697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2028193307051671e-05,
+      "loss": 0.9525,
+      "step": 4698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2025142233404875e-05,
+      "loss": 1.0732,
+      "step": 4699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2022090963161595e-05,
+      "loss": 1.145,
+      "step": 4700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2019039496618044e-05,
+      "loss": 1.0415,
+      "step": 4701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2015987834070457e-05,
+      "loss": 1.016,
+      "step": 4702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2012935975815076e-05,
+      "loss": 1.1074,
+      "step": 4703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2009883922148173e-05,
+      "loss": 1.074,
+      "step": 4704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2006831673366034e-05,
+      "loss": 1.1182,
+      "step": 4705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2003779229764967e-05,
+      "loss": 1.0581,
+      "step": 4706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2000726591641293e-05,
+      "loss": 1.0569,
+      "step": 4707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1997673759291357e-05,
+      "loss": 1.0815,
+      "step": 4708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994620733011521e-05,
+      "loss": 1.1941,
+      "step": 4709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1991567513098168e-05,
+      "loss": 1.1565,
+      "step": 4710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1988514099847697e-05,
+      "loss": 1.0276,
+      "step": 4711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1985460493556525e-05,
+      "loss": 1.1338,
+      "step": 4712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1982406694521091e-05,
+      "loss": 1.0159,
+      "step": 4713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1979352703037854e-05,
+      "loss": 1.1643,
+      "step": 4714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1976298519403284e-05,
+      "loss": 1.0094,
+      "step": 4715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1973244143913879e-05,
+      "loss": 1.0571,
+      "step": 4716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1970189576866148e-05,
+      "loss": 1.001,
+      "step": 4717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1967134818556625e-05,
+      "loss": 1.0793,
+      "step": 4718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1964079869281855e-05,
+      "loss": 1.1167,
+      "step": 4719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196102472933841e-05,
+      "loss": 1.0897,
+      "step": 4720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1957969399022874e-05,
+      "loss": 1.0686,
+      "step": 4721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1954913878631855e-05,
+      "loss": 1.0632,
+      "step": 4722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1951858168461972e-05,
+      "loss": 1.0101,
+      "step": 4723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1948802268809869e-05,
+      "loss": 1.1187,
+      "step": 4724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1945746179972208e-05,
+      "loss": 1.0696,
+      "step": 4725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1942689902245665e-05,
+      "loss": 1.0898,
+      "step": 4726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.193963343592694e-05,
+      "loss": 1.1184,
+      "step": 4727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1936576781312743e-05,
+      "loss": 1.1064,
+      "step": 4728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1933519938699814e-05,
+      "loss": 1.1604,
+      "step": 4729
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1930462908384899e-05,
+      "loss": 1.1833,
+      "step": 4730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1927405690664769e-05,
+      "loss": 1.1323,
+      "step": 4731
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1924348285836216e-05,
+      "loss": 1.1616,
+      "step": 4732
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1921290694196042e-05,
+      "loss": 1.0764,
+      "step": 4733
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1918232916041075e-05,
+      "loss": 1.1782,
+      "step": 4734
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1915174951668153e-05,
+      "loss": 1.0215,
+      "step": 4735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191211680137414e-05,
+      "loss": 1.0526,
+      "step": 4736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1909058465455917e-05,
+      "loss": 1.001,
+      "step": 4737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1905999944210376e-05,
+      "loss": 1.1743,
+      "step": 4738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1902941237934433e-05,
+      "loss": 1.2117,
+      "step": 4739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1899882346925022e-05,
+      "loss": 1.012,
+      "step": 4740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1896823271479092e-05,
+      "loss": 1.0063,
+      "step": 4741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1893764011893616e-05,
+      "loss": 1.1271,
+      "step": 4742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1890704568465572e-05,
+      "loss": 1.0082,
+      "step": 4743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188764494149197e-05,
+      "loss": 1.0382,
+      "step": 4744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1884585131269831e-05,
+      "loss": 0.9137,
+      "step": 4745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1881525138096198e-05,
+      "loss": 1.0325,
+      "step": 4746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1878464962268122e-05,
+      "loss": 1.1682,
+      "step": 4747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187540460408268e-05,
+      "loss": 0.9888,
+      "step": 4748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1872344063836972e-05,
+      "loss": 1.0938,
+      "step": 4749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1869283341828101e-05,
+      "loss": 0.9861,
+      "step": 4750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1866222438353196e-05,
+      "loss": 1.1853,
+      "step": 4751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1863161353709407e-05,
+      "loss": 1.1233,
+      "step": 4752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1860100088193897e-05,
+      "loss": 1.0791,
+      "step": 4753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1857038642103845e-05,
+      "loss": 1.1384,
+      "step": 4754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1853977015736445e-05,
+      "loss": 1.0458,
+      "step": 4755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1850915209388923e-05,
+      "loss": 1.0416,
+      "step": 4756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1847853223358507e-05,
+      "loss": 1.0156,
+      "step": 4757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1844791057942448e-05,
+      "loss": 1.0995,
+      "step": 4758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1841728713438015e-05,
+      "loss": 1.1326,
+      "step": 4759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838666190142496e-05,
+      "loss": 1.1641,
+      "step": 4760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1835603488353194e-05,
+      "loss": 1.0457,
+      "step": 4761
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1832540608367421e-05,
+      "loss": 1.0583,
+      "step": 4762
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182947755048253e-05,
+      "loss": 1.137,
+      "step": 4763
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1826414314995867e-05,
+      "loss": 1.0742,
+      "step": 4764
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1823350902204802e-05,
+      "loss": 1.0437,
+      "step": 4765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1820287312406726e-05,
+      "loss": 1.0513,
+      "step": 4766
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1817223545899052e-05,
+      "loss": 1.1067,
+      "step": 4767
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18141596029792e-05,
+      "loss": 1.1211,
+      "step": 4768
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181109548394461e-05,
+      "loss": 1.0596,
+      "step": 4769
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1808031189092737e-05,
+      "loss": 1.1416,
+      "step": 4770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1804966718721062e-05,
+      "loss": 1.0459,
+      "step": 4771
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1801902073127078e-05,
+      "loss": 1.1472,
+      "step": 4772
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179883725260829e-05,
+      "loss": 1.1401,
+      "step": 4773
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1795772257462224e-05,
+      "loss": 1.188,
+      "step": 4774
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1792707087986426e-05,
+      "loss": 0.9443,
+      "step": 4775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1789641744478455e-05,
+      "loss": 1.1703,
+      "step": 4776
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1786576227235888e-05,
+      "loss": 1.093,
+      "step": 4777
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783510536556314e-05,
+      "loss": 1.1189,
+      "step": 4778
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1780444672737357e-05,
+      "loss": 1.1167,
+      "step": 4779
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1777378636076634e-05,
+      "loss": 1.1177,
+      "step": 4780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1774312426871788e-05,
+      "loss": 1.131,
+      "step": 4781
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177124604542048e-05,
+      "loss": 1.1345,
+      "step": 4782
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17681794920204e-05,
+      "loss": 1.2324,
+      "step": 4783
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176511276696923e-05,
+      "loss": 1.1328,
+      "step": 4784
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1762045870564683e-05,
+      "loss": 1.0046,
+      "step": 4785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1758978803104489e-05,
+      "loss": 1.1633,
+      "step": 4786
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1755911564886392e-05,
+      "loss": 1.0552,
+      "step": 4787
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752844156208152e-05,
+      "loss": 1.2075,
+      "step": 4788
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1749776577367547e-05,
+      "loss": 1.0426,
+      "step": 4789
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1746708828662372e-05,
+      "loss": 1.0488,
+      "step": 4790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1743640910390437e-05,
+      "loss": 0.9946,
+      "step": 4791
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1740572822849568e-05,
+      "loss": 1.1902,
+      "step": 4792
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1737504566337608e-05,
+      "loss": 1.1661,
+      "step": 4793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1734436141152419e-05,
+      "loss": 1.0647,
+      "step": 4794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1731367547591876e-05,
+      "loss": 0.97,
+      "step": 4795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1728298785953871e-05,
+      "loss": 1.0269,
+      "step": 4796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1725229856536313e-05,
+      "loss": 1.0466,
+      "step": 4797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1722160759637126e-05,
+      "loss": 1.0349,
+      "step": 4798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1719091495554255e-05,
+      "loss": 0.9769,
+      "step": 4799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1716022064585655e-05,
+      "loss": 0.9963,
+      "step": 4800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1712952467029301e-05,
+      "loss": 1.0968,
+      "step": 4801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.170988270318318e-05,
+      "loss": 1.0918,
+      "step": 4802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706812773345301e-05,
+      "loss": 0.9534,
+      "step": 4803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1703742677813685e-05,
+      "loss": 1.0297,
+      "step": 4804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1700672416886372e-05,
+      "loss": 0.9342,
+      "step": 4805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1697601990861416e-05,
+      "loss": 1.1179,
+      "step": 4806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1694531400036884e-05,
+      "loss": 1.168,
+      "step": 4807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1691460644710866e-05,
+      "loss": 1.0657,
+      "step": 4808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1688389725181464e-05,
+      "loss": 1.0161,
+      "step": 4809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685318641746795e-05,
+      "loss": 1.1228,
+      "step": 4810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1682247394704993e-05,
+      "loss": 1.0813,
+      "step": 4811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167917598435421e-05,
+      "loss": 1.2942,
+      "step": 4812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167610441099261e-05,
+      "loss": 1.231,
+      "step": 4813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1673032674918376e-05,
+      "loss": 1.1104,
+      "step": 4814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1669960776429706e-05,
+      "loss": 1.0461,
+      "step": 4815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1666888715824812e-05,
+      "loss": 1.0049,
+      "step": 4816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663816493401924e-05,
+      "loss": 1.0486,
+      "step": 4817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1660744109459283e-05,
+      "loss": 1.1548,
+      "step": 4818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165767156429516e-05,
+      "loss": 1.1079,
+      "step": 4819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1654598858207815e-05,
+      "loss": 1.0696,
+      "step": 4820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1651525991495553e-05,
+      "loss": 1.062,
+      "step": 4821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1648452964456678e-05,
+      "loss": 1.2166,
+      "step": 4822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1645379777389513e-05,
+      "loss": 0.9778,
+      "step": 4823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1642306430592392e-05,
+      "loss": 1.0752,
+      "step": 4824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1639232924363675e-05,
+      "loss": 0.9645,
+      "step": 4825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1636159259001726e-05,
+      "loss": 1.1555,
+      "step": 4826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1633085434804939e-05,
+      "loss": 1.1833,
+      "step": 4827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1630011452071701e-05,
+      "loss": 1.0645,
+      "step": 4828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1626937311100438e-05,
+      "loss": 1.0461,
+      "step": 4829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1623863012189576e-05,
+      "loss": 1.0898,
+      "step": 4830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1620788555637563e-05,
+      "loss": 1.1155,
+      "step": 4831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161771394174286e-05,
+      "loss": 1.0276,
+      "step": 4832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1614639170803947e-05,
+      "loss": 1.0352,
+      "step": 4833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1611564243119314e-05,
+      "loss": 0.974,
+      "step": 4834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1608489158987463e-05,
+      "loss": 1.1177,
+      "step": 4835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1605413918706927e-05,
+      "loss": 1.1243,
+      "step": 4836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1602338522576238e-05,
+      "loss": 1.1375,
+      "step": 4837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159926297089395e-05,
+      "loss": 1.1106,
+      "step": 4838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1596187263958626e-05,
+      "loss": 1.0071,
+      "step": 4839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1593111402068857e-05,
+      "loss": 1.0505,
+      "step": 4840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159003538552324e-05,
+      "loss": 1.0315,
+      "step": 4841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586959214620384e-05,
+      "loss": 0.9435,
+      "step": 4842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1583882889658913e-05,
+      "loss": 1.0027,
+      "step": 4843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1580806410937481e-05,
+      "loss": 0.9972,
+      "step": 4844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577729778754743e-05,
+      "loss": 1.0864,
+      "step": 4845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1574652993409368e-05,
+      "loss": 1.1494,
+      "step": 4846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157157605520004e-05,
+      "loss": 1.1698,
+      "step": 4847
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1568498964425472e-05,
+      "loss": 1.0168,
+      "step": 4848
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1565421721384378e-05,
+      "loss": 1.1604,
+      "step": 4849
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1562344326375487e-05,
+      "loss": 1.0608,
+      "step": 4850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1559266779697545e-05,
+      "loss": 1.1376,
+      "step": 4851
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155618908164932e-05,
+      "loss": 1.1863,
+      "step": 4852
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1553111232529583e-05,
+      "loss": 1.1428,
+      "step": 4853
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1550033232637126e-05,
+      "loss": 1.0458,
+      "step": 4854
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1546955082270752e-05,
+      "loss": 1.0764,
+      "step": 4855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1543876781729291e-05,
+      "loss": 1.1646,
+      "step": 4856
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154079833131157e-05,
+      "loss": 1.1697,
+      "step": 4857
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1537719731316437e-05,
+      "loss": 1.1411,
+      "step": 4858
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1534640982042758e-05,
+      "loss": 1.0253,
+      "step": 4859
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1531562083789413e-05,
+      "loss": 1.0127,
+      "step": 4860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1528483036855296e-05,
+      "loss": 0.9624,
+      "step": 4861
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1525403841539311e-05,
+      "loss": 1.1843,
+      "step": 4862
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152232449814038e-05,
+      "loss": 1.1257,
+      "step": 4863
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151924500695744e-05,
+      "loss": 1.0302,
+      "step": 4864
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1516165368289442e-05,
+      "loss": 1.0781,
+      "step": 4865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1513085582435349e-05,
+      "loss": 1.0984,
+      "step": 4866
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510005649694142e-05,
+      "loss": 1.0472,
+      "step": 4867
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1506925570364814e-05,
+      "loss": 1.064,
+      "step": 4868
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1503845344746372e-05,
+      "loss": 1.1484,
+      "step": 4869
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1500764973137835e-05,
+      "loss": 1.156,
+      "step": 4870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497684455838244e-05,
+      "loss": 1.2231,
+      "step": 4871
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1494603793146646e-05,
+      "loss": 1.0529,
+      "step": 4872
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1491522985362106e-05,
+      "loss": 1.1296,
+      "step": 4873
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1488442032783702e-05,
+      "loss": 1.0409,
+      "step": 4874
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1485360935710528e-05,
+      "loss": 1.0083,
+      "step": 4875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1482279694441684e-05,
+      "loss": 1.0024,
+      "step": 4876
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1479198309276298e-05,
+      "loss": 1.1968,
+      "step": 4877
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1476116780513502e-05,
+      "loss": 1.026,
+      "step": 4878
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1473035108452441e-05,
+      "loss": 1.0337,
+      "step": 4879
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1469953293392282e-05,
+      "loss": 1.0933,
+      "step": 4880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1466871335632197e-05,
+      "loss": 1.1133,
+      "step": 4881
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1463789235471378e-05,
+      "loss": 1.0593,
+      "step": 4882
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1460706993209028e-05,
+      "loss": 1.0869,
+      "step": 4883
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1457624609144365e-05,
+      "loss": 1.1733,
+      "step": 4884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1454542083576619e-05,
+      "loss": 1.0834,
+      "step": 4885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1451459416805035e-05,
+      "loss": 1.1418,
+      "step": 4886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1448376609128875e-05,
+      "loss": 1.0996,
+      "step": 4887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1445293660847407e-05,
+      "loss": 1.0908,
+      "step": 4888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1442210572259922e-05,
+      "loss": 1.1201,
+      "step": 4889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439127343665715e-05,
+      "loss": 1.0527,
+      "step": 4890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14360439753641e-05,
+      "loss": 1.144,
+      "step": 4891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1432960467654404e-05,
+      "loss": 1.0826,
+      "step": 4892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142987682083597e-05,
+      "loss": 1.1206,
+      "step": 4893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1426793035208147e-05,
+      "loss": 1.0405,
+      "step": 4894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423709111070309e-05,
+      "loss": 1.0962,
+      "step": 4895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1420625048721832e-05,
+      "loss": 1.1882,
+      "step": 4896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141754084846211e-05,
+      "loss": 1.125,
+      "step": 4897
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1414456510590553e-05,
+      "loss": 1.1816,
+      "step": 4898
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1411372035406578e-05,
+      "loss": 0.9919,
+      "step": 4899
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1408287423209626e-05,
+      "loss": 1.0115,
+      "step": 4900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1405202674299139e-05,
+      "loss": 1.137,
+      "step": 4901
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402117788974575e-05,
+      "loss": 1.1317,
+      "step": 4902
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399032767535418e-05,
+      "loss": 1.1124,
+      "step": 4903
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1395947610281146e-05,
+      "loss": 1.0311,
+      "step": 4904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1392862317511266e-05,
+      "loss": 1.0364,
+      "step": 4905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1389776889525288e-05,
+      "loss": 1.0812,
+      "step": 4906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.138669132662274e-05,
+      "loss": 1.0721,
+      "step": 4907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1383605629103158e-05,
+      "loss": 1.0977,
+      "step": 4908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1380519797266102e-05,
+      "loss": 1.1028,
+      "step": 4909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1377433831411133e-05,
+      "loss": 1.05,
+      "step": 4910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1374347731837834e-05,
+      "loss": 1.1025,
+      "step": 4911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137126149884579e-05,
+      "loss": 1.1995,
+      "step": 4912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1368175132734609e-05,
+      "loss": 1.1366,
+      "step": 4913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1365088633803913e-05,
+      "loss": 0.9263,
+      "step": 4914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1362002002353331e-05,
+      "loss": 0.9991,
+      "step": 4915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1358915238682502e-05,
+      "loss": 1.1274,
+      "step": 4916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1355828343091087e-05,
+      "loss": 1.0962,
+      "step": 4917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1352741315878755e-05,
+      "loss": 1.1062,
+      "step": 4918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1349654157345188e-05,
+      "loss": 0.9893,
+      "step": 4919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1346566867790073e-05,
+      "loss": 0.9916,
+      "step": 4920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1343479447513129e-05,
+      "loss": 1.0623,
+      "step": 4921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1340391896814074e-05,
+      "loss": 1.1956,
+      "step": 4922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1337304215992636e-05,
+      "loss": 1.1099,
+      "step": 4923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1334216405348557e-05,
+      "loss": 0.9899,
+      "step": 4924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1331128465181605e-05,
+      "loss": 1.0681,
+      "step": 4925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1328040395791549e-05,
+      "loss": 1.1504,
+      "step": 4926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1324952197478165e-05,
+      "loss": 1.0708,
+      "step": 4927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1321863870541252e-05,
+      "loss": 1.0065,
+      "step": 4928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1318775415280622e-05,
+      "loss": 1.0,
+      "step": 4929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1315686831996096e-05,
+      "loss": 0.9871,
+      "step": 4930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13125981209875e-05,
+      "loss": 1.2129,
+      "step": 4931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1309509282554682e-05,
+      "loss": 1.0745,
+      "step": 4932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1306420316997507e-05,
+      "loss": 1.0793,
+      "step": 4933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1303331224615835e-05,
+      "loss": 1.0618,
+      "step": 4934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1300242005709551e-05,
+      "loss": 1.0154,
+      "step": 4935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1297152660578557e-05,
+      "loss": 1.0891,
+      "step": 4936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1294063189522753e-05,
+      "loss": 1.0001,
+      "step": 4937
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.129097359284206e-05,
+      "loss": 1.0181,
+      "step": 4938
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1287883870836407e-05,
+      "loss": 1.2551,
+      "step": 4939
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128479402380574e-05,
+      "loss": 1.1848,
+      "step": 4940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128170405205002e-05,
+      "loss": 1.1013,
+      "step": 4941
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1278613955869206e-05,
+      "loss": 1.1758,
+      "step": 4942
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1275523735563281e-05,
+      "loss": 1.1182,
+      "step": 4943
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127243339143224e-05,
+      "loss": 1.1003,
+      "step": 4944
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269342923776083e-05,
+      "loss": 1.0881,
+      "step": 4945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1266252332894827e-05,
+      "loss": 1.0823,
+      "step": 4946
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1263161619088498e-05,
+      "loss": 1.0403,
+      "step": 4947
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1260070782657144e-05,
+      "loss": 1.0276,
+      "step": 4948
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125697982390081e-05,
+      "loss": 0.9521,
+      "step": 4949
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1253888743119558e-05,
+      "loss": 1.0684,
+      "step": 4950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1250797540613465e-05,
+      "loss": 1.1206,
+      "step": 4951
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1247706216682626e-05,
+      "loss": 1.0454,
+      "step": 4952
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1244614771627129e-05,
+      "loss": 1.1387,
+      "step": 4953
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1241523205747092e-05,
+      "loss": 1.0338,
+      "step": 4954
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1238431519342634e-05,
+      "loss": 1.048,
+      "step": 4955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1235339712713894e-05,
+      "loss": 1.1179,
+      "step": 4956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123224778616101e-05,
+      "loss": 1.106,
+      "step": 4957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1229155739984145e-05,
+      "loss": 1.0801,
+      "step": 4958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.122606357448347e-05,
+      "loss": 1.1138,
+      "step": 4959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1222971289959161e-05,
+      "loss": 1.007,
+      "step": 4960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1219878886711413e-05,
+      "loss": 1.0703,
+      "step": 4961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121678636504043e-05,
+      "loss": 1.0101,
+      "step": 4962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1213693725246426e-05,
+      "loss": 1.0542,
+      "step": 4963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121060096762963e-05,
+      "loss": 1.0828,
+      "step": 4964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120750809249028e-05,
+      "loss": 1.0078,
+      "step": 4965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1204415100128624e-05,
+      "loss": 1.0479,
+      "step": 4966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1201321990844921e-05,
+      "loss": 0.9503,
+      "step": 4967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1198228764939451e-05,
+      "loss": 1.1743,
+      "step": 4968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1195135422712492e-05,
+      "loss": 0.975,
+      "step": 4969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1192041964464339e-05,
+      "loss": 1.1292,
+      "step": 4970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11889483904953e-05,
+      "loss": 1.0828,
+      "step": 4971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1185854701105691e-05,
+      "loss": 1.1931,
+      "step": 4972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1182760896595843e-05,
+      "loss": 1.0164,
+      "step": 4973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1179666977266095e-05,
+      "loss": 1.0502,
+      "step": 4974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1176572943416797e-05,
+      "loss": 1.059,
+      "step": 4975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1173478795348312e-05,
+      "loss": 1.0251,
+      "step": 4976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1170384533361014e-05,
+      "loss": 1.0935,
+      "step": 4977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1167290157755287e-05,
+      "loss": 1.0966,
+      "step": 4978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1164195668831523e-05,
+      "loss": 0.9357,
+      "step": 4979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1161101066890134e-05,
+      "loss": 1.1929,
+      "step": 4980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1158006352231535e-05,
+      "loss": 1.0923,
+      "step": 4981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1154911525156152e-05,
+      "loss": 0.9121,
+      "step": 4982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1151816585964428e-05,
+      "loss": 1.0773,
+      "step": 4983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114872153495681e-05,
+      "loss": 1.0591,
+      "step": 4984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114562637243376e-05,
+      "loss": 1.1279,
+      "step": 4985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1142531098695752e-05,
+      "loss": 1.0774,
+      "step": 4986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1139435714043265e-05,
+      "loss": 1.1965,
+      "step": 4987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1136340218776796e-05,
+      "loss": 0.9939,
+      "step": 4988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133244613196843e-05,
+      "loss": 1.1372,
+      "step": 4989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1130148897603928e-05,
+      "loss": 0.9846,
+      "step": 4990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.112705307229857e-05,
+      "loss": 0.9062,
+      "step": 4991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1123957137581313e-05,
+      "loss": 1.1223,
+      "step": 4992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1120861093752695e-05,
+      "loss": 1.061,
+      "step": 4993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1117764941113278e-05,
+      "loss": 1.0696,
+      "step": 4994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111466867996363e-05,
+      "loss": 0.9655,
+      "step": 4995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1111572310604331e-05,
+      "loss": 1.2224,
+      "step": 4996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.110847583333596e-05,
+      "loss": 1.0847,
+      "step": 4997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1105379248459132e-05,
+      "loss": 1.053,
+      "step": 4998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102282556274446e-05,
+      "loss": 1.1062,
+      "step": 4999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1099185757082522e-05,
+      "loss": 1.0297,
+      "step": 5000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1096088851183997e-05,
+      "loss": 1.1392,
+      "step": 5001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1092991838879507e-05,
+      "loss": 0.9287,
+      "step": 5002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108989472046971e-05,
+      "loss": 1.0913,
+      "step": 5003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086797496255256e-05,
+      "loss": 1.0889,
+      "step": 5004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1083700166536825e-05,
+      "loss": 1.0129,
+      "step": 5005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10806027316151e-05,
+      "loss": 1.1365,
+      "step": 5006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1077505191790772e-05,
+      "loss": 0.9408,
+      "step": 5007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1074407547364538e-05,
+      "loss": 1.0732,
+      "step": 5008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1071309798637118e-05,
+      "loss": 1.0339,
+      "step": 5009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1068211945909233e-05,
+      "loss": 1.116,
+      "step": 5010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1065113989481614e-05,
+      "loss": 1.095,
+      "step": 5011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1062015929655004e-05,
+      "loss": 1.2131,
+      "step": 5012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1058917766730158e-05,
+      "loss": 1.092,
+      "step": 5013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105581950100784e-05,
+      "loss": 1.1292,
+      "step": 5014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1052721132788819e-05,
+      "loss": 1.04,
+      "step": 5015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.104962266237388e-05,
+      "loss": 1.0785,
+      "step": 5016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1046524090063815e-05,
+      "loss": 1.0764,
+      "step": 5017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1043425416159432e-05,
+      "loss": 1.0972,
+      "step": 5018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1040326640961537e-05,
+      "loss": 1.0056,
+      "step": 5019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1037227764770953e-05,
+      "loss": 0.9651,
+      "step": 5020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.103412878788852e-05,
+      "loss": 0.9468,
+      "step": 5021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1031029710615067e-05,
+      "loss": 1.0857,
+      "step": 5022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1027930533251456e-05,
+      "loss": 1.0327,
+      "step": 5023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1024831256098544e-05,
+      "loss": 1.032,
+      "step": 5024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1021731879457207e-05,
+      "loss": 1.147,
+      "step": 5025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1018632403628322e-05,
+      "loss": 1.1091,
+      "step": 5026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1015532828912779e-05,
+      "loss": 0.9604,
+      "step": 5027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1012433155611476e-05,
+      "loss": 1.0393,
+      "step": 5028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1009333384025331e-05,
+      "loss": 1.0165,
+      "step": 5029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1006233514455256e-05,
+      "loss": 1.0173,
+      "step": 5030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1003133547202181e-05,
+      "loss": 1.0914,
+      "step": 5031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1000033482567045e-05,
+      "loss": 1.0049,
+      "step": 5032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0996933320850798e-05,
+      "loss": 0.9751,
+      "step": 5033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0993833062354393e-05,
+      "loss": 0.957,
+      "step": 5034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0990732707378795e-05,
+      "loss": 1.0708,
+      "step": 5035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0987632256224989e-05,
+      "loss": 1.0315,
+      "step": 5036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.098453170919395e-05,
+      "loss": 1.1372,
+      "step": 5037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0981431066586679e-05,
+      "loss": 1.0615,
+      "step": 5038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0978330328704173e-05,
+      "loss": 1.1455,
+      "step": 5039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0975229495847455e-05,
+      "loss": 1.126,
+      "step": 5040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.097212856831754e-05,
+      "loss": 1.0815,
+      "step": 5041
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0969027546415459e-05,
+      "loss": 1.0989,
+      "step": 5042
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0965926430442256e-05,
+      "loss": 1.0129,
+      "step": 5043
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0962825220698979e-05,
+      "loss": 1.1265,
+      "step": 5044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0959723917486687e-05,
+      "loss": 0.9847,
+      "step": 5045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0956622521106448e-05,
+      "loss": 1.0813,
+      "step": 5046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.095352103185934e-05,
+      "loss": 1.0989,
+      "step": 5047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950419450046448e-05,
+      "loss": 1.1533,
+      "step": 5048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0947317775968864e-05,
+      "loss": 1.1589,
+      "step": 5049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0944216009927697e-05,
+      "loss": 0.9646,
+      "step": 5050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0941114152224058e-05,
+      "loss": 1.0562,
+      "step": 5051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0938012203159067e-05,
+      "loss": 1.126,
+      "step": 5052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934910163033856e-05,
+      "loss": 1.231,
+      "step": 5053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0931808032149565e-05,
+      "loss": 1.0605,
+      "step": 5054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092870581080734e-05,
+      "loss": 1.0876,
+      "step": 5055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925603499308342e-05,
+      "loss": 1.1508,
+      "step": 5056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0922501097953733e-05,
+      "loss": 1.0646,
+      "step": 5057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0919398607044688e-05,
+      "loss": 1.0557,
+      "step": 5058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0916296026882393e-05,
+      "loss": 1.1609,
+      "step": 5059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0913193357768036e-05,
+      "loss": 1.0476,
+      "step": 5060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091009060000282e-05,
+      "loss": 1.0194,
+      "step": 5061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0906987753887955e-05,
+      "loss": 1.0732,
+      "step": 5062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0903884819724654e-05,
+      "loss": 0.9648,
+      "step": 5063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0900781797814148e-05,
+      "loss": 1.1436,
+      "step": 5064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0897678688457672e-05,
+      "loss": 1.156,
+      "step": 5065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0894575491956462e-05,
+      "loss": 1.1423,
+      "step": 5066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089147220861178e-05,
+      "loss": 1.1611,
+      "step": 5067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0888368838724879e-05,
+      "loss": 1.0837,
+      "step": 5068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088526538259703e-05,
+      "loss": 1.1431,
+      "step": 5069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0882161840529508e-05,
+      "loss": 1.0498,
+      "step": 5070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0879058212823602e-05,
+      "loss": 1.1306,
+      "step": 5071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0875954499780601e-05,
+      "loss": 1.0664,
+      "step": 5072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0872850701701811e-05,
+      "loss": 1.0463,
+      "step": 5073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086974681888854e-05,
+      "loss": 0.9388,
+      "step": 5074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0866642851642106e-05,
+      "loss": 1.187,
+      "step": 5075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0863538800263836e-05,
+      "loss": 1.1724,
+      "step": 5076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0860434665055064e-05,
+      "loss": 1.0801,
+      "step": 5077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0857330446317134e-05,
+      "loss": 1.2004,
+      "step": 5078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0854226144351397e-05,
+      "loss": 1.2153,
+      "step": 5079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0851121759459213e-05,
+      "loss": 1.0879,
+      "step": 5080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0848017291941943e-05,
+      "loss": 1.1211,
+      "step": 5081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0844912742100969e-05,
+      "loss": 1.0967,
+      "step": 5082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084180811023767e-05,
+      "loss": 1.0836,
+      "step": 5083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0838703396653442e-05,
+      "loss": 0.9924,
+      "step": 5084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0835598601649677e-05,
+      "loss": 1.0865,
+      "step": 5085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0832493725527788e-05,
+      "loss": 1.0439,
+      "step": 5086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0829388768589186e-05,
+      "loss": 0.9695,
+      "step": 5087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0826283731135296e-05,
+      "loss": 1.0574,
+      "step": 5088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0823178613467544e-05,
+      "loss": 1.0562,
+      "step": 5089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0820073415887373e-05,
+      "loss": 1.042,
+      "step": 5090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0816968138696232e-05,
+      "loss": 1.1029,
+      "step": 5091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0813862782195566e-05,
+      "loss": 1.0898,
+      "step": 5092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0810757346686835e-05,
+      "loss": 1.1062,
+      "step": 5093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080765183247152e-05,
+      "loss": 1.1624,
+      "step": 5094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080454623985109e-05,
+      "loss": 1.1086,
+      "step": 5095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080144056912703e-05,
+      "loss": 1.1018,
+      "step": 5096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0798334820600831e-05,
+      "loss": 1.1201,
+      "step": 5097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0795228994573994e-05,
+      "loss": 0.9803,
+      "step": 5098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0792123091348028e-05,
+      "loss": 1.1025,
+      "step": 5099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0789017111224443e-05,
+      "loss": 1.1169,
+      "step": 5100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0785911054504762e-05,
+      "loss": 1.1116,
+      "step": 5101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078280492149052e-05,
+      "loss": 1.0422,
+      "step": 5102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0779698712483245e-05,
+      "loss": 1.0269,
+      "step": 5103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0776592427784484e-05,
+      "loss": 1.0608,
+      "step": 5104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.077348606769579e-05,
+      "loss": 1.0901,
+      "step": 5105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0770379632518726e-05,
+      "loss": 1.1194,
+      "step": 5106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0767273122554851e-05,
+      "loss": 1.0674,
+      "step": 5107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0764166538105738e-05,
+      "loss": 1.1292,
+      "step": 5108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0761059879472975e-05,
+      "loss": 1.1421,
+      "step": 5109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0757953146958144e-05,
+      "loss": 1.1298,
+      "step": 5110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0754846340862841e-05,
+      "loss": 1.0771,
+      "step": 5111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751739461488667e-05,
+      "loss": 1.1685,
+      "step": 5112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0748632509137235e-05,
+      "loss": 1.1338,
+      "step": 5113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0745525484110157e-05,
+      "loss": 1.1917,
+      "step": 5114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074241838670906e-05,
+      "loss": 1.1813,
+      "step": 5115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0739311217235568e-05,
+      "loss": 1.1025,
+      "step": 5116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0736203975991329e-05,
+      "loss": 1.036,
+      "step": 5117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0733096663277977e-05,
+      "loss": 1.1708,
+      "step": 5118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.072998927939717e-05,
+      "loss": 0.9629,
+      "step": 5119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0726881824650561e-05,
+      "loss": 1.0342,
+      "step": 5120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723774299339822e-05,
+      "loss": 1.1904,
+      "step": 5121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0720666703766617e-05,
+      "loss": 1.1804,
+      "step": 5122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071755903823263e-05,
+      "loss": 1.0608,
+      "step": 5123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0714451303039545e-05,
+      "loss": 1.1387,
+      "step": 5124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0711343498489055e-05,
+      "loss": 1.1431,
+      "step": 5125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0708235624882857e-05,
+      "loss": 1.0776,
+      "step": 5126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0705127682522658e-05,
+      "loss": 1.1569,
+      "step": 5127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070201967171017e-05,
+      "loss": 0.9714,
+      "step": 5128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0698911592747114e-05,
+      "loss": 0.9753,
+      "step": 5129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0695803445935213e-05,
+      "loss": 1.0662,
+      "step": 5130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0692695231576202e-05,
+      "loss": 1.0828,
+      "step": 5131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0689586949971818e-05,
+      "loss": 1.1145,
+      "step": 5132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0686478601423804e-05,
+      "loss": 0.9921,
+      "step": 5133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0683370186233917e-05,
+      "loss": 0.9825,
+      "step": 5134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.068026170470391e-05,
+      "loss": 1.0153,
+      "step": 5135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0677153157135553e-05,
+      "loss": 1.1013,
+      "step": 5136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0674044543830616e-05,
+      "loss": 1.0416,
+      "step": 5137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.067093586509087e-05,
+      "loss": 1.0155,
+      "step": 5138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0667827121218108e-05,
+      "loss": 1.0173,
+      "step": 5139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0664718312514114e-05,
+      "loss": 1.1807,
+      "step": 5140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066160943928069e-05,
+      "loss": 1.1755,
+      "step": 5141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0658500501819633e-05,
+      "loss": 1.1499,
+      "step": 5142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0655391500432754e-05,
+      "loss": 1.093,
+      "step": 5143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.065228243542187e-05,
+      "loss": 1.1855,
+      "step": 5144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.06491733070888e-05,
+      "loss": 1.0016,
+      "step": 5145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0646064115735373e-05,
+      "loss": 1.2461,
+      "step": 5146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0642954861663424e-05,
+      "loss": 1.115,
+      "step": 5147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063984554517479e-05,
+      "loss": 0.9879,
+      "step": 5148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0636736166571316e-05,
+      "loss": 1.0355,
+      "step": 5149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0633626726154858e-05,
+      "loss": 1.1411,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063051722422727e-05,
+      "loss": 1.0344,
+      "step": 5151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0627407661090418e-05,
+      "loss": 1.1006,
+      "step": 5152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062429803704617e-05,
+      "loss": 1.0166,
+      "step": 5153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0621188352396406e-05,
+      "loss": 1.0569,
+      "step": 5154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0618078607443e-05,
+      "loss": 0.9875,
+      "step": 5155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0614968802487845e-05,
+      "loss": 1.0681,
+      "step": 5156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0611858937832835e-05,
+      "loss": 1.0701,
+      "step": 5157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0608749013779864e-05,
+      "loss": 1.1326,
+      "step": 5158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0605639030630841e-05,
+      "loss": 1.0098,
+      "step": 5159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602528988687675e-05,
+      "loss": 1.2053,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0599418888252284e-05,
+      "loss": 1.0538,
+      "step": 5161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0596308729626585e-05,
+      "loss": 1.1992,
+      "step": 5162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059319851311251e-05,
+      "loss": 1.0425,
+      "step": 5163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0590088239011994e-05,
+      "loss": 1.1345,
+      "step": 5164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058697790762697e-05,
+      "loss": 0.9846,
+      "step": 5165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0583867519259383e-05,
+      "loss": 1.1721,
+      "step": 5166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0580757074211187e-05,
+      "loss": 1.062,
+      "step": 5167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0577646572784339e-05,
+      "loss": 1.0408,
+      "step": 5168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057453601528079e-05,
+      "loss": 1.0759,
+      "step": 5169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0571425402002515e-05,
+      "loss": 1.2031,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0568314733251484e-05,
+      "loss": 1.0081,
+      "step": 5171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0565204009329674e-05,
+      "loss": 1.124,
+      "step": 5172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0562093230539064e-05,
+      "loss": 1.0511,
+      "step": 5173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0558982397181646e-05,
+      "loss": 1.0693,
+      "step": 5174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055587150955941e-05,
+      "loss": 1.1172,
+      "step": 5175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0552760567974359e-05,
+      "loss": 1.0591,
+      "step": 5176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0549649572728489e-05,
+      "loss": 1.147,
+      "step": 5177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0546538524123815e-05,
+      "loss": 1.1508,
+      "step": 5178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0543427422462352e-05,
+      "loss": 1.0883,
+      "step": 5179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0540316268046114e-05,
+      "loss": 1.1074,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0537205061177122e-05,
+      "loss": 1.1406,
+      "step": 5181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0534093802157417e-05,
+      "loss": 1.0532,
+      "step": 5182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0530982491289029e-05,
+      "loss": 1.1296,
+      "step": 5183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052787112887399e-05,
+      "loss": 0.9822,
+      "step": 5184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052475971521435e-05,
+      "loss": 1.0978,
+      "step": 5185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052164825061216e-05,
+      "loss": 1.0635,
+      "step": 5186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0518536735369476e-05,
+      "loss": 1.1021,
+      "step": 5187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051542516978835e-05,
+      "loss": 1.0776,
+      "step": 5188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0512313554170848e-05,
+      "loss": 1.1731,
+      "step": 5189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0509201888819045e-05,
+      "loss": 1.1392,
+      "step": 5190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0506090174035008e-05,
+      "loss": 0.9744,
+      "step": 5191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0502978410120821e-05,
+      "loss": 1.037,
+      "step": 5192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0499866597378561e-05,
+      "loss": 1.0555,
+      "step": 5193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0496754736110324e-05,
+      "loss": 1.0938,
+      "step": 5194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0493642826618198e-05,
+      "loss": 1.2067,
+      "step": 5195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0490530869204281e-05,
+      "loss": 1.0175,
+      "step": 5196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0487418864170672e-05,
+      "loss": 1.0426,
+      "step": 5197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0484306811819487e-05,
+      "loss": 1.0303,
+      "step": 5198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048119471245283e-05,
+      "loss": 1.0703,
+      "step": 5199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0478082566372817e-05,
+      "loss": 1.1279,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0474970373881568e-05,
+      "loss": 1.0908,
+      "step": 5201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0471858135281215e-05,
+      "loss": 1.0901,
+      "step": 5202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046874585087388e-05,
+      "loss": 1.1616,
+      "step": 5203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0465633520961697e-05,
+      "loss": 1.1582,
+      "step": 5204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0462521145846811e-05,
+      "loss": 1.1414,
+      "step": 5205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045940872583136e-05,
+      "loss": 1.073,
+      "step": 5206
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0456296261217491e-05,
+      "loss": 1.021,
+      "step": 5207
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0453183752307354e-05,
+      "loss": 1.0504,
+      "step": 5208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0450071199403112e-05,
+      "loss": 1.0851,
+      "step": 5209
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0446958602806916e-05,
+      "loss": 1.0023,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0443845962820936e-05,
+      "loss": 1.0994,
+      "step": 5211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0440733279747332e-05,
+      "loss": 1.0643,
+      "step": 5212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0437620553888293e-05,
+      "loss": 1.1482,
+      "step": 5213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043450778554598e-05,
+      "loss": 1.1492,
+      "step": 5214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043139497502258e-05,
+      "loss": 1.0801,
+      "step": 5215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042828212262028e-05,
+      "loss": 0.9116,
+      "step": 5216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0425169228641264e-05,
+      "loss": 1.0564,
+      "step": 5217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0422056293387728e-05,
+      "loss": 1.0923,
+      "step": 5218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0418943317161872e-05,
+      "loss": 1.0167,
+      "step": 5219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041583030026589e-05,
+      "loss": 1.0403,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0412717243001993e-05,
+      "loss": 1.103,
+      "step": 5221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040960414567239e-05,
+      "loss": 0.9114,
+      "step": 5222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040649100857929e-05,
+      "loss": 1.0947,
+      "step": 5223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040337783202491e-05,
+      "loss": 1.0698,
+      "step": 5224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0400264616311473e-05,
+      "loss": 1.027,
+      "step": 5225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0397151361741204e-05,
+      "loss": 1.0891,
+      "step": 5226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0394038068616328e-05,
+      "loss": 1.0853,
+      "step": 5227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039092473723908e-05,
+      "loss": 1.0183,
+      "step": 5228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0387811367911694e-05,
+      "loss": 1.0715,
+      "step": 5229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0384697960936408e-05,
+      "loss": 1.2502,
+      "step": 5230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0381584516615464e-05,
+      "loss": 1.1477,
+      "step": 5231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0378471035251112e-05,
+      "loss": 1.1636,
+      "step": 5232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0375357517145603e-05,
+      "loss": 1.1796,
+      "step": 5233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0372243962601185e-05,
+      "loss": 1.1531,
+      "step": 5234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0369130371920121e-05,
+      "loss": 0.9888,
+      "step": 5235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036601674540467e-05,
+      "loss": 1.0773,
+      "step": 5236
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0362903083357092e-05,
+      "loss": 1.146,
+      "step": 5237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0359789386079662e-05,
+      "loss": 1.0093,
+      "step": 5238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0356675653874646e-05,
+      "loss": 1.1506,
+      "step": 5239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0353561887044319e-05,
+      "loss": 1.241,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0350448085890958e-05,
+      "loss": 0.9359,
+      "step": 5241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034733425071685e-05,
+      "loss": 0.9746,
+      "step": 5242
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0344220381824273e-05,
+      "loss": 1.0854,
+      "step": 5243
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0341106479515517e-05,
+      "loss": 1.0789,
+      "step": 5244
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0337992544092872e-05,
+      "loss": 1.2278,
+      "step": 5245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0334878575858634e-05,
+      "loss": 1.1108,
+      "step": 5246
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.03317645751151e-05,
+      "loss": 1.1528,
+      "step": 5247
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.032865054216457e-05,
+      "loss": 1.1396,
+      "step": 5248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0325536477309349e-05,
+      "loss": 0.9979,
+      "step": 5249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0322422380851737e-05,
+      "loss": 1.0283,
+      "step": 5250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0319308253094054e-05,
+      "loss": 1.0615,
+      "step": 5251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0316194094338604e-05,
+      "loss": 1.1787,
+      "step": 5252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0313079904887712e-05,
+      "loss": 1.1052,
+      "step": 5253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030996568504369e-05,
+      "loss": 1.052,
+      "step": 5254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030685143510886e-05,
+      "loss": 1.0615,
+      "step": 5255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0303737155385552e-05,
+      "loss": 1.0918,
+      "step": 5256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030062284617609e-05,
+      "loss": 1.0486,
+      "step": 5257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0297508507782799e-05,
+      "loss": 0.9834,
+      "step": 5258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0294394140508022e-05,
+      "loss": 1.2092,
+      "step": 5259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0291279744654093e-05,
+      "loss": 1.2146,
+      "step": 5260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0288165320523347e-05,
+      "loss": 1.0806,
+      "step": 5261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0285050868418123e-05,
+      "loss": 1.0542,
+      "step": 5262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0281936388640774e-05,
+      "loss": 1.1423,
+      "step": 5263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278821881493644e-05,
+      "loss": 1.1106,
+      "step": 5264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027570734727908e-05,
+      "loss": 1.0085,
+      "step": 5265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0272592786299433e-05,
+      "loss": 1.0618,
+      "step": 5266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0269478198857066e-05,
+      "loss": 1.1709,
+      "step": 5267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0266363585254331e-05,
+      "loss": 1.106,
+      "step": 5268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0263248945793585e-05,
+      "loss": 1.0896,
+      "step": 5269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0260134280777193e-05,
+      "loss": 1.1465,
+      "step": 5270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0257019590507527e-05,
+      "loss": 1.1497,
+      "step": 5271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0253904875286944e-05,
+      "loss": 1.1478,
+      "step": 5272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0250790135417816e-05,
+      "loss": 1.1189,
+      "step": 5273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0247675371202521e-05,
+      "loss": 0.9617,
+      "step": 5274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0244560582943432e-05,
+      "loss": 0.9491,
+      "step": 5275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0241445770942924e-05,
+      "loss": 1.186,
+      "step": 5276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023833093550337e-05,
+      "loss": 0.9941,
+      "step": 5277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0235216076927163e-05,
+      "loss": 1.0684,
+      "step": 5278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0232101195516686e-05,
+      "loss": 1.1055,
+      "step": 5279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0228986291574317e-05,
+      "loss": 1.0544,
+      "step": 5280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0225871365402445e-05,
+      "loss": 1.0369,
+      "step": 5281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0222756417303468e-05,
+      "loss": 1.0291,
+      "step": 5282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0219641447579772e-05,
+      "loss": 1.0615,
+      "step": 5283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0216526456533756e-05,
+      "loss": 1.0476,
+      "step": 5284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021341144446781e-05,
+      "loss": 1.1174,
+      "step": 5285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021029641168434e-05,
+      "loss": 1.1125,
+      "step": 5286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0207181358485742e-05,
+      "loss": 1.2019,
+      "step": 5287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0204066285174423e-05,
+      "loss": 1.0305,
+      "step": 5288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0200951192052781e-05,
+      "loss": 1.084,
+      "step": 5289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197836079423231e-05,
+      "loss": 1.1277,
+      "step": 5290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0194720947588178e-05,
+      "loss": 1.0874,
+      "step": 5291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0191605796850027e-05,
+      "loss": 1.1248,
+      "step": 5292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0188490627511196e-05,
+      "loss": 1.1851,
+      "step": 5293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0185375439874098e-05,
+      "loss": 1.0237,
+      "step": 5294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018226023424115e-05,
+      "loss": 1.0338,
+      "step": 5295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0179145010914769e-05,
+      "loss": 1.009,
+      "step": 5296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0176029770197369e-05,
+      "loss": 1.0594,
+      "step": 5297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0172914512391379e-05,
+      "loss": 0.9708,
+      "step": 5298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0169799237799214e-05,
+      "loss": 1.0579,
+      "step": 5299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166683946723303e-05,
+      "loss": 1.053,
+      "step": 5300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0163568639466073e-05,
+      "loss": 1.0862,
+      "step": 5301
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0160453316329946e-05,
+      "loss": 1.0665,
+      "step": 5302
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0157337977617357e-05,
+      "loss": 1.106,
+      "step": 5303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0154222623630732e-05,
+      "loss": 1.2026,
+      "step": 5304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0151107254672503e-05,
+      "loss": 1.0775,
+      "step": 5305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0147991871045104e-05,
+      "loss": 1.0986,
+      "step": 5306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0144876473050973e-05,
+      "loss": 1.0066,
+      "step": 5307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0141761060992544e-05,
+      "loss": 1.0864,
+      "step": 5308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0138645635172251e-05,
+      "loss": 1.1135,
+      "step": 5309
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0135530195892537e-05,
+      "loss": 1.1367,
+      "step": 5310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0132414743455842e-05,
+      "loss": 0.9606,
+      "step": 5311
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0129299278164609e-05,
+      "loss": 1.0479,
+      "step": 5312
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0126183800321274e-05,
+      "loss": 0.9836,
+      "step": 5313
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0123068310228287e-05,
+      "loss": 1.0107,
+      "step": 5314
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0119952808188094e-05,
+      "loss": 0.9758,
+      "step": 5315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116837294503134e-05,
+      "loss": 1.0741,
+      "step": 5316
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0113721769475863e-05,
+      "loss": 1.0586,
+      "step": 5317
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0110606233408723e-05,
+      "loss": 1.1064,
+      "step": 5318
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0107490686604169e-05,
+      "loss": 1.0504,
+      "step": 5319
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0104375129364648e-05,
+      "loss": 1.0551,
+      "step": 5320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0101259561992612e-05,
+      "loss": 1.099,
+      "step": 5321
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0098143984790513e-05,
+      "loss": 1.0878,
+      "step": 5322
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0095028398060807e-05,
+      "loss": 1.0603,
+      "step": 5323
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0091912802105947e-05,
+      "loss": 1.0762,
+      "step": 5324
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0088797197228389e-05,
+      "loss": 1.1157,
+      "step": 5325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008568158373059e-05,
+      "loss": 1.04,
+      "step": 5326
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0082565961915001e-05,
+      "loss": 1.1436,
+      "step": 5327
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0079450332084089e-05,
+      "loss": 0.9762,
+      "step": 5328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0076334694540308e-05,
+      "loss": 1.0598,
+      "step": 5329
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0073219049586118e-05,
+      "loss": 1.1167,
+      "step": 5330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0070103397523979e-05,
+      "loss": 1.0044,
+      "step": 5331
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006698773865635e-05,
+      "loss": 1.1221,
+      "step": 5332
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0063872073285696e-05,
+      "loss": 1.2236,
+      "step": 5333
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0060756401714479e-05,
+      "loss": 1.0494,
+      "step": 5334
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0057640724245158e-05,
+      "loss": 1.2339,
+      "step": 5335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0054525041180201e-05,
+      "loss": 1.1648,
+      "step": 5336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005140935282207e-05,
+      "loss": 1.0227,
+      "step": 5337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0048293659473225e-05,
+      "loss": 0.9399,
+      "step": 5338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0045177961436136e-05,
+      "loss": 0.8982,
+      "step": 5339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042062259013269e-05,
+      "loss": 0.9667,
+      "step": 5340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003894655250709e-05,
+      "loss": 1.0187,
+      "step": 5341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0035830842220059e-05,
+      "loss": 1.0227,
+      "step": 5342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0032715128454646e-05,
+      "loss": 1.0421,
+      "step": 5343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002959941151332e-05,
+      "loss": 1.0476,
+      "step": 5344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0026483691698548e-05,
+      "loss": 1.0923,
+      "step": 5345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002336796931279e-05,
+      "loss": 0.9956,
+      "step": 5346
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0020252244658524e-05,
+      "loss": 1.1201,
+      "step": 5347
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0017136518038215e-05,
+      "loss": 1.179,
+      "step": 5348
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0014020789754327e-05,
+      "loss": 1.0798,
+      "step": 5349
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0010905060109329e-05,
+      "loss": 1.1362,
+      "step": 5350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000778932940569e-05,
+      "loss": 1.2003,
+      "step": 5351
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0004673597945886e-05,
+      "loss": 1.1338,
+      "step": 5352
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0001557866032374e-05,
+      "loss": 1.114,
+      "step": 5353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.998442133967628e-06,
+      "loss": 1.0818,
+      "step": 5354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.995326402054117e-06,
+      "loss": 1.0261,
+      "step": 5355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.992210670594309e-06,
+      "loss": 0.9434,
+      "step": 5356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.989094939890673e-06,
+      "loss": 1.0123,
+      "step": 5357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.985979210245678e-06,
+      "loss": 1.0762,
+      "step": 5358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98286348196179e-06,
+      "loss": 1.1675,
+      "step": 5359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.979747755341479e-06,
+      "loss": 0.8909,
+      "step": 5360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.976632030687212e-06,
+      "loss": 1.1365,
+      "step": 5361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.973516308301457e-06,
+      "loss": 1.2249,
+      "step": 5362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.970400588486681e-06,
+      "loss": 1.2216,
+      "step": 5363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.967284871545356e-06,
+      "loss": 1.0233,
+      "step": 5364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.964169157779946e-06,
+      "loss": 1.0704,
+      "step": 5365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.961053447492915e-06,
+      "loss": 1.1316,
+      "step": 5366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.957937740986732e-06,
+      "loss": 1.0708,
+      "step": 5367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.954822038563865e-06,
+      "loss": 1.052,
+      "step": 5368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.951706340526777e-06,
+      "loss": 1.0682,
+      "step": 5369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.948590647177933e-06,
+      "loss": 1.1772,
+      "step": 5370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.945474958819799e-06,
+      "loss": 0.988,
+      "step": 5371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.942359275754843e-06,
+      "loss": 1.136,
+      "step": 5372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.939243598285524e-06,
+      "loss": 1.2,
+      "step": 5373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.936127926714306e-06,
+      "loss": 1.1663,
+      "step": 5374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.933012261343651e-06,
+      "loss": 1.004,
+      "step": 5375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.929896602476025e-06,
+      "loss": 1.0564,
+      "step": 5376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.926780950413884e-06,
+      "loss": 1.1174,
+      "step": 5377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.923665305459693e-06,
+      "loss": 1.1316,
+      "step": 5378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.920549667915911e-06,
+      "loss": 1.1946,
+      "step": 5379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.917434038085e-06,
+      "loss": 0.9387,
+      "step": 5380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.914318416269416e-06,
+      "loss": 1.0902,
+      "step": 5381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.911202802771615e-06,
+      "loss": 1.0879,
+      "step": 5382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.908087197894056e-06,
+      "loss": 1.0144,
+      "step": 5383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.904971601939197e-06,
+      "loss": 1.1066,
+      "step": 5384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.901856015209488e-06,
+      "loss": 1.0425,
+      "step": 5385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89874043800739e-06,
+      "loss": 1.0408,
+      "step": 5386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895624870635357e-06,
+      "loss": 0.9628,
+      "step": 5387
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.892509313395835e-06,
+      "loss": 0.9622,
+      "step": 5388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.889393766591278e-06,
+      "loss": 1.0778,
+      "step": 5389
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88627823052414e-06,
+      "loss": 1.1011,
+      "step": 5390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883162705496867e-06,
+      "loss": 1.1322,
+      "step": 5391
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88004719181191e-06,
+      "loss": 1.1445,
+      "step": 5392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.876931689771713e-06,
+      "loss": 1.2876,
+      "step": 5393
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.873816199678731e-06,
+      "loss": 1.0483,
+      "step": 5394
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.870700721835396e-06,
+      "loss": 1.0605,
+      "step": 5395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86758525654416e-06,
+      "loss": 1.0922,
+      "step": 5396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864469804107464e-06,
+      "loss": 0.872,
+      "step": 5397
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86135436482775e-06,
+      "loss": 0.9927,
+      "step": 5398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.858238939007461e-06,
+      "loss": 1.0415,
+      "step": 5399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85512352694903e-06,
+      "loss": 0.9495,
+      "step": 5400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.852008128954895e-06,
+      "loss": 1.173,
+      "step": 5401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.848892745327504e-06,
+      "loss": 1.0774,
+      "step": 5402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.845777376369272e-06,
+      "loss": 1.1045,
+      "step": 5403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.842662022382647e-06,
+      "loss": 1.0593,
+      "step": 5404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.839546683670055e-06,
+      "loss": 1.197,
+      "step": 5405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83643136053393e-06,
+      "loss": 1.0342,
+      "step": 5406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833316053276699e-06,
+      "loss": 1.0137,
+      "step": 5407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.830200762200786e-06,
+      "loss": 0.9722,
+      "step": 5408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.827085487608628e-06,
+      "loss": 1.0139,
+      "step": 5409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.823970229802635e-06,
+      "loss": 1.0608,
+      "step": 5410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.820854989085235e-06,
+      "loss": 1.0781,
+      "step": 5411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.817739765758853e-06,
+      "loss": 1.0474,
+      "step": 5412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.814624560125903e-06,
+      "loss": 1.1035,
+      "step": 5413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.811509372488805e-06,
+      "loss": 1.0466,
+      "step": 5414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808394203149976e-06,
+      "loss": 1.0842,
+      "step": 5415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.805279052411829e-06,
+      "loss": 1.1184,
+      "step": 5416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802163920576774e-06,
+      "loss": 1.1365,
+      "step": 5417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.799048807947222e-06,
+      "loss": 1.0035,
+      "step": 5418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79593371482558e-06,
+      "loss": 1.0211,
+      "step": 5419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79281864151426e-06,
+      "loss": 1.146,
+      "step": 5420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789703588315663e-06,
+      "loss": 1.1047,
+      "step": 5421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.786588555532192e-06,
+      "loss": 1.1089,
+      "step": 5422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.783473543466249e-06,
+      "loss": 1.058,
+      "step": 5423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780358552420233e-06,
+      "loss": 1.208,
+      "step": 5424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.777243582696537e-06,
+      "loss": 1.2312,
+      "step": 5425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.774128634597558e-06,
+      "loss": 1.1467,
+      "step": 5426
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.771013708425688e-06,
+      "loss": 0.9419,
+      "step": 5427
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.767898804483317e-06,
+      "loss": 1.1368,
+      "step": 5428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.764783923072837e-06,
+      "loss": 1.092,
+      "step": 5429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76166906449663e-06,
+      "loss": 1.1179,
+      "step": 5430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.758554229057081e-06,
+      "loss": 0.9973,
+      "step": 5431
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.755439417056573e-06,
+      "loss": 1.0614,
+      "step": 5432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75232462879748e-06,
+      "loss": 1.0928,
+      "step": 5433
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.749209864582185e-06,
+      "loss": 1.093,
+      "step": 5434
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.746095124713058e-06,
+      "loss": 0.9966,
+      "step": 5435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.742980409492476e-06,
+      "loss": 0.9668,
+      "step": 5436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.739865719222806e-06,
+      "loss": 1.067,
+      "step": 5437
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.736751054206416e-06,
+      "loss": 1.2063,
+      "step": 5438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.733636414745674e-06,
+      "loss": 1.126,
+      "step": 5439
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.730521801142939e-06,
+      "loss": 1.0786,
+      "step": 5440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.727407213700568e-06,
+      "loss": 1.1228,
+      "step": 5441
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.724292652720922e-06,
+      "loss": 1.1077,
+      "step": 5442
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.721178118506358e-06,
+      "loss": 1.0845,
+      "step": 5443
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.718063611359226e-06,
+      "loss": 1.1211,
+      "step": 5444
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.714949131581877e-06,
+      "loss": 1.031,
+      "step": 5445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.71183467947666e-06,
+      "loss": 1.1216,
+      "step": 5446
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.708720255345912e-06,
+      "loss": 1.1201,
+      "step": 5447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.705605859491981e-06,
+      "loss": 1.0566,
+      "step": 5448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.702491492217203e-06,
+      "loss": 1.0796,
+      "step": 5449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.699377153823913e-06,
+      "loss": 1.074,
+      "step": 5450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69626284461445e-06,
+      "loss": 1.0879,
+      "step": 5451
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69314856489114e-06,
+      "loss": 1.1624,
+      "step": 5452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.690034314956313e-06,
+      "loss": 1.0189,
+      "step": 5453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.686920095112291e-06,
+      "loss": 1.1658,
+      "step": 5454
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.683805905661397e-06,
+      "loss": 1.0203,
+      "step": 5455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68069174690595e-06,
+      "loss": 1.062,
+      "step": 5456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.677577619148266e-06,
+      "loss": 1.0508,
+      "step": 5457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.674463522690656e-06,
+      "loss": 1.0454,
+      "step": 5458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.671349457835432e-06,
+      "loss": 1.0793,
+      "step": 5459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.668235424884902e-06,
+      "loss": 1.1248,
+      "step": 5460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66512142414137e-06,
+      "loss": 1.0665,
+      "step": 5461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.662007455907131e-06,
+      "loss": 1.1255,
+      "step": 5462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658893520484486e-06,
+      "loss": 1.0374,
+      "step": 5463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.655779618175732e-06,
+      "loss": 1.0256,
+      "step": 5464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.652665749283154e-06,
+      "loss": 0.8899,
+      "step": 5465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649551914109042e-06,
+      "loss": 1.0205,
+      "step": 5466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.646438112955681e-06,
+      "loss": 1.136,
+      "step": 5467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64332434612536e-06,
+      "loss": 1.103,
+      "step": 5468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.640210613920341e-06,
+      "loss": 1.1736,
+      "step": 5469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63709691664291e-06,
+      "loss": 1.105,
+      "step": 5470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.633983254595334e-06,
+      "loss": 1.1428,
+      "step": 5471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.630869628079882e-06,
+      "loss": 1.0658,
+      "step": 5472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.627756037398816e-06,
+      "loss": 1.1179,
+      "step": 5473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.624642482854399e-06,
+      "loss": 1.1143,
+      "step": 5474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.621528964748891e-06,
+      "loss": 1.0759,
+      "step": 5475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.618415483384539e-06,
+      "loss": 1.0869,
+      "step": 5476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.615302039063597e-06,
+      "loss": 1.0339,
+      "step": 5477
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.612188632088311e-06,
+      "loss": 1.0653,
+      "step": 5478
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609075262760922e-06,
+      "loss": 0.9581,
+      "step": 5479
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.605961931383674e-06,
+      "loss": 1.0159,
+      "step": 5480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.602848638258798e-06,
+      "loss": 1.1633,
+      "step": 5481
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.599735383688527e-06,
+      "loss": 1.2002,
+      "step": 5482
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596622167975094e-06,
+      "loss": 1.0023,
+      "step": 5483
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.593508991420715e-06,
+      "loss": 1.1069,
+      "step": 5484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.590395854327614e-06,
+      "loss": 1.0715,
+      "step": 5485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58728275699801e-06,
+      "loss": 1.1267,
+      "step": 5486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.584169699734113e-06,
+      "loss": 1.0168,
+      "step": 5487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.581056682838133e-06,
+      "loss": 1.0864,
+      "step": 5488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.577943706612272e-06,
+      "loss": 1.1201,
+      "step": 5489
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.574830771358741e-06,
+      "loss": 1.0358,
+      "step": 5490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.571717877379727e-06,
+      "loss": 1.1812,
+      "step": 5491
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.568605024977423e-06,
+      "loss": 1.135,
+      "step": 5492
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.565492214454022e-06,
+      "loss": 1.0082,
+      "step": 5493
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.562379446111712e-06,
+      "loss": 1.0297,
+      "step": 5494
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.559266720252666e-06,
+      "loss": 0.9586,
+      "step": 5495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.556154037179069e-06,
+      "loss": 1.0815,
+      "step": 5496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.553041397193089e-06,
+      "loss": 1.082,
+      "step": 5497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.549928800596893e-06,
+      "loss": 1.0935,
+      "step": 5498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54681624769265e-06,
+      "loss": 1.1567,
+      "step": 5499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.543703738782512e-06,
+      "loss": 1.0789,
+      "step": 5500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.540591274168643e-06,
+      "loss": 1.0984,
+      "step": 5501
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53747885415319e-06,
+      "loss": 0.9653,
+      "step": 5502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.534366479038304e-06,
+      "loss": 1.1689,
+      "step": 5503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.531254149126124e-06,
+      "loss": 0.9576,
+      "step": 5504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.52814186471879e-06,
+      "loss": 1.0464,
+      "step": 5505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.525029626118436e-06,
+      "loss": 0.9819,
+      "step": 5506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.521917433627187e-06,
+      "loss": 1.02,
+      "step": 5507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.518805287547175e-06,
+      "loss": 0.8936,
+      "step": 5508
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515693188180518e-06,
+      "loss": 0.9094,
+      "step": 5509
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.512581135829328e-06,
+      "loss": 1.1577,
+      "step": 5510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.509469130795724e-06,
+      "loss": 1.0901,
+      "step": 5511
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.506357173381807e-06,
+      "loss": 1.0382,
+      "step": 5512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.503245263889681e-06,
+      "loss": 1.1523,
+      "step": 5513
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.500133402621442e-06,
+      "loss": 1.0547,
+      "step": 5514
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.497021589879182e-06,
+      "loss": 1.2136,
+      "step": 5515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.493909825964994e-06,
+      "loss": 1.1111,
+      "step": 5516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.490798111180957e-06,
+      "loss": 1.0471,
+      "step": 5517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487686445829153e-06,
+      "loss": 1.1042,
+      "step": 5518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.484574830211655e-06,
+      "loss": 1.2048,
+      "step": 5519
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48146326463053e-06,
+      "loss": 1.1084,
+      "step": 5520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.478351749387843e-06,
+      "loss": 1.1566,
+      "step": 5521
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.475240284785653e-06,
+      "loss": 0.9861,
+      "step": 5522
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472128871126014e-06,
+      "loss": 1.145,
+      "step": 5523
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.469017508710976e-06,
+      "loss": 1.1848,
+      "step": 5524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.465906197842583e-06,
+      "loss": 1.1052,
+      "step": 5525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.462794938822878e-06,
+      "loss": 1.0496,
+      "step": 5526
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459683731953891e-06,
+      "loss": 1.066,
+      "step": 5527
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.456572577537653e-06,
+      "loss": 1.1133,
+      "step": 5528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.453461475876188e-06,
+      "loss": 1.1157,
+      "step": 5529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.450350427271513e-06,
+      "loss": 1.1333,
+      "step": 5530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.447239432025643e-06,
+      "loss": 1.0898,
+      "step": 5531
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44412849044059e-06,
+      "loss": 1.0242,
+      "step": 5532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.441017602818356e-06,
+      "loss": 1.0527,
+      "step": 5533
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.437906769460938e-06,
+      "loss": 1.0828,
+      "step": 5534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43479599067033e-06,
+      "loss": 1.1826,
+      "step": 5535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.431685266748517e-06,
+      "loss": 1.1177,
+      "step": 5536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.428574597997487e-06,
+      "loss": 0.9346,
+      "step": 5537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425463984719212e-06,
+      "loss": 0.9758,
+      "step": 5538
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.422353427215664e-06,
+      "loss": 1.066,
+      "step": 5539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.419242925788813e-06,
+      "loss": 1.051,
+      "step": 5540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.416132480740619e-06,
+      "loss": 1.0503,
+      "step": 5541
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413022092373034e-06,
+      "loss": 1.0068,
+      "step": 5542
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.409911760988011e-06,
+      "loss": 1.0249,
+      "step": 5543
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.406801486887493e-06,
+      "loss": 1.1216,
+      "step": 5544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.403691270373418e-06,
+      "loss": 1.0881,
+      "step": 5545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40058111174772e-06,
+      "loss": 1.0388,
+      "step": 5546
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397471011312325e-06,
+      "loss": 1.1008,
+      "step": 5547
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.39436096936916e-06,
+      "loss": 1.0374,
+      "step": 5548
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.391250986220138e-06,
+      "loss": 1.0278,
+      "step": 5549
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388141062167168e-06,
+      "loss": 1.0763,
+      "step": 5550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.385031197512158e-06,
+      "loss": 0.9521,
+      "step": 5551
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.381921392557003e-06,
+      "loss": 1.0409,
+      "step": 5552
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.378811647603598e-06,
+      "loss": 1.1318,
+      "step": 5553
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.375701962953831e-06,
+      "loss": 1.0454,
+      "step": 5554
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.372592338909584e-06,
+      "loss": 1.054,
+      "step": 5555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.369482775772734e-06,
+      "loss": 1.1064,
+      "step": 5556
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.366373273845145e-06,
+      "loss": 1.1416,
+      "step": 5557
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.363263833428686e-06,
+      "loss": 1.145,
+      "step": 5558
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.360154454825215e-06,
+      "loss": 1.0166,
+      "step": 5559
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.35704513833658e-06,
+      "loss": 1.071,
+      "step": 5560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.353935884264629e-06,
+      "loss": 1.0978,
+      "step": 5561
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350826692911201e-06,
+      "loss": 1.2578,
+      "step": 5562
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.347717564578136e-06,
+      "loss": 1.0179,
+      "step": 5563
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.344608499567251e-06,
+      "loss": 1.1355,
+      "step": 5564
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34149949818037e-06,
+      "loss": 1.0632,
+      "step": 5565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.338390560719314e-06,
+      "loss": 1.0452,
+      "step": 5566
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.335281687485887e-06,
+      "loss": 1.0459,
+      "step": 5567
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.332172878781896e-06,
+      "loss": 1.1628,
+      "step": 5568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.329064134909131e-06,
+      "loss": 1.0212,
+      "step": 5569
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.325955456169386e-06,
+      "loss": 1.0967,
+      "step": 5570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.322846842864452e-06,
+      "loss": 1.0352,
+      "step": 5571
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.319738295296094e-06,
+      "loss": 0.996,
+      "step": 5572
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.316629813766087e-06,
+      "loss": 1.0841,
+      "step": 5573
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.313521398576199e-06,
+      "loss": 1.0232,
+      "step": 5574
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.310413050028187e-06,
+      "loss": 1.115,
+      "step": 5575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.307304768423801e-06,
+      "loss": 1.1609,
+      "step": 5576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.304196554064787e-06,
+      "loss": 1.0081,
+      "step": 5577
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30108840725289e-06,
+      "loss": 1.0266,
+      "step": 5578
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.297980328289835e-06,
+      "loss": 0.9978,
+      "step": 5579
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.294872317477345e-06,
+      "loss": 1.0035,
+      "step": 5580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.291764375117146e-06,
+      "loss": 1.2488,
+      "step": 5581
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.288656501510948e-06,
+      "loss": 1.1567,
+      "step": 5582
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.285548696960458e-06,
+      "loss": 1.0327,
+      "step": 5583
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.282440961767372e-06,
+      "loss": 1.0338,
+      "step": 5584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.279333296233386e-06,
+      "loss": 1.1033,
+      "step": 5585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276225700660185e-06,
+      "loss": 1.2021,
+      "step": 5586
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.273118175349442e-06,
+      "loss": 1.0718,
+      "step": 5587
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.270010720602833e-06,
+      "loss": 1.1904,
+      "step": 5588
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.266903336722025e-06,
+      "loss": 0.9683,
+      "step": 5589
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.263796024008675e-06,
+      "loss": 1.2292,
+      "step": 5590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.260688782764432e-06,
+      "loss": 1.0613,
+      "step": 5591
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.257581613290943e-06,
+      "loss": 1.1516,
+      "step": 5592
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.254474515889848e-06,
+      "loss": 1.0742,
+      "step": 5593
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25136749086277e-06,
+      "loss": 1.0436,
+      "step": 5594
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.248260538511337e-06,
+      "loss": 0.9635,
+      "step": 5595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.245153659137162e-06,
+      "loss": 1.0632,
+      "step": 5596
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.242046853041858e-06,
+      "loss": 1.1597,
+      "step": 5597
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.238940120527027e-06,
+      "loss": 1.1707,
+      "step": 5598
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.235833461894262e-06,
+      "loss": 1.0898,
+      "step": 5599
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.232726877445154e-06,
+      "loss": 1.1791,
+      "step": 5600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.229620367481279e-06,
+      "loss": 1.1072,
+      "step": 5601
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22651393230421e-06,
+      "loss": 1.1587,
+      "step": 5602
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22340757221552e-06,
+      "loss": 1.1152,
+      "step": 5603
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.220301287516757e-06,
+      "loss": 1.05,
+      "step": 5604
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.217195078509483e-06,
+      "loss": 1.0977,
+      "step": 5605
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21408894549524e-06,
+      "loss": 1.0316,
+      "step": 5606
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.210982888775559e-06,
+      "loss": 1.1184,
+      "step": 5607
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.207876908651977e-06,
+      "loss": 1.1221,
+      "step": 5608
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.20477100542601e-06,
+      "loss": 1.2202,
+      "step": 5609
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201665179399172e-06,
+      "loss": 1.0139,
+      "step": 5610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.198559430872972e-06,
+      "loss": 1.0562,
+      "step": 5611
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195453760148912e-06,
+      "loss": 1.0477,
+      "step": 5612
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192348167528482e-06,
+      "loss": 1.1636,
+      "step": 5613
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.189242653313165e-06,
+      "loss": 1.0145,
+      "step": 5614
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18613721780444e-06,
+      "loss": 0.9648,
+      "step": 5615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.183031861303773e-06,
+      "loss": 1.1306,
+      "step": 5616
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179926584112629e-06,
+      "loss": 1.1733,
+      "step": 5617
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.176821386532458e-06,
+      "loss": 1.0675,
+      "step": 5618
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.173716268864706e-06,
+      "loss": 1.1851,
+      "step": 5619
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.170611231410816e-06,
+      "loss": 1.0951,
+      "step": 5620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.167506274472214e-06,
+      "loss": 1.0994,
+      "step": 5621
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.164401398350326e-06,
+      "loss": 1.0177,
+      "step": 5622
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.161296603346563e-06,
+      "loss": 1.103,
+      "step": 5623
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.158191889762333e-06,
+      "loss": 1.0427,
+      "step": 5624
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.155087257899034e-06,
+      "loss": 1.0676,
+      "step": 5625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15198270805806e-06,
+      "loss": 1.1101,
+      "step": 5626
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14887824054079e-06,
+      "loss": 0.9995,
+      "step": 5627
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.145773855648605e-06,
+      "loss": 0.9105,
+      "step": 5628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.142669553682868e-06,
+      "loss": 1.0779,
+      "step": 5629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13956533494494e-06,
+      "loss": 1.0518,
+      "step": 5630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.136461199736167e-06,
+      "loss": 1.0525,
+      "step": 5631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.133357148357897e-06,
+      "loss": 1.1475,
+      "step": 5632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.130253181111464e-06,
+      "loss": 1.0891,
+      "step": 5633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12714929829819e-06,
+      "loss": 1.1047,
+      "step": 5634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.124045500219399e-06,
+      "loss": 1.1243,
+      "step": 5635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.120941787176398e-06,
+      "loss": 0.856,
+      "step": 5636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.117838159470495e-06,
+      "loss": 1.1936,
+      "step": 5637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.114734617402973e-06,
+      "loss": 1.0525,
+      "step": 5638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.111631161275125e-06,
+      "loss": 1.0436,
+      "step": 5639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.108527791388224e-06,
+      "loss": 1.0496,
+      "step": 5640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10542450804354e-06,
+      "loss": 1.0347,
+      "step": 5641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.102321311542333e-06,
+      "loss": 1.0901,
+      "step": 5642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.099218202185852e-06,
+      "loss": 1.0962,
+      "step": 5643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09611518027535e-06,
+      "loss": 1.1577,
+      "step": 5644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.093012246112049e-06,
+      "loss": 1.1619,
+      "step": 5645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.089909399997183e-06,
+      "loss": 1.1528,
+      "step": 5646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.086806642231965e-06,
+      "loss": 1.0625,
+      "step": 5647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.08370397311761e-06,
+      "loss": 1.0861,
+      "step": 5648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.080601392955313e-06,
+      "loss": 1.0353,
+      "step": 5649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.077498902046267e-06,
+      "loss": 1.0927,
+      "step": 5650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.074396500691658e-06,
+      "loss": 1.1489,
+      "step": 5651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.071294189192663e-06,
+      "loss": 1.2422,
+      "step": 5652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.068191967850436e-06,
+      "loss": 1.0502,
+      "step": 5653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.065089836966146e-06,
+      "loss": 1.0652,
+      "step": 5654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061987796840936e-06,
+      "loss": 1.0002,
+      "step": 5655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.058885847775944e-06,
+      "loss": 0.9111,
+      "step": 5656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.055783990072305e-06,
+      "loss": 1.0615,
+      "step": 5657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.052682224031136e-06,
+      "loss": 1.0867,
+      "step": 5658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.049580549953559e-06,
+      "loss": 1.146,
+      "step": 5659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046478968140665e-06,
+      "loss": 0.9984,
+      "step": 5660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.043377478893554e-06,
+      "loss": 1.1409,
+      "step": 5661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.040276082513315e-06,
+      "loss": 1.0154,
+      "step": 5662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.037174779301023e-06,
+      "loss": 0.994,
+      "step": 5663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.034073569557746e-06,
+      "loss": 0.9751,
+      "step": 5664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.030972453584543e-06,
+      "loss": 1.0696,
+      "step": 5665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.027871431682467e-06,
+      "loss": 1.0664,
+      "step": 5666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02477050415255e-06,
+      "loss": 1.1744,
+      "step": 5667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.021669671295829e-06,
+      "loss": 1.0895,
+      "step": 5668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.018568933413326e-06,
+      "loss": 1.075,
+      "step": 5669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015468290806051e-06,
+      "loss": 1.0852,
+      "step": 5670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.012367743775015e-06,
+      "loss": 1.0349,
+      "step": 5671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.009267292621206e-06,
+      "loss": 1.014,
+      "step": 5672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00616693764561e-06,
+      "loss": 0.9738,
+      "step": 5673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.003066679149208e-06,
+      "loss": 1.1018,
+      "step": 5674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.999966517432958e-06,
+      "loss": 0.9834,
+      "step": 5675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99686645279782e-06,
+      "loss": 1.0073,
+      "step": 5676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.993766485544748e-06,
+      "loss": 1.0662,
+      "step": 5677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.990666615974672e-06,
+      "loss": 1.1089,
+      "step": 5678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.987566844388524e-06,
+      "loss": 1.0806,
+      "step": 5679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984467171087226e-06,
+      "loss": 1.113,
+      "step": 5680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.981367596371685e-06,
+      "loss": 1.1167,
+      "step": 5681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.978268120542798e-06,
+      "loss": 1.0345,
+      "step": 5682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.975168743901459e-06,
+      "loss": 1.0664,
+      "step": 5683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.972069466748546e-06,
+      "loss": 1.0298,
+      "step": 5684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.968970289384935e-06,
+      "loss": 0.9877,
+      "step": 5685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.965871212111486e-06,
+      "loss": 1.0596,
+      "step": 5686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962772235229049e-06,
+      "loss": 0.9882,
+      "step": 5687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.959673359038468e-06,
+      "loss": 1.1177,
+      "step": 5688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.956574583840573e-06,
+      "loss": 1.0529,
+      "step": 5689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.953475909936187e-06,
+      "loss": 1.0111,
+      "step": 5690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.950377337626124e-06,
+      "loss": 1.0327,
+      "step": 5691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.947278867211183e-06,
+      "loss": 1.0842,
+      "step": 5692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.944180498992163e-06,
+      "loss": 1.1055,
+      "step": 5693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941082233269842e-06,
+      "loss": 1.0674,
+      "step": 5694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.937984070344996e-06,
+      "loss": 1.0447,
+      "step": 5695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93488601051839e-06,
+      "loss": 1.0925,
+      "step": 5696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93178805409077e-06,
+      "loss": 1.0375,
+      "step": 5697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.928690201362885e-06,
+      "loss": 1.1002,
+      "step": 5698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.925592452635465e-06,
+      "loss": 0.9614,
+      "step": 5699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922494808209231e-06,
+      "loss": 1.1218,
+      "step": 5700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.919397268384901e-06,
+      "loss": 1.0718,
+      "step": 5701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916299833463176e-06,
+      "loss": 1.0837,
+      "step": 5702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913202503744748e-06,
+      "loss": 0.9268,
+      "step": 5703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.910105279530295e-06,
+      "loss": 1.1057,
+      "step": 5704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.907008161120495e-06,
+      "loss": 0.98,
+      "step": 5705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.903911148816006e-06,
+      "loss": 1.01,
+      "step": 5706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90081424291748e-06,
+      "loss": 1.0906,
+      "step": 5707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897717443725556e-06,
+      "loss": 1.1995,
+      "step": 5708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89462075154087e-06,
+      "loss": 1.03,
+      "step": 5709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891524166664041e-06,
+      "loss": 1.1401,
+      "step": 5710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.888427689395674e-06,
+      "loss": 1.1423,
+      "step": 5711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.885331320036374e-06,
+      "loss": 1.0796,
+      "step": 5712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.882235058886725e-06,
+      "loss": 1.03,
+      "step": 5713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87913890624731e-06,
+      "loss": 1.1362,
+      "step": 5714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876042862418688e-06,
+      "loss": 1.1846,
+      "step": 5715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87294692770143e-06,
+      "loss": 1.0466,
+      "step": 5716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.869851102396074e-06,
+      "loss": 1.1167,
+      "step": 5717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86675538680316e-06,
+      "loss": 0.9922,
+      "step": 5718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.863659781223208e-06,
+      "loss": 1.0454,
+      "step": 5719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.860564285956738e-06,
+      "loss": 1.0962,
+      "step": 5720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.857468901304251e-06,
+      "loss": 1.0413,
+      "step": 5721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.854373627566242e-06,
+      "loss": 1.1516,
+      "step": 5722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85127846504319e-06,
+      "loss": 1.0239,
+      "step": 5723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848183414035574e-06,
+      "loss": 1.0076,
+      "step": 5724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.845088474843853e-06,
+      "loss": 0.9889,
+      "step": 5725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.841993647768469e-06,
+      "loss": 1.0562,
+      "step": 5726
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.838898933109869e-06,
+      "loss": 1.0947,
+      "step": 5727
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83580433116848e-06,
+      "loss": 1.1445,
+      "step": 5728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.832709842244718e-06,
+      "loss": 1.0328,
+      "step": 5729
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82961546663899e-06,
+      "loss": 0.9528,
+      "step": 5730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82652120465169e-06,
+      "loss": 1.0227,
+      "step": 5731
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.823427056583208e-06,
+      "loss": 1.0835,
+      "step": 5732
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820333022733911e-06,
+      "loss": 1.0876,
+      "step": 5733
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817239103404159e-06,
+      "loss": 1.1208,
+      "step": 5734
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81414529889431e-06,
+      "loss": 1.0925,
+      "step": 5735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811051609504704e-06,
+      "loss": 1.0739,
+      "step": 5736
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.807958035535664e-06,
+      "loss": 1.1453,
+      "step": 5737
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.804864577287512e-06,
+      "loss": 1.1106,
+      "step": 5738
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80177123506055e-06,
+      "loss": 1.0601,
+      "step": 5739
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.798678009155082e-06,
+      "loss": 1.147,
+      "step": 5740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.795584899871381e-06,
+      "loss": 1.043,
+      "step": 5741
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.792491907509722e-06,
+      "loss": 0.9493,
+      "step": 5742
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.789399032370372e-06,
+      "loss": 1.1504,
+      "step": 5743
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.786306274753575e-06,
+      "loss": 1.0874,
+      "step": 5744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.783213634959573e-06,
+      "loss": 1.0667,
+      "step": 5745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.780121113288588e-06,
+      "loss": 1.2188,
+      "step": 5746
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.777028710040844e-06,
+      "loss": 1.0232,
+      "step": 5747
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.773936425516536e-06,
+      "loss": 0.9672,
+      "step": 5748
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.770844260015856e-06,
+      "loss": 0.9269,
+      "step": 5749
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.767752213838993e-06,
+      "loss": 1.041,
+      "step": 5750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.764660287286111e-06,
+      "loss": 1.1543,
+      "step": 5751
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761568480657367e-06,
+      "loss": 1.0115,
+      "step": 5752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75847679425291e-06,
+      "loss": 1.092,
+      "step": 5753
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755385228372871e-06,
+      "loss": 0.9352,
+      "step": 5754
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.752293783317379e-06,
+      "loss": 1.1332,
+      "step": 5755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.749202459386536e-06,
+      "loss": 1.1038,
+      "step": 5756
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.746111256880445e-06,
+      "loss": 1.1824,
+      "step": 5757
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.743020176099194e-06,
+      "loss": 1.0743,
+      "step": 5758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.739929217342859e-06,
+      "loss": 1.1265,
+      "step": 5759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.736838380911502e-06,
+      "loss": 1.0334,
+      "step": 5760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.733747667105176e-06,
+      "loss": 1.0405,
+      "step": 5761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730657076223922e-06,
+      "loss": 1.0991,
+      "step": 5762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.727566608567766e-06,
+      "loss": 1.092,
+      "step": 5763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724476264436724e-06,
+      "loss": 1.0717,
+      "step": 5764
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.721386044130797e-06,
+      "loss": 1.1953,
+      "step": 5765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718295947949983e-06,
+      "loss": 1.1223,
+      "step": 5766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71520597619426e-06,
+      "loss": 1.0601,
+      "step": 5767
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.712116129163595e-06,
+      "loss": 1.1328,
+      "step": 5768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.709026407157945e-06,
+      "loss": 1.054,
+      "step": 5769
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70593681047725e-06,
+      "loss": 1.1038,
+      "step": 5770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.702847339421448e-06,
+      "loss": 1.0815,
+      "step": 5771
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69975799429045e-06,
+      "loss": 1.0269,
+      "step": 5772
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.696668775384167e-06,
+      "loss": 0.9951,
+      "step": 5773
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.693579683002496e-06,
+      "loss": 1.2224,
+      "step": 5774
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.690490717445318e-06,
+      "loss": 1.0159,
+      "step": 5775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.687401879012502e-06,
+      "loss": 1.1262,
+      "step": 5776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68431316800391e-06,
+      "loss": 1.0413,
+      "step": 5777
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68122458471938e-06,
+      "loss": 1.0693,
+      "step": 5778
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67813612945875e-06,
+      "loss": 1.0729,
+      "step": 5779
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675047802521838e-06,
+      "loss": 1.1316,
+      "step": 5780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.671959604208455e-06,
+      "loss": 1.1335,
+      "step": 5781
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.668871534818397e-06,
+      "loss": 1.1125,
+      "step": 5782
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.665783594651443e-06,
+      "loss": 1.0574,
+      "step": 5783
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.662695784007371e-06,
+      "loss": 1.0034,
+      "step": 5784
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.659608103185933e-06,
+      "loss": 0.9585,
+      "step": 5785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.656520552486873e-06,
+      "loss": 1.0149,
+      "step": 5786
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65343313220993e-06,
+      "loss": 1.0432,
+      "step": 5787
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.650345842654816e-06,
+      "loss": 1.1306,
+      "step": 5788
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.647258684121247e-06,
+      "loss": 1.0432,
+      "step": 5789
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.644171656908914e-06,
+      "loss": 1.1934,
+      "step": 5790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641084761317501e-06,
+      "loss": 1.134,
+      "step": 5791
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.637997997646674e-06,
+      "loss": 1.0481,
+      "step": 5792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63491136619609e-06,
+      "loss": 1.1079,
+      "step": 5793
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.631824867265393e-06,
+      "loss": 1.1099,
+      "step": 5794
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.628738501154214e-06,
+      "loss": 1.0551,
+      "step": 5795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.625652268162169e-06,
+      "loss": 1.0471,
+      "step": 5796
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.622566168588867e-06,
+      "loss": 0.9894,
+      "step": 5797
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.619480202733898e-06,
+      "loss": 0.9623,
+      "step": 5798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.616394370896843e-06,
+      "loss": 1.0537,
+      "step": 5799
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613308673377264e-06,
+      "loss": 1.1384,
+      "step": 5800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.610223110474717e-06,
+      "loss": 1.0684,
+      "step": 5801
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607137682488736e-06,
+      "loss": 1.178,
+      "step": 5802
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.604052389718856e-06,
+      "loss": 1.0588,
+      "step": 5803
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600967232464583e-06,
+      "loss": 1.0361,
+      "step": 5804
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.597882211025423e-06,
+      "loss": 1.0959,
+      "step": 5805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.594797325700868e-06,
+      "loss": 1.1443,
+      "step": 5806
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59171257679038e-06,
+      "loss": 1.0552,
+      "step": 5807
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.588627964593424e-06,
+      "loss": 1.0828,
+      "step": 5808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58554348940945e-06,
+      "loss": 1.1583,
+      "step": 5809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.582459151537893e-06,
+      "loss": 1.0671,
+      "step": 5810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.579374951278171e-06,
+      "loss": 1.0465,
+      "step": 5811
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576290888929691e-06,
+      "loss": 1.1421,
+      "step": 5812
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.573206964791856e-06,
+      "loss": 1.166,
+      "step": 5813
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570123179164034e-06,
+      "loss": 1.0609,
+      "step": 5814
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.567039532345597e-06,
+      "loss": 1.1348,
+      "step": 5815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.563956024635905e-06,
+      "loss": 0.97,
+      "step": 5816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560872656334289e-06,
+      "loss": 1.1777,
+      "step": 5817
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.557789427740083e-06,
+      "loss": 1.0627,
+      "step": 5818
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.554706339152593e-06,
+      "loss": 1.1282,
+      "step": 5819
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.551623390871125e-06,
+      "loss": 1.1248,
+      "step": 5820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.548540583194968e-06,
+      "loss": 0.9816,
+      "step": 5821
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.545457916423384e-06,
+      "loss": 0.9998,
+      "step": 5822
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.542375390855638e-06,
+      "loss": 1.0757,
+      "step": 5823
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.539293006790975e-06,
+      "loss": 1.001,
+      "step": 5824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.536210764528625e-06,
+      "loss": 0.9495,
+      "step": 5825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533128664367807e-06,
+      "loss": 1.1177,
+      "step": 5826
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53004670660772e-06,
+      "loss": 0.9802,
+      "step": 5827
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.526964891547564e-06,
+      "loss": 1.1152,
+      "step": 5828
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.523883219486504e-06,
+      "loss": 0.9923,
+      "step": 5829
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.520801690723704e-06,
+      "loss": 1.1133,
+      "step": 5830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.517720305558318e-06,
+      "loss": 1.0022,
+      "step": 5831
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.514639064289477e-06,
+      "loss": 1.0454,
+      "step": 5832
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511557967216301e-06,
+      "loss": 1.1543,
+      "step": 5833
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.508477014637895e-06,
+      "loss": 1.0742,
+      "step": 5834
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.505396206853358e-06,
+      "loss": 1.1235,
+      "step": 5835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50231554416176e-06,
+      "loss": 1.1465,
+      "step": 5836
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49923502686217e-06,
+      "loss": 1.1116,
+      "step": 5837
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496154655253632e-06,
+      "loss": 1.0874,
+      "step": 5838
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.493074429635188e-06,
+      "loss": 1.1178,
+      "step": 5839
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48999435030586e-06,
+      "loss": 1.0564,
+      "step": 5840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.486914417564653e-06,
+      "loss": 1.0159,
+      "step": 5841
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.483834631710561e-06,
+      "loss": 1.0203,
+      "step": 5842
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.480754993042565e-06,
+      "loss": 0.9839,
+      "step": 5843
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.477675501859625e-06,
+      "loss": 1.0903,
+      "step": 5844
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.474596158460694e-06,
+      "loss": 1.0685,
+      "step": 5845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.471516963144707e-06,
+      "loss": 1.0613,
+      "step": 5846
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.468437916210588e-06,
+      "loss": 1.1404,
+      "step": 5847
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.465359017957244e-06,
+      "loss": 1.147,
+      "step": 5848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.462280268683568e-06,
+      "loss": 1.1116,
+      "step": 5849
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.459201668688437e-06,
+      "loss": 1.0779,
+      "step": 5850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.456123218270714e-06,
+      "loss": 1.1606,
+      "step": 5851
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45304491772925e-06,
+      "loss": 1.0854,
+      "step": 5852
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.449966767362877e-06,
+      "loss": 1.166,
+      "step": 5853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44688876747042e-06,
+      "loss": 1.0627,
+      "step": 5854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.443810918350683e-06,
+      "loss": 1.073,
+      "step": 5855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.440733220302455e-06,
+      "loss": 1.0981,
+      "step": 5856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43765567362452e-06,
+      "loss": 1.0059,
+      "step": 5857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.434578278615627e-06,
+      "loss": 1.0303,
+      "step": 5858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.431501035574532e-06,
+      "loss": 1.1643,
+      "step": 5859
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.428423944799962e-06,
+      "loss": 1.0969,
+      "step": 5860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.425347006590637e-06,
+      "loss": 1.0779,
+      "step": 5861
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42227022124526e-06,
+      "loss": 0.96,
+      "step": 5862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41919358906252e-06,
+      "loss": 1.1263,
+      "step": 5863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.416117110341087e-06,
+      "loss": 1.0604,
+      "step": 5864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.413040785379623e-06,
+      "loss": 1.0034,
+      "step": 5865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.409964614476766e-06,
+      "loss": 0.999,
+      "step": 5866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.406888597931145e-06,
+      "loss": 0.9456,
+      "step": 5867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.403812736041375e-06,
+      "loss": 0.9711,
+      "step": 5868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.400737029106051e-06,
+      "loss": 1.0767,
+      "step": 5869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.397661477423764e-06,
+      "loss": 1.2629,
+      "step": 5870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.394586081293073e-06,
+      "loss": 0.9072,
+      "step": 5871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.391510841012539e-06,
+      "loss": 1.0136,
+      "step": 5872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.388435756880691e-06,
+      "loss": 1.1704,
+      "step": 5873
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.385360829196056e-06,
+      "loss": 1.1187,
+      "step": 5874
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.382286058257143e-06,
+      "loss": 1.1459,
+      "step": 5875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37921144436244e-06,
+      "loss": 1.1188,
+      "step": 5876
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.376136987810426e-06,
+      "loss": 1.0062,
+      "step": 5877
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.373062688899564e-06,
+      "loss": 1.1418,
+      "step": 5878
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.369988547928302e-06,
+      "loss": 0.9919,
+      "step": 5879
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.366914565195066e-06,
+      "loss": 1.116,
+      "step": 5880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.363840740998275e-06,
+      "loss": 1.0537,
+      "step": 5881
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.360767075636328e-06,
+      "loss": 1.0837,
+      "step": 5882
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35769356940761e-06,
+      "loss": 1.0781,
+      "step": 5883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.354620222610488e-06,
+      "loss": 0.999,
+      "step": 5884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.351547035543323e-06,
+      "loss": 1.1423,
+      "step": 5885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348474008504447e-06,
+      "loss": 1.0713,
+      "step": 5886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.345401141792188e-06,
+      "loss": 1.2041,
+      "step": 5887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.342328435704847e-06,
+      "loss": 1.0793,
+      "step": 5888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33925589054072e-06,
+      "loss": 1.043,
+      "step": 5889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.336183506598081e-06,
+      "loss": 1.1736,
+      "step": 5890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.333111284175193e-06,
+      "loss": 1.1263,
+      "step": 5891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.330039223570296e-06,
+      "loss": 1.1033,
+      "step": 5892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.326967325081626e-06,
+      "loss": 1.2583,
+      "step": 5893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.323895589007395e-06,
+      "loss": 1.1316,
+      "step": 5894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.320824015645794e-06,
+      "loss": 1.0044,
+      "step": 5895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31775260529501e-06,
+      "loss": 0.95,
+      "step": 5896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.314681358253208e-06,
+      "loss": 1.0864,
+      "step": 5897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31161027481854e-06,
+      "loss": 1.1377,
+      "step": 5898
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.308539355289137e-06,
+      "loss": 1.1138,
+      "step": 5899
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.305468599963116e-06,
+      "loss": 1.0312,
+      "step": 5900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.302398009138586e-06,
+      "loss": 1.0361,
+      "step": 5901
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299327583113633e-06,
+      "loss": 1.0972,
+      "step": 5902
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.296257322186316e-06,
+      "loss": 1.0984,
+      "step": 5903
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2931872266547e-06,
+      "loss": 0.9863,
+      "step": 5904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.290117296816822e-06,
+      "loss": 1.0989,
+      "step": 5905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.287047532970702e-06,
+      "loss": 1.1716,
+      "step": 5906
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.283977935414346e-06,
+      "loss": 1.0275,
+      "step": 5907
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.280908504445744e-06,
+      "loss": 1.1179,
+      "step": 5908
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277839240362877e-06,
+      "loss": 1.0208,
+      "step": 5909
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274770143463692e-06,
+      "loss": 1.01,
+      "step": 5910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.271701214046132e-06,
+      "loss": 1.0111,
+      "step": 5911
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.268632452408127e-06,
+      "loss": 1.0323,
+      "step": 5912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.265563858847584e-06,
+      "loss": 1.2598,
+      "step": 5913
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262495433662396e-06,
+      "loss": 1.0505,
+      "step": 5914
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.259427177150434e-06,
+      "loss": 1.2119,
+      "step": 5915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25635908960957e-06,
+      "loss": 1.0374,
+      "step": 5916
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.253291171337635e-06,
+      "loss": 1.0898,
+      "step": 5917
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.250223422632456e-06,
+      "loss": 1.0918,
+      "step": 5918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247155843791852e-06,
+      "loss": 1.1367,
+      "step": 5919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.244088435113613e-06,
+      "loss": 1.1194,
+      "step": 5920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.241021196895516e-06,
+      "loss": 1.0508,
+      "step": 5921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.237954129435322e-06,
+      "loss": 1.0205,
+      "step": 5922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.234887233030772e-06,
+      "loss": 1.1165,
+      "step": 5923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.231820507979607e-06,
+      "loss": 0.9902,
+      "step": 5924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22875395457952e-06,
+      "loss": 1.1558,
+      "step": 5925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.225687573128216e-06,
+      "loss": 1.1433,
+      "step": 5926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.222621363923371e-06,
+      "loss": 1.1128,
+      "step": 5927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.219555327262647e-06,
+      "loss": 1.0494,
+      "step": 5928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.216489463443684e-06,
+      "loss": 1.0781,
+      "step": 5929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213423772764115e-06,
+      "loss": 1.0686,
+      "step": 5930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21035825552155e-06,
+      "loss": 1.1504,
+      "step": 5931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.207292912013578e-06,
+      "loss": 1.1453,
+      "step": 5932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20422774253778e-06,
+      "loss": 0.9481,
+      "step": 5933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.201162747391713e-06,
+      "loss": 1.1799,
+      "step": 5934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.198097926872925e-06,
+      "loss": 1.0876,
+      "step": 5935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.195033281278937e-06,
+      "loss": 1.1116,
+      "step": 5936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.191968810907264e-06,
+      "loss": 1.1213,
+      "step": 5937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.188904516055395e-06,
+      "loss": 1.1836,
+      "step": 5938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.185840397020805e-06,
+      "loss": 1.0398,
+      "step": 5939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.182776454100951e-06,
+      "loss": 1.113,
+      "step": 5940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.179712687593276e-06,
+      "loss": 1.157,
+      "step": 5941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1766490977952e-06,
+      "loss": 1.1194,
+      "step": 5942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.173585685004137e-06,
+      "loss": 1.0657,
+      "step": 5943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.170522449517472e-06,
+      "loss": 1.1421,
+      "step": 5944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.167459391632577e-06,
+      "loss": 1.1541,
+      "step": 5945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.164396511646811e-06,
+      "loss": 1.0522,
+      "step": 5946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161333809857508e-06,
+      "loss": 1.1665,
+      "step": 5947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.158271286561987e-06,
+      "loss": 1.1582,
+      "step": 5948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.155208942057555e-06,
+      "loss": 0.9901,
+      "step": 5949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.152146776641496e-06,
+      "loss": 1.0437,
+      "step": 5950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.149084790611079e-06,
+      "loss": 1.0549,
+      "step": 5951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.146022984263557e-06,
+      "loss": 1.1101,
+      "step": 5952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142961357896162e-06,
+      "loss": 1.123,
+      "step": 5953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.139899911806108e-06,
+      "loss": 1.0762,
+      "step": 5954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.136838646290596e-06,
+      "loss": 1.0884,
+      "step": 5955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.133777561646807e-06,
+      "loss": 1.1492,
+      "step": 5956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1307166581719e-06,
+      "loss": 1.0087,
+      "step": 5957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12765593616303e-06,
+      "loss": 1.0302,
+      "step": 5958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12459539591732e-06,
+      "loss": 0.9835,
+      "step": 5959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121535037731883e-06,
+      "loss": 1.0829,
+      "step": 5960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.118474861903807e-06,
+      "loss": 1.1096,
+      "step": 5961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11541486873017e-06,
+      "loss": 1.0781,
+      "step": 5962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.112355058508034e-06,
+      "loss": 1.198,
+      "step": 5963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.109295431534432e-06,
+      "loss": 1.0735,
+      "step": 5964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.106235988106387e-06,
+      "loss": 1.0757,
+      "step": 5965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103176728520908e-06,
+      "loss": 1.0131,
+      "step": 5966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10011765307498e-06,
+      "loss": 1.0277,
+      "step": 5967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.097058762065569e-06,
+      "loss": 1.0352,
+      "step": 5968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094000055789628e-06,
+      "loss": 1.0298,
+      "step": 5969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.090941534544087e-06,
+      "loss": 0.9861,
+      "step": 5970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.087883198625861e-06,
+      "loss": 0.9946,
+      "step": 5971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08482504833185e-06,
+      "loss": 1.0994,
+      "step": 5972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.081767083958926e-06,
+      "loss": 1.0991,
+      "step": 5973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.078709305803958e-06,
+      "loss": 1.1895,
+      "step": 5974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07565171416379e-06,
+      "loss": 0.9946,
+      "step": 5975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.072594309335233e-06,
+      "loss": 1.1389,
+      "step": 5976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.069537091615105e-06,
+      "loss": 1.1719,
+      "step": 5977
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.066480061300191e-06,
+      "loss": 1.0251,
+      "step": 5978
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.063423218687259e-06,
+      "loss": 1.0686,
+      "step": 5979
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.060366564073064e-06,
+      "loss": 1.053,
+      "step": 5980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.057310097754333e-06,
+      "loss": 1.2031,
+      "step": 5981
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.054253820027797e-06,
+      "loss": 0.9764,
+      "step": 5982
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051197731190133e-06,
+      "loss": 1.1165,
+      "step": 5983
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04814183153803e-06,
+      "loss": 1.1252,
+      "step": 5984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045086121368148e-06,
+      "loss": 0.9778,
+      "step": 5985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042030600977129e-06,
+      "loss": 1.0774,
+      "step": 5986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.038975270661593e-06,
+      "loss": 0.9856,
+      "step": 5987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.035920130718146e-06,
+      "loss": 1.1519,
+      "step": 5988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.032865181443376e-06,
+      "loss": 1.1418,
+      "step": 5989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.029810423133857e-06,
+      "loss": 1.1704,
+      "step": 5990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.026755856086123e-06,
+      "loss": 1.0483,
+      "step": 5991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.023701480596718e-06,
+      "loss": 1.0908,
+      "step": 5992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02064729696215e-06,
+      "loss": 1.1555,
+      "step": 5993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01759330547891e-06,
+      "loss": 1.1548,
+      "step": 5994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.014539506443477e-06,
+      "loss": 1.0471,
+      "step": 5995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.011485900152305e-06,
+      "loss": 0.9203,
+      "step": 5996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.008432486901837e-06,
+      "loss": 1.2131,
+      "step": 5997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005379266988482e-06,
+      "loss": 1.0493,
+      "step": 5998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002326240708645e-06,
+      "loss": 1.0271,
+      "step": 5999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999273408358708e-06,
+      "loss": 1.0774,
+      "step": 6000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.996220770235034e-06,
+      "loss": 1.0812,
+      "step": 6001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.993168326633967e-06,
+      "loss": 1.0818,
+      "step": 6002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.990116077851829e-06,
+      "loss": 1.0947,
+      "step": 6003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987064024184929e-06,
+      "loss": 1.1672,
+      "step": 6004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98401216592955e-06,
+      "loss": 1.095,
+      "step": 6005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.980960503381961e-06,
+      "loss": 1.2012,
+      "step": 6006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97790903683841e-06,
+      "loss": 1.0668,
+      "step": 6007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97485776659513e-06,
+      "loss": 1.1846,
+      "step": 6008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97180669294833e-06,
+      "loss": 0.9441,
+      "step": 6009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.968755816194203e-06,
+      "loss": 1.0249,
+      "step": 6010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.965705136628921e-06,
+      "loss": 0.8698,
+      "step": 6011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.962654654548639e-06,
+      "loss": 0.976,
+      "step": 6012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.959604370249485e-06,
+      "loss": 1.0046,
+      "step": 6013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95655428402758e-06,
+      "loss": 0.9976,
+      "step": 6014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.953504396179017e-06,
+      "loss": 0.9904,
+      "step": 6015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.950454706999875e-06,
+      "loss": 1.021,
+      "step": 6016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.94740521678621e-06,
+      "loss": 1.0693,
+      "step": 6017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.94435592583406e-06,
+      "loss": 1.0426,
+      "step": 6018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.941306834439447e-06,
+      "loss": 0.9531,
+      "step": 6019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.938257942898365e-06,
+      "loss": 1.0251,
+      "step": 6020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.935209251506794e-06,
+      "loss": 0.9955,
+      "step": 6021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932160760560696e-06,
+      "loss": 1.1252,
+      "step": 6022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.929112470356014e-06,
+      "loss": 1.0747,
+      "step": 6023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926064381188669e-06,
+      "loss": 1.0554,
+      "step": 6024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.923016493354561e-06,
+      "loss": 1.0532,
+      "step": 6025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.919968807149575e-06,
+      "loss": 1.1193,
+      "step": 6026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91692132286957e-06,
+      "loss": 1.0464,
+      "step": 6027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.913874040810392e-06,
+      "loss": 1.0774,
+      "step": 6028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.910826961267863e-06,
+      "loss": 1.0535,
+      "step": 6029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.907780084537787e-06,
+      "loss": 1.0747,
+      "step": 6030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.904733410915952e-06,
+      "loss": 1.1121,
+      "step": 6031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.901686940698118e-06,
+      "loss": 1.1421,
+      "step": 6032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.898640674180032e-06,
+      "loss": 1.0991,
+      "step": 6033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.895594611657422e-06,
+      "loss": 1.1111,
+      "step": 6034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.892548753425987e-06,
+      "loss": 1.1138,
+      "step": 6035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.889503099781416e-06,
+      "loss": 0.9572,
+      "step": 6036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.886457651019373e-06,
+      "loss": 1.0273,
+      "step": 6037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.883412407435503e-06,
+      "loss": 1.1462,
+      "step": 6038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.880367369325436e-06,
+      "loss": 1.0071,
+      "step": 6039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.877322536984775e-06,
+      "loss": 1.1548,
+      "step": 6040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.874277910709107e-06,
+      "loss": 1.1116,
+      "step": 6041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871233490793995e-06,
+      "loss": 1.016,
+      "step": 6042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868189277534987e-06,
+      "loss": 1.0747,
+      "step": 6043
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.865145271227609e-06,
+      "loss": 1.0525,
+      "step": 6044
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.862101472167364e-06,
+      "loss": 1.0952,
+      "step": 6045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.859057880649738e-06,
+      "loss": 1.0007,
+      "step": 6046
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8560144969702e-06,
+      "loss": 1.127,
+      "step": 6047
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.852971321424197e-06,
+      "loss": 0.9426,
+      "step": 6048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.849928354307145e-06,
+      "loss": 0.9862,
+      "step": 6049
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.846885595914455e-06,
+      "loss": 1.1467,
+      "step": 6050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84384304654151e-06,
+      "loss": 1.0291,
+      "step": 6051
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.840800706483674e-06,
+      "loss": 1.0818,
+      "step": 6052
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.837758576036289e-06,
+      "loss": 1.0957,
+      "step": 6053
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.834716655494681e-06,
+      "loss": 1.1096,
+      "step": 6054
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.831674945154156e-06,
+      "loss": 1.0137,
+      "step": 6055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.828633445309995e-06,
+      "loss": 1.0349,
+      "step": 6056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.825592156257455e-06,
+      "loss": 1.051,
+      "step": 6057
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822551078291783e-06,
+      "loss": 0.9681,
+      "step": 6058
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.819510211708198e-06,
+      "loss": 1.0312,
+      "step": 6059
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.816469556801904e-06,
+      "loss": 1.2151,
+      "step": 6060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.813429113868075e-06,
+      "loss": 1.1116,
+      "step": 6061
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.810388883201879e-06,
+      "loss": 1.0242,
+      "step": 6062
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.807348865098456e-06,
+      "loss": 1.1035,
+      "step": 6063
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80430905985291e-06,
+      "loss": 1.1826,
+      "step": 6064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.801269467760353e-06,
+      "loss": 1.1511,
+      "step": 6065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.798230089115855e-06,
+      "loss": 1.026,
+      "step": 6066
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.795190924214476e-06,
+      "loss": 1.0933,
+      "step": 6067
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.792151973351251e-06,
+      "loss": 1.0911,
+      "step": 6068
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789113236821191e-06,
+      "loss": 1.1121,
+      "step": 6069
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.786074714919296e-06,
+      "loss": 1.1395,
+      "step": 6070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78303640794054e-06,
+      "loss": 1.1724,
+      "step": 6071
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.779998316179867e-06,
+      "loss": 0.9662,
+      "step": 6072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.776960439932214e-06,
+      "loss": 1.0791,
+      "step": 6073
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.773922779492492e-06,
+      "loss": 1.0933,
+      "step": 6074
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.770885335155588e-06,
+      "loss": 1.0294,
+      "step": 6075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.767848107216373e-06,
+      "loss": 0.9968,
+      "step": 6076
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.764811095969693e-06,
+      "loss": 1.1177,
+      "step": 6077
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76177430171038e-06,
+      "loss": 1.0994,
+      "step": 6078
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75873772473323e-06,
+      "loss": 1.1067,
+      "step": 6079
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.755701365333029e-06,
+      "loss": 1.0918,
+      "step": 6080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.752665223804547e-06,
+      "loss": 1.0616,
+      "step": 6081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.749629300442522e-06,
+      "loss": 1.063,
+      "step": 6082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.746593595541676e-06,
+      "loss": 0.9576,
+      "step": 6083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.743558109396706e-06,
+      "loss": 0.9863,
+      "step": 6084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.740522842302299e-06,
+      "loss": 1.0891,
+      "step": 6085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.737487794553101e-06,
+      "loss": 1.0323,
+      "step": 6086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.734452966443751e-06,
+      "loss": 1.0513,
+      "step": 6087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73141835826887e-06,
+      "loss": 1.0109,
+      "step": 6088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.728383970323044e-06,
+      "loss": 0.9421,
+      "step": 6089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.725349802900848e-06,
+      "loss": 1.0531,
+      "step": 6090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.722315856296833e-06,
+      "loss": 1.1798,
+      "step": 6091
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.719282130805524e-06,
+      "loss": 1.1255,
+      "step": 6092
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71624862672144e-06,
+      "loss": 1.0574,
+      "step": 6093
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713215344339053e-06,
+      "loss": 1.0913,
+      "step": 6094
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.710182283952832e-06,
+      "loss": 1.0759,
+      "step": 6095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707149445857225e-06,
+      "loss": 1.2104,
+      "step": 6096
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.704116830346649e-06,
+      "loss": 1.0452,
+      "step": 6097
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.701084437715503e-06,
+      "loss": 1.0925,
+      "step": 6098
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.698052268258169e-06,
+      "loss": 1.1284,
+      "step": 6099
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695020322269004e-06,
+      "loss": 1.1709,
+      "step": 6100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.691988600042337e-06,
+      "loss": 1.177,
+      "step": 6101
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688957101872484e-06,
+      "loss": 1.0306,
+      "step": 6102
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.685925828053735e-06,
+      "loss": 1.21,
+      "step": 6103
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.682894778880365e-06,
+      "loss": 1.0464,
+      "step": 6104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.679863954646618e-06,
+      "loss": 1.208,
+      "step": 6105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67683335564672e-06,
+      "loss": 1.1313,
+      "step": 6106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.673802982174877e-06,
+      "loss": 1.0261,
+      "step": 6107
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670772834525265e-06,
+      "loss": 0.9587,
+      "step": 6108
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667742912992051e-06,
+      "loss": 1.1235,
+      "step": 6109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664713217869371e-06,
+      "loss": 1.0586,
+      "step": 6110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.661683749451338e-06,
+      "loss": 1.1936,
+      "step": 6111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.658654508032053e-06,
+      "loss": 1.0459,
+      "step": 6112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655625493905584e-06,
+      "loss": 1.0157,
+      "step": 6113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.652596707365982e-06,
+      "loss": 1.0857,
+      "step": 6114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.649568148707279e-06,
+      "loss": 1.106,
+      "step": 6115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.646539818223473e-06,
+      "loss": 1.0044,
+      "step": 6116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643511716208553e-06,
+      "loss": 0.9849,
+      "step": 6117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.640483842956478e-06,
+      "loss": 1.0012,
+      "step": 6118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637456198761191e-06,
+      "loss": 0.9561,
+      "step": 6119
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.634428783916608e-06,
+      "loss": 1.2666,
+      "step": 6120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.631401598716622e-06,
+      "loss": 1.0408,
+      "step": 6121
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.628374643455111e-06,
+      "loss": 1.0226,
+      "step": 6122
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.625347918425918e-06,
+      "loss": 1.1023,
+      "step": 6123
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.622321423922875e-06,
+      "loss": 1.0298,
+      "step": 6124
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.619295160239787e-06,
+      "loss": 1.0927,
+      "step": 6125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.616269127670435e-06,
+      "loss": 1.0159,
+      "step": 6126
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.613243326508584e-06,
+      "loss": 1.0247,
+      "step": 6127
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.61021775704797e-06,
+      "loss": 0.9688,
+      "step": 6128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607192419582312e-06,
+      "loss": 1.0339,
+      "step": 6129
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.604167314405298e-06,
+      "loss": 1.0925,
+      "step": 6130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.601142441810601e-06,
+      "loss": 1.0261,
+      "step": 6131
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598117802091871e-06,
+      "loss": 1.0437,
+      "step": 6132
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.595093395542731e-06,
+      "loss": 1.0139,
+      "step": 6133
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.592069222456781e-06,
+      "loss": 1.0659,
+      "step": 6134
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5890452831276095e-06,
+      "loss": 1.1538,
+      "step": 6135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.586021577848769e-06,
+      "loss": 0.9998,
+      "step": 6136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.582998106913797e-06,
+      "loss": 1.1145,
+      "step": 6137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.579974870616201e-06,
+      "loss": 0.9995,
+      "step": 6138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5769518692494735e-06,
+      "loss": 1.1123,
+      "step": 6139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.573929103107081e-06,
+      "loss": 1.01,
+      "step": 6140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.570906572482466e-06,
+      "loss": 0.9889,
+      "step": 6141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567884277669047e-06,
+      "loss": 1.1453,
+      "step": 6142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.564862218960228e-06,
+      "loss": 1.0498,
+      "step": 6143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.561840396649386e-06,
+      "loss": 1.0374,
+      "step": 6144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.55881881102986e-06,
+      "loss": 0.9529,
+      "step": 6145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.555797462394989e-06,
+      "loss": 1.1704,
+      "step": 6146
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.552776351038078e-06,
+      "loss": 1.0892,
+      "step": 6147
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.549755477252409e-06,
+      "loss": 1.1573,
+      "step": 6148
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.546734841331242e-06,
+      "loss": 1.1082,
+      "step": 6149
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.543714443567812e-06,
+      "loss": 1.1724,
+      "step": 6150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.540694284255341e-06,
+      "loss": 0.981,
+      "step": 6151
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5376743636870065e-06,
+      "loss": 1.0474,
+      "step": 6152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534654682155985e-06,
+      "loss": 1.137,
+      "step": 6153
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53163523995542e-06,
+      "loss": 1.1418,
+      "step": 6154
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.528616037378429e-06,
+      "loss": 1.2051,
+      "step": 6155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525597074718112e-06,
+      "loss": 1.0874,
+      "step": 6156
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.522578352267542e-06,
+      "loss": 1.1865,
+      "step": 6157
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.519559870319772e-06,
+      "loss": 1.0779,
+      "step": 6158
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.516541629167835e-06,
+      "loss": 0.9523,
+      "step": 6159
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513523629104722e-06,
+      "loss": 0.9957,
+      "step": 6160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510505870423423e-06,
+      "loss": 1.0054,
+      "step": 6161
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.507488353416896e-06,
+      "loss": 1.1475,
+      "step": 6162
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5044710783780725e-06,
+      "loss": 1.061,
+      "step": 6163
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.501454045599863e-06,
+      "loss": 1.2612,
+      "step": 6164
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.498437255375153e-06,
+      "loss": 1.0331,
+      "step": 6165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.495420707996816e-06,
+      "loss": 1.0894,
+      "step": 6166
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.492404403757679e-06,
+      "loss": 1.1553,
+      "step": 6167
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489388342950562e-06,
+      "loss": 1.2063,
+      "step": 6168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4863725258682615e-06,
+      "loss": 1.1101,
+      "step": 6169
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4833569528035445e-06,
+      "loss": 0.9735,
+      "step": 6170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.480341624049156e-06,
+      "loss": 1.1385,
+      "step": 6171
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.47732653989782e-06,
+      "loss": 1.0164,
+      "step": 6172
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.474311700642234e-06,
+      "loss": 1.0508,
+      "step": 6173
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.471297106575067e-06,
+      "loss": 1.0725,
+      "step": 6174
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.468282757988974e-06,
+      "loss": 1.0969,
+      "step": 6175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.465268655176578e-06,
+      "loss": 1.1946,
+      "step": 6176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.462254798430487e-06,
+      "loss": 1.0818,
+      "step": 6177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.459241188043277e-06,
+      "loss": 1.033,
+      "step": 6178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.456227824307503e-06,
+      "loss": 0.9656,
+      "step": 6179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.453214707515695e-06,
+      "loss": 1.0837,
+      "step": 6180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.450201837960363e-06,
+      "loss": 1.0964,
+      "step": 6181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.447189215933985e-06,
+      "loss": 1.2,
+      "step": 6182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444176841729022e-06,
+      "loss": 0.9932,
+      "step": 6183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.44116471563791e-06,
+      "loss": 1.1968,
+      "step": 6184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4381528379530596e-06,
+      "loss": 1.0444,
+      "step": 6185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.435141208966858e-06,
+      "loss": 1.0325,
+      "step": 6186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.432129828971666e-06,
+      "loss": 1.0579,
+      "step": 6187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.429118698259824e-06,
+      "loss": 1.0283,
+      "step": 6188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.426107817123644e-06,
+      "loss": 0.9807,
+      "step": 6189
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.423097185855416e-06,
+      "loss": 1.0708,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4200868047474036e-06,
+      "loss": 1.0348,
+      "step": 6191
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.417076674091854e-06,
+      "loss": 1.1333,
+      "step": 6192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.414066794180981e-06,
+      "loss": 1.0828,
+      "step": 6193
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.411057165306976e-06,
+      "loss": 1.0122,
+      "step": 6194
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408047787762013e-06,
+      "loss": 1.0454,
+      "step": 6195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.405038661838229e-06,
+      "loss": 1.2048,
+      "step": 6196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.402029787827746e-06,
+      "loss": 1.0454,
+      "step": 6197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3990211660226595e-06,
+      "loss": 1.1689,
+      "step": 6198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3960127967150375e-06,
+      "loss": 1.0078,
+      "step": 6199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.39300468019693e-06,
+      "loss": 1.0898,
+      "step": 6200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3899968167603585e-06,
+      "loss": 1.0641,
+      "step": 6201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386989206697317e-06,
+      "loss": 1.1289,
+      "step": 6202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.383981850299782e-06,
+      "loss": 1.0135,
+      "step": 6203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.380974747859697e-06,
+      "loss": 1.0547,
+      "step": 6204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.377967899668985e-06,
+      "loss": 1.209,
+      "step": 6205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3749613060195476e-06,
+      "loss": 1.0386,
+      "step": 6206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3719549672032546e-06,
+      "loss": 0.9929,
+      "step": 6207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.36894888351196e-06,
+      "loss": 1.0486,
+      "step": 6208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3659430552374836e-06,
+      "loss": 1.0688,
+      "step": 6209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362937482671628e-06,
+      "loss": 0.9529,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.359932166106165e-06,
+      "loss": 1.1335,
+      "step": 6211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356927105832846e-06,
+      "loss": 1.0863,
+      "step": 6212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.353922302143395e-06,
+      "loss": 1.1268,
+      "step": 6213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.350917755329512e-06,
+      "loss": 0.9357,
+      "step": 6214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34791346568287e-06,
+      "loss": 0.9741,
+      "step": 6215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.344909433495124e-06,
+      "loss": 1.0183,
+      "step": 6216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.341905659057895e-06,
+      "loss": 0.9977,
+      "step": 6217
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3389021426627886e-06,
+      "loss": 1.0269,
+      "step": 6218
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3358988846013715e-06,
+      "loss": 1.0881,
+      "step": 6219
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.332895885165198e-06,
+      "loss": 1.0066,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.329893144645792e-06,
+      "loss": 1.0681,
+      "step": 6221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3268906633346515e-06,
+      "loss": 1.0444,
+      "step": 6222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.323888441523255e-06,
+      "loss": 1.1553,
+      "step": 6223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.32088647950305e-06,
+      "loss": 1.0719,
+      "step": 6224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.317884777565462e-06,
+      "loss": 1.0881,
+      "step": 6225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.314883336001885e-06,
+      "loss": 1.031,
+      "step": 6226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.311882155103698e-06,
+      "loss": 1.0171,
+      "step": 6227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.308881235162246e-06,
+      "loss": 0.9718,
+      "step": 6228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.305880576468853e-06,
+      "loss": 1.1624,
+      "step": 6229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302880179314813e-06,
+      "loss": 1.0574,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2998800439914045e-06,
+      "loss": 1.0994,
+      "step": 6231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.296880170789875e-06,
+      "loss": 1.123,
+      "step": 6232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.293880560001435e-06,
+      "loss": 1.0809,
+      "step": 6233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29088121191729e-06,
+      "loss": 1.0232,
+      "step": 6234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.287882126828607e-06,
+      "loss": 1.053,
+      "step": 6235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2848833050265335e-06,
+      "loss": 1.0167,
+      "step": 6236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2818847468021834e-06,
+      "loss": 0.98,
+      "step": 6237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.278886452446654e-06,
+      "loss": 0.9697,
+      "step": 6238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.275888422251013e-06,
+      "loss": 1.0195,
+      "step": 6239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2728906565063074e-06,
+      "loss": 1.1624,
+      "step": 6240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.269893155503543e-06,
+      "loss": 0.9771,
+      "step": 6241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.266895919533719e-06,
+      "loss": 1.0746,
+      "step": 6242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.263898948887797e-06,
+      "loss": 1.1418,
+      "step": 6243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2609022438567205e-06,
+      "loss": 1.1775,
+      "step": 6244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.257905804731398e-06,
+      "loss": 0.9857,
+      "step": 6245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25490963180272e-06,
+      "loss": 1.1522,
+      "step": 6246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.251913725361553e-06,
+      "loss": 1.1499,
+      "step": 6247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.248918085698726e-06,
+      "loss": 1.127,
+      "step": 6248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24592271310505e-06,
+      "loss": 1.0771,
+      "step": 6249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.242927607871314e-06,
+      "loss": 1.0072,
+      "step": 6250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.239932770288274e-06,
+      "loss": 0.9768,
+      "step": 6251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236938200646661e-06,
+      "loss": 1.0232,
+      "step": 6252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2339438992371815e-06,
+      "loss": 1.1489,
+      "step": 6253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.230949866350524e-06,
+      "loss": 0.9775,
+      "step": 6254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.227956102277332e-06,
+      "loss": 1.0098,
+      "step": 6255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.224962607308234e-06,
+      "loss": 1.094,
+      "step": 6256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.22196938173384e-06,
+      "loss": 1.0374,
+      "step": 6257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.218976425844719e-06,
+      "loss": 1.099,
+      "step": 6258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.215983739931426e-06,
+      "loss": 1.0056,
+      "step": 6259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.212991324284482e-06,
+      "loss": 1.2014,
+      "step": 6260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.209999179194381e-06,
+      "loss": 1.2014,
+      "step": 6261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.207007304951604e-06,
+      "loss": 1.1133,
+      "step": 6262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204015701846585e-06,
+      "loss": 1.0975,
+      "step": 6263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.201024370169745e-06,
+      "loss": 1.039,
+      "step": 6264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.19803331021148e-06,
+      "loss": 0.9624,
+      "step": 6265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.195042522262152e-06,
+      "loss": 1.1582,
+      "step": 6266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.192052006612102e-06,
+      "loss": 1.0981,
+      "step": 6267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.189061763551644e-06,
+      "loss": 1.063,
+      "step": 6268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186071793371066e-06,
+      "loss": 1.1226,
+      "step": 6269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.183082096360621e-06,
+      "loss": 1.168,
+      "step": 6270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.180092672810546e-06,
+      "loss": 1.0591,
+      "step": 6271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.177103523011047e-06,
+      "loss": 0.9135,
+      "step": 6272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.174114647252306e-06,
+      "loss": 1.0774,
+      "step": 6273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.171126045824477e-06,
+      "loss": 0.9811,
+      "step": 6274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.168137719017687e-06,
+      "loss": 1.1021,
+      "step": 6275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.165149667122036e-06,
+      "loss": 1.1436,
+      "step": 6276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.162161890427597e-06,
+      "loss": 1.1133,
+      "step": 6277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.159174389224417e-06,
+      "loss": 1.0339,
+      "step": 6278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.156187163802515e-06,
+      "loss": 0.9769,
+      "step": 6279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.153200214451885e-06,
+      "loss": 1.0022,
+      "step": 6280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1502135414624966e-06,
+      "loss": 1.1096,
+      "step": 6281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.147227145124287e-06,
+      "loss": 0.9929,
+      "step": 6282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1442410257271695e-06,
+      "loss": 1.1083,
+      "step": 6283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.141255183561033e-06,
+      "loss": 1.0293,
+      "step": 6284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.138269618915732e-06,
+      "loss": 1.0806,
+      "step": 6285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1352843320811005e-06,
+      "loss": 1.157,
+      "step": 6286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.132299323346943e-06,
+      "loss": 1.0098,
+      "step": 6287
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1293145930030395e-06,
+      "loss": 1.0063,
+      "step": 6288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.126330141339142e-06,
+      "loss": 1.0854,
+      "step": 6289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.123345968644972e-06,
+      "loss": 1.088,
+      "step": 6290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.120362075210232e-06,
+      "loss": 1.0804,
+      "step": 6291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.117378461324584e-06,
+      "loss": 1.2141,
+      "step": 6292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.114395127277676e-06,
+      "loss": 1.1277,
+      "step": 6293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.111412073359121e-06,
+      "loss": 1.0222,
+      "step": 6294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.108429299858508e-06,
+      "loss": 1.042,
+      "step": 6295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.105446807065402e-06,
+      "loss": 1.1096,
+      "step": 6296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1024645952693325e-06,
+      "loss": 1.2825,
+      "step": 6297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.099482664759812e-06,
+      "loss": 1.0664,
+      "step": 6298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0965010158263125e-06,
+      "loss": 1.0522,
+      "step": 6299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.09351964875829e-06,
+      "loss": 1.1206,
+      "step": 6300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.090538563845168e-06,
+      "loss": 1.106,
+      "step": 6301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.087557761376347e-06,
+      "loss": 1.1714,
+      "step": 6302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.084577241641191e-06,
+      "loss": 1.0759,
+      "step": 6303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.081597004929048e-06,
+      "loss": 1.0479,
+      "step": 6304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078617051529232e-06,
+      "loss": 1.2302,
+      "step": 6305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0756373817310306e-06,
+      "loss": 1.1384,
+      "step": 6306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.072657995823702e-06,
+      "loss": 1.0387,
+      "step": 6307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.069678894096479e-06,
+      "loss": 1.1292,
+      "step": 6308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.066700076838566e-06,
+      "loss": 1.0541,
+      "step": 6309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.063721544339142e-06,
+      "loss": 0.9625,
+      "step": 6310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.060743296887352e-06,
+      "loss": 1.186,
+      "step": 6311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.057765334772325e-06,
+      "loss": 1.1165,
+      "step": 6312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.054787658283155e-06,
+      "loss": 0.9534,
+      "step": 6313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0518102677089006e-06,
+      "loss": 1.0796,
+      "step": 6314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.048833163338607e-06,
+      "loss": 1.1118,
+      "step": 6315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0458563454612815e-06,
+      "loss": 1.0752,
+      "step": 6316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0428798143659095e-06,
+      "loss": 1.0886,
+      "step": 6317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.039903570341447e-06,
+      "loss": 0.9954,
+      "step": 6318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.036927613676816e-06,
+      "loss": 1.0513,
+      "step": 6319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.033951944660928e-06,
+      "loss": 1.0278,
+      "step": 6320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.03097656358264e-06,
+      "loss": 1.0337,
+      "step": 6321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.028001470730804e-06,
+      "loss": 1.187,
+      "step": 6322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.025026666394235e-06,
+      "loss": 0.911,
+      "step": 6323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022052150861719e-06,
+      "loss": 1.1205,
+      "step": 6324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0190779244220175e-06,
+      "loss": 1.0146,
+      "step": 6325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.016103987363857e-06,
+      "loss": 1.0325,
+      "step": 6326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013130339975948e-06,
+      "loss": 0.9875,
+      "step": 6327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.010156982546966e-06,
+      "loss": 1.0306,
+      "step": 6328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0071839153655496e-06,
+      "loss": 1.0005,
+      "step": 6329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.004211138720325e-06,
+      "loss": 1.0997,
+      "step": 6330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00123865289988e-06,
+      "loss": 1.001,
+      "step": 6331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.998266458192778e-06,
+      "loss": 0.9335,
+      "step": 6332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.995294554887555e-06,
+      "loss": 1.1128,
+      "step": 6333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.992322943272712e-06,
+      "loss": 1.1262,
+      "step": 6334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9893516236367374e-06,
+      "loss": 0.9895,
+      "step": 6335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.98638059626807e-06,
+      "loss": 0.9657,
+      "step": 6336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.98340986145513e-06,
+      "loss": 1.0952,
+      "step": 6337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.980439419486318e-06,
+      "loss": 1.1001,
+      "step": 6338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.977469270649995e-06,
+      "loss": 1.0532,
+      "step": 6339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9744994152344934e-06,
+      "loss": 0.9814,
+      "step": 6340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971529853528125e-06,
+      "loss": 1.1091,
+      "step": 6341
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.968560585819168e-06,
+      "loss": 1.1257,
+      "step": 6342
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.965591612395869e-06,
+      "loss": 1.0957,
+      "step": 6343
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.962622933546452e-06,
+      "loss": 1.156,
+      "step": 6344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959654549559108e-06,
+      "loss": 1.0325,
+      "step": 6345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.956686460722004e-06,
+      "loss": 1.0723,
+      "step": 6346
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9537186673232744e-06,
+      "loss": 1.0267,
+      "step": 6347
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9507511696510285e-06,
+      "loss": 1.2749,
+      "step": 6348
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.947783967993342e-06,
+      "loss": 1.1238,
+      "step": 6349
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9448170626382675e-06,
+      "loss": 1.0065,
+      "step": 6350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.941850453873822e-06,
+      "loss": 0.988,
+      "step": 6351
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.938884141987996e-06,
+      "loss": 1.2178,
+      "step": 6352
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9359181272687595e-06,
+      "loss": 1.0054,
+      "step": 6353
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9329524100040415e-06,
+      "loss": 1.1191,
+      "step": 6354
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929986990481749e-06,
+      "loss": 1.0547,
+      "step": 6355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.927021868989761e-06,
+      "loss": 1.2253,
+      "step": 6356
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9240570458159225e-06,
+      "loss": 1.1047,
+      "step": 6357
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.921092521248052e-06,
+      "loss": 1.0183,
+      "step": 6358
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.918128295573941e-06,
+      "loss": 0.9823,
+      "step": 6359
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.915164369081344e-06,
+      "loss": 1.1069,
+      "step": 6360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9122007420580014e-06,
+      "loss": 0.997,
+      "step": 6361
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909237414791614e-06,
+      "loss": 1.0186,
+      "step": 6362
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906274387569851e-06,
+      "loss": 1.0276,
+      "step": 6363
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.90331166068036e-06,
+      "loss": 1.084,
+      "step": 6364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.900349234410757e-06,
+      "loss": 1.0305,
+      "step": 6365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.897387109048625e-06,
+      "loss": 1.1191,
+      "step": 6366
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894425284881521e-06,
+      "loss": 1.1351,
+      "step": 6367
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.891463762196973e-06,
+      "loss": 1.019,
+      "step": 6368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88850254128248e-06,
+      "loss": 1.1445,
+      "step": 6369
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.885541622425514e-06,
+      "loss": 1.0493,
+      "step": 6370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.882581005913509e-06,
+      "loss": 1.0443,
+      "step": 6371
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879620692033882e-06,
+      "loss": 1.0183,
+      "step": 6372
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.876660681074007e-06,
+      "loss": 1.1179,
+      "step": 6373
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.873700973321238e-06,
+      "loss": 1.0188,
+      "step": 6374
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.870741569062898e-06,
+      "loss": 0.9949,
+      "step": 6375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.867782468586277e-06,
+      "loss": 1.0452,
+      "step": 6376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.864823672178642e-06,
+      "loss": 1.1545,
+      "step": 6377
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.861865180127226e-06,
+      "loss": 1.0657,
+      "step": 6378
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.858906992719234e-06,
+      "loss": 1.1118,
+      "step": 6379
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8559491102418376e-06,
+      "loss": 1.22,
+      "step": 6380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.852991532982181e-06,
+      "loss": 1.0825,
+      "step": 6381
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.850034261227381e-06,
+      "loss": 1.0869,
+      "step": 6382
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.847077295264524e-06,
+      "loss": 0.8716,
+      "step": 6383
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.844120635380664e-06,
+      "loss": 1.0869,
+      "step": 6384
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.84116428186283e-06,
+      "loss": 1.1355,
+      "step": 6385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.838208234998016e-06,
+      "loss": 1.1431,
+      "step": 6386
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.835252495073194e-06,
+      "loss": 1.085,
+      "step": 6387
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.832297062375293e-06,
+      "loss": 1.0647,
+      "step": 6388
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829341937191225e-06,
+      "loss": 1.1636,
+      "step": 6389
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.826387119807864e-06,
+      "loss": 0.9513,
+      "step": 6390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8234326105120595e-06,
+      "loss": 1.1924,
+      "step": 6391
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.820478409590631e-06,
+      "loss": 1.1877,
+      "step": 6392
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.817524517330363e-06,
+      "loss": 1.0469,
+      "step": 6393
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.814570934018015e-06,
+      "loss": 1.1079,
+      "step": 6394
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811617659940312e-06,
+      "loss": 1.0808,
+      "step": 6395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808664695383953e-06,
+      "loss": 1.0592,
+      "step": 6396
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.805712040635606e-06,
+      "loss": 0.9019,
+      "step": 6397
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.802759695981909e-06,
+      "loss": 1.1228,
+      "step": 6398
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799807661709463e-06,
+      "loss": 1.1147,
+      "step": 6399
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.796855938104855e-06,
+      "loss": 0.9547,
+      "step": 6400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.793904525454631e-06,
+      "loss": 1.032,
+      "step": 6401
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.790953424045297e-06,
+      "loss": 1.0096,
+      "step": 6402
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7880026341633495e-06,
+      "loss": 1.1218,
+      "step": 6403
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.785052156095242e-06,
+      "loss": 1.0371,
+      "step": 6404
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782101990127401e-06,
+      "loss": 1.0056,
+      "step": 6405
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.779152136546223e-06,
+      "loss": 1.0801,
+      "step": 6406
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.77620259563807e-06,
+      "loss": 1.0513,
+      "step": 6407
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.773253367689281e-06,
+      "loss": 1.0786,
+      "step": 6408
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.770304452986164e-06,
+      "loss": 1.1008,
+      "step": 6409
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767355851814983e-06,
+      "loss": 1.0544,
+      "step": 6410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.76440756446199e-06,
+      "loss": 1.0691,
+      "step": 6411
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.761459591213395e-06,
+      "loss": 1.036,
+      "step": 6412
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758511932355383e-06,
+      "loss": 1.1152,
+      "step": 6413
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.755564588174106e-06,
+      "loss": 0.9852,
+      "step": 6414
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.752617558955682e-06,
+      "loss": 1.2292,
+      "step": 6415
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.749670844986214e-06,
+      "loss": 1.1301,
+      "step": 6416
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.746724446551748e-06,
+      "loss": 1.0458,
+      "step": 6417
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.743778363938318e-06,
+      "loss": 1.0052,
+      "step": 6418
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.74083259743193e-06,
+      "loss": 1.0568,
+      "step": 6419
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.737887147318548e-06,
+      "loss": 1.0535,
+      "step": 6420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7349420138841105e-06,
+      "loss": 1.0295,
+      "step": 6421
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.731997197414522e-06,
+      "loss": 1.0764,
+      "step": 6422
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729052698195669e-06,
+      "loss": 1.1143,
+      "step": 6423
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.726108516513386e-06,
+      "loss": 1.0524,
+      "step": 6424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723164652653489e-06,
+      "loss": 1.0081,
+      "step": 6425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.720221106901768e-06,
+      "loss": 1.015,
+      "step": 6426
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7172778795439704e-06,
+      "loss": 1.0693,
+      "step": 6427
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.714334970865823e-06,
+      "loss": 1.1772,
+      "step": 6428
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7113923811530165e-06,
+      "loss": 0.9609,
+      "step": 6429
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.708450110691206e-06,
+      "loss": 1.0408,
+      "step": 6430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.705508159766032e-06,
+      "loss": 1.1084,
+      "step": 6431
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.70256652866308e-06,
+      "loss": 0.99,
+      "step": 6432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699625217667922e-06,
+      "loss": 1.1582,
+      "step": 6433
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.696684227066096e-06,
+      "loss": 1.2095,
+      "step": 6434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6937435571431065e-06,
+      "loss": 0.9758,
+      "step": 6435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.690803208184426e-06,
+      "loss": 0.9452,
+      "step": 6436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6878631804755e-06,
+      "loss": 1.0204,
+      "step": 6437
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.684923474301739e-06,
+      "loss": 1.1418,
+      "step": 6438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.681984089948521e-06,
+      "loss": 1.083,
+      "step": 6439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.679045027701198e-06,
+      "loss": 1.074,
+      "step": 6440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.676106287845083e-06,
+      "loss": 1.1799,
+      "step": 6441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.67316787066547e-06,
+      "loss": 1.0334,
+      "step": 6442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.67022977644761e-06,
+      "loss": 1.0754,
+      "step": 6443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.667292005476729e-06,
+      "loss": 1.0798,
+      "step": 6444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664354558038017e-06,
+      "loss": 1.0952,
+      "step": 6445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6614174344166375e-06,
+      "loss": 1.0813,
+      "step": 6446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.658480634897717e-06,
+      "loss": 1.0403,
+      "step": 6447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.655544159766356e-06,
+      "loss": 1.0867,
+      "step": 6448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652608009307618e-06,
+      "loss": 0.9739,
+      "step": 6449
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6496721838065435e-06,
+      "loss": 1.024,
+      "step": 6450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.646736683548134e-06,
+      "loss": 1.093,
+      "step": 6451
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.643801508817362e-06,
+      "loss": 1.0693,
+      "step": 6452
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.640866659899169e-06,
+      "loss": 1.1619,
+      "step": 6453
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.63793213707846e-06,
+      "loss": 0.8839,
+      "step": 6454
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6349979406401135e-06,
+      "loss": 1.0217,
+      "step": 6455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632064070868975e-06,
+      "loss": 1.1194,
+      "step": 6456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.629130528049861e-06,
+      "loss": 1.0193,
+      "step": 6457
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.62619731246755e-06,
+      "loss": 1.0793,
+      "step": 6458
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.623264424406796e-06,
+      "loss": 1.1848,
+      "step": 6459
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.620331864152318e-06,
+      "loss": 1.032,
+      "step": 6460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.617399631988797e-06,
+      "loss": 1.0918,
+      "step": 6461
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.614467728200893e-06,
+      "loss": 1.1147,
+      "step": 6462
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611536153073225e-06,
+      "loss": 1.0461,
+      "step": 6463
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.608604906890384e-06,
+      "loss": 1.1147,
+      "step": 6464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.605673989936934e-06,
+      "loss": 1.1312,
+      "step": 6465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.602743402497399e-06,
+      "loss": 1.0142,
+      "step": 6466
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.599813144856278e-06,
+      "loss": 1.0212,
+      "step": 6467
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596883217298027e-06,
+      "loss": 1.0769,
+      "step": 6468
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593953620107081e-06,
+      "loss": 1.0112,
+      "step": 6469
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.591024353567839e-06,
+      "loss": 0.9856,
+      "step": 6470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.588095417964668e-06,
+      "loss": 1.1306,
+      "step": 6471
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5851668135819005e-06,
+      "loss": 1.0608,
+      "step": 6472
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.582238540703843e-06,
+      "loss": 1.0159,
+      "step": 6473
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.579310599614764e-06,
+      "loss": 1.0972,
+      "step": 6474
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.576382990598904e-06,
+      "loss": 1.145,
+      "step": 6475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5734557139404645e-06,
+      "loss": 0.9723,
+      "step": 6476
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.570528769923623e-06,
+      "loss": 1.0325,
+      "step": 6477
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.567602158832519e-06,
+      "loss": 1.0538,
+      "step": 6478
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564675880951263e-06,
+      "loss": 1.0693,
+      "step": 6479
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561749936563928e-06,
+      "loss": 0.8887,
+      "step": 6480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5588243259545645e-06,
+      "loss": 1.0923,
+      "step": 6481
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.555899049407185e-06,
+      "loss": 1.002,
+      "step": 6482
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.55297410720576e-06,
+      "loss": 1.0637,
+      "step": 6483
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.550049499634245e-06,
+      "loss": 0.9835,
+      "step": 6484
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.547125226976552e-06,
+      "loss": 1.0291,
+      "step": 6485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544201289516562e-06,
+      "loss": 1.0664,
+      "step": 6486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5412776875381255e-06,
+      "loss": 1.0923,
+      "step": 6487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.53835442132506e-06,
+      "loss": 1.0481,
+      "step": 6488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.535431491161152e-06,
+      "loss": 1.0471,
+      "step": 6489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532508897330149e-06,
+      "loss": 1.0818,
+      "step": 6490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.529586640115772e-06,
+      "loss": 1.0613,
+      "step": 6491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526664719801708e-06,
+      "loss": 1.1086,
+      "step": 6492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523743136671611e-06,
+      "loss": 1.0862,
+      "step": 6493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.520821891009102e-06,
+      "loss": 1.0952,
+      "step": 6494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.517900983097764e-06,
+      "loss": 1.1296,
+      "step": 6495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.514980413221162e-06,
+      "loss": 1.0999,
+      "step": 6496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.512060181662818e-06,
+      "loss": 1.1848,
+      "step": 6497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.509140288706211e-06,
+      "loss": 0.9482,
+      "step": 6498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506220734634807e-06,
+      "loss": 1.104,
+      "step": 6499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503301519732027e-06,
+      "loss": 1.1008,
+      "step": 6500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.500382644281265e-06,
+      "loss": 1.0248,
+      "step": 6501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.497464108565874e-06,
+      "loss": 1.1577,
+      "step": 6502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.494545912869184e-06,
+      "loss": 1.0432,
+      "step": 6503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.491628057474489e-06,
+      "loss": 1.0669,
+      "step": 6504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.488710542665042e-06,
+      "loss": 0.9668,
+      "step": 6505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4857933687240695e-06,
+      "loss": 1.0745,
+      "step": 6506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.482876535934768e-06,
+      "loss": 0.9403,
+      "step": 6507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479960044580298e-06,
+      "loss": 1.001,
+      "step": 6508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.477043894943783e-06,
+      "loss": 1.0938,
+      "step": 6509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.474128087308319e-06,
+      "loss": 1.042,
+      "step": 6510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.471212621956967e-06,
+      "loss": 1.1208,
+      "step": 6511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4682974991727506e-06,
+      "loss": 1.0947,
+      "step": 6512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.465382719238666e-06,
+      "loss": 1.1001,
+      "step": 6513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4624682824376716e-06,
+      "loss": 1.0791,
+      "step": 6514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4595541890526995e-06,
+      "loss": 1.1956,
+      "step": 6515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.456640439366639e-06,
+      "loss": 0.9985,
+      "step": 6516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453727033662353e-06,
+      "loss": 0.9691,
+      "step": 6517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.450813972222669e-06,
+      "loss": 1.0387,
+      "step": 6518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447901255330381e-06,
+      "loss": 0.9691,
+      "step": 6519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.444988883268246e-06,
+      "loss": 1.0262,
+      "step": 6520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.442076856318993e-06,
+      "loss": 1.0215,
+      "step": 6521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439165174765315e-06,
+      "loss": 1.0515,
+      "step": 6522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.436253838889874e-06,
+      "loss": 1.058,
+      "step": 6523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4333428489752935e-06,
+      "loss": 1.043,
+      "step": 6524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.430432205304168e-06,
+      "loss": 1.0237,
+      "step": 6525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.427521908159058e-06,
+      "loss": 0.9353,
+      "step": 6526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.424611957822484e-06,
+      "loss": 1.0719,
+      "step": 6527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.421702354576941e-06,
+      "loss": 1.1692,
+      "step": 6528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.418793098704885e-06,
+      "loss": 1.0558,
+      "step": 6529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4158841904887426e-06,
+      "loss": 1.2039,
+      "step": 6530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.412975630210905e-06,
+      "loss": 1.0641,
+      "step": 6531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.410067418153727e-06,
+      "loss": 1.1235,
+      "step": 6532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.407159554599532e-06,
+      "loss": 1.0752,
+      "step": 6533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4042520398306115e-06,
+      "loss": 1.2017,
+      "step": 6534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.401344874129218e-06,
+      "loss": 1.0173,
+      "step": 6535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.398438057777574e-06,
+      "loss": 1.1301,
+      "step": 6536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.395531591057863e-06,
+      "loss": 1.0991,
+      "step": 6537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.392625474252245e-06,
+      "loss": 1.0457,
+      "step": 6538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.389719707642838e-06,
+      "loss": 1.0913,
+      "step": 6539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386814291511726e-06,
+      "loss": 1.0916,
+      "step": 6540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.383909226140963e-06,
+      "loss": 1.1086,
+      "step": 6541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.381004511812563e-06,
+      "loss": 1.0496,
+      "step": 6542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37810014880851e-06,
+      "loss": 0.8806,
+      "step": 6543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375196137410755e-06,
+      "loss": 1.027,
+      "step": 6544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.372292477901212e-06,
+      "loss": 1.1284,
+      "step": 6545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.369389170561764e-06,
+      "loss": 1.1477,
+      "step": 6546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3664862156742565e-06,
+      "loss": 0.9924,
+      "step": 6547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.363583613520505e-06,
+      "loss": 1.0542,
+      "step": 6548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.360681364382282e-06,
+      "loss": 0.9421,
+      "step": 6549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357779468541336e-06,
+      "loss": 1.1111,
+      "step": 6550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.354877926279375e-06,
+      "loss": 1.1194,
+      "step": 6551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.351976737878076e-06,
+      "loss": 1.1145,
+      "step": 6552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.349075903619077e-06,
+      "loss": 1.0701,
+      "step": 6553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.346175423783989e-06,
+      "loss": 1.0615,
+      "step": 6554
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.343275298654384e-06,
+      "loss": 1.0228,
+      "step": 6555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3403755285118e-06,
+      "loss": 1.0938,
+      "step": 6556
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.337476113637735e-06,
+      "loss": 0.9335,
+      "step": 6557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.334577054313664e-06,
+      "loss": 1.041,
+      "step": 6558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.33167835082102e-06,
+      "loss": 1.0962,
+      "step": 6559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3287800034412e-06,
+      "loss": 0.9924,
+      "step": 6560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.325882012455573e-06,
+      "loss": 0.9501,
+      "step": 6561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.322984378145468e-06,
+      "loss": 1.0211,
+      "step": 6562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.320087100792185e-06,
+      "loss": 1.0895,
+      "step": 6563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.31719018067698e-06,
+      "loss": 1.0535,
+      "step": 6564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.314293618081082e-06,
+      "loss": 1.0986,
+      "step": 6565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311397413285684e-06,
+      "loss": 0.9651,
+      "step": 6566
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.308501566571942e-06,
+      "loss": 1.0253,
+      "step": 6567
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.305606078220976e-06,
+      "loss": 1.1379,
+      "step": 6568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.302710948513881e-06,
+      "loss": 1.0701,
+      "step": 6569
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.299816177731708e-06,
+      "loss": 1.1208,
+      "step": 6570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.296921766155468e-06,
+      "loss": 1.145,
+      "step": 6571
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.294027714066152e-06,
+      "loss": 1.0942,
+      "step": 6572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.291134021744706e-06,
+      "loss": 1.1138,
+      "step": 6573
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.288240689472043e-06,
+      "loss": 1.1222,
+      "step": 6574
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.28534771752904e-06,
+      "loss": 1.1143,
+      "step": 6575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.282455106196543e-06,
+      "loss": 0.9679,
+      "step": 6576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2795628557553616e-06,
+      "loss": 1.0566,
+      "step": 6577
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.276670966486269e-06,
+      "loss": 1.0354,
+      "step": 6578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273779438669998e-06,
+      "loss": 1.1135,
+      "step": 6579
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.270888272587259e-06,
+      "loss": 1.0222,
+      "step": 6580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267997468518716e-06,
+      "loss": 1.0242,
+      "step": 6581
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.265107026745003e-06,
+      "loss": 0.9722,
+      "step": 6582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26221694754672e-06,
+      "loss": 1.0505,
+      "step": 6583
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.259327231204425e-06,
+      "loss": 0.9868,
+      "step": 6584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.256437877998654e-06,
+      "loss": 1.0214,
+      "step": 6585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25354888820989e-06,
+      "loss": 1.1172,
+      "step": 6586
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.250660262118593e-06,
+      "loss": 0.9944,
+      "step": 6587
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247772000005188e-06,
+      "loss": 1.1072,
+      "step": 6588
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.244884102150057e-06,
+      "loss": 0.9658,
+      "step": 6589
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.241996568833552e-06,
+      "loss": 1.1042,
+      "step": 6590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.239109400335989e-06,
+      "loss": 0.9325,
+      "step": 6591
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2362225969376535e-06,
+      "loss": 1.0894,
+      "step": 6592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.233336158918781e-06,
+      "loss": 0.9542,
+      "step": 6593
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230450086559583e-06,
+      "loss": 1.1543,
+      "step": 6594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.227564380140238e-06,
+      "loss": 1.0378,
+      "step": 6595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22467903994088e-06,
+      "loss": 1.1316,
+      "step": 6596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221794066241613e-06,
+      "loss": 1.0767,
+      "step": 6597
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.218909459322504e-06,
+      "loss": 1.0754,
+      "step": 6598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2160252194635815e-06,
+      "loss": 1.0151,
+      "step": 6599
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.213141346944853e-06,
+      "loss": 0.9982,
+      "step": 6600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.210257842046263e-06,
+      "loss": 1.1252,
+      "step": 6601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.207374705047744e-06,
+      "loss": 1.0297,
+      "step": 6602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204491936229184e-06,
+      "loss": 1.1255,
+      "step": 6603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.201609535870436e-06,
+      "loss": 1.0298,
+      "step": 6604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.198727504251316e-06,
+      "loss": 0.9844,
+      "step": 6605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.195845841651609e-06,
+      "loss": 1.015,
+      "step": 6606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.192964548351058e-06,
+      "loss": 1.1165,
+      "step": 6607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.190083624629373e-06,
+      "loss": 1.127,
+      "step": 6608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.187203070766225e-06,
+      "loss": 1.166,
+      "step": 6609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1843228870412555e-06,
+      "loss": 1.1294,
+      "step": 6610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.181443073734067e-06,
+      "loss": 1.156,
+      "step": 6611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1785636311242245e-06,
+      "loss": 1.208,
+      "step": 6612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.175684559491259e-06,
+      "loss": 1.0386,
+      "step": 6613
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.172805859114665e-06,
+      "loss": 1.1899,
+      "step": 6614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.169927530273898e-06,
+      "loss": 1.1155,
+      "step": 6615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.167049573248381e-06,
+      "loss": 1.1338,
+      "step": 6616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164171988317501e-06,
+      "loss": 1.0205,
+      "step": 6617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.161294775760604e-06,
+      "loss": 1.0208,
+      "step": 6618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.158417935857011e-06,
+      "loss": 1.0441,
+      "step": 6619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.155541468885994e-06,
+      "loss": 1.1489,
+      "step": 6620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.152665375126795e-06,
+      "loss": 1.0662,
+      "step": 6621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.149789654858622e-06,
+      "loss": 1.0166,
+      "step": 6622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.14691430836064e-06,
+      "loss": 1.0657,
+      "step": 6623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144039335911982e-06,
+      "loss": 1.0798,
+      "step": 6624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141164737791744e-06,
+      "loss": 1.2234,
+      "step": 6625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.138290514278988e-06,
+      "loss": 1.033,
+      "step": 6626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.135416665652737e-06,
+      "loss": 1.1526,
+      "step": 6627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.132543192191978e-06,
+      "loss": 1.0521,
+      "step": 6628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.129670094175661e-06,
+      "loss": 1.0828,
+      "step": 6629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1267973718827e-06,
+      "loss": 1.0085,
+      "step": 6630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.123925025591972e-06,
+      "loss": 1.0779,
+      "step": 6631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12105305558232e-06,
+      "loss": 1.0002,
+      "step": 6632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.118181462132546e-06,
+      "loss": 0.9542,
+      "step": 6633
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.115310245521423e-06,
+      "loss": 1.0667,
+      "step": 6634
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1124394060276795e-06,
+      "loss": 1.0693,
+      "step": 6635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.109568943930011e-06,
+      "loss": 1.0691,
+      "step": 6636
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.106698859507075e-06,
+      "loss": 1.156,
+      "step": 6637
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.103829153037494e-06,
+      "loss": 1.0709,
+      "step": 6638
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.100959824799853e-06,
+      "loss": 1.1555,
+      "step": 6639
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.098090875072698e-06,
+      "loss": 1.1379,
+      "step": 6640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.095222304134541e-06,
+      "loss": 1.001,
+      "step": 6641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.092354112263861e-06,
+      "loss": 1.1042,
+      "step": 6642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.089486299739093e-06,
+      "loss": 1.1604,
+      "step": 6643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0866188668386405e-06,
+      "loss": 0.9678,
+      "step": 6644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.083751813840864e-06,
+      "loss": 1.0697,
+      "step": 6645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0808851410240895e-06,
+      "loss": 1.0209,
+      "step": 6646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.078018848666614e-06,
+      "loss": 1.0488,
+      "step": 6647
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.075152937046685e-06,
+      "loss": 1.0261,
+      "step": 6648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0722874064425185e-06,
+      "loss": 1.0696,
+      "step": 6649
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.069422257132301e-06,
+      "loss": 1.1279,
+      "step": 6650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.066557489394176e-06,
+      "loss": 1.1084,
+      "step": 6651
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.063693103506236e-06,
+      "loss": 1.0371,
+      "step": 6652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.06082909974656e-06,
+      "loss": 1.0193,
+      "step": 6653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.057965478393177e-06,
+      "loss": 1.1648,
+      "step": 6654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.055102239724082e-06,
+      "loss": 0.9342,
+      "step": 6655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.052239384017231e-06,
+      "loss": 1.0459,
+      "step": 6656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.049376911550542e-06,
+      "loss": 1.0762,
+      "step": 6657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.046514822601905e-06,
+      "loss": 1.0667,
+      "step": 6658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0436531174491565e-06,
+      "loss": 1.0735,
+      "step": 6659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040791796370109e-06,
+      "loss": 0.9807,
+      "step": 6660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.037930859642533e-06,
+      "loss": 1.0508,
+      "step": 6661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0350703075441645e-06,
+      "loss": 1.1399,
+      "step": 6662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032210140352695e-06,
+      "loss": 1.1177,
+      "step": 6663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.029350358345785e-06,
+      "loss": 1.1028,
+      "step": 6664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.026490961801058e-06,
+      "loss": 1.0645,
+      "step": 6665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.023631950996101e-06,
+      "loss": 1.0381,
+      "step": 6666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0207733262084515e-06,
+      "loss": 1.0378,
+      "step": 6667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.017915087715626e-06,
+      "loss": 1.147,
+      "step": 6668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.015057235795093e-06,
+      "loss": 1.0747,
+      "step": 6669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012199770724287e-06,
+      "loss": 1.1277,
+      "step": 6670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0093426927806074e-06,
+      "loss": 1.0317,
+      "step": 6671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0064860022414085e-06,
+      "loss": 1.0385,
+      "step": 6672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.00362969938402e-06,
+      "loss": 1.0105,
+      "step": 6673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.000773784485716e-06,
+      "loss": 1.0623,
+      "step": 6674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.997918257823745e-06,
+      "loss": 0.9963,
+      "step": 6675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.995063119675319e-06,
+      "loss": 1.0566,
+      "step": 6676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.992208370317607e-06,
+      "loss": 1.0605,
+      "step": 6677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.98935401002774e-06,
+      "loss": 1.0573,
+      "step": 6678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.986500039082816e-06,
+      "loss": 1.0591,
+      "step": 6679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983646457759891e-06,
+      "loss": 1.0286,
+      "step": 6680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9807932663359895e-06,
+      "loss": 1.0945,
+      "step": 6681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9779404650880835e-06,
+      "loss": 1.1523,
+      "step": 6682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9750880542931215e-06,
+      "loss": 0.9882,
+      "step": 6683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.972236034228011e-06,
+      "loss": 1.249,
+      "step": 6684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96938440516962e-06,
+      "loss": 1.0796,
+      "step": 6685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.966533167394777e-06,
+      "loss": 1.0227,
+      "step": 6686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.963682321180274e-06,
+      "loss": 1.0331,
+      "step": 6687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960831866802868e-06,
+      "loss": 1.0358,
+      "step": 6688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.95798180453927e-06,
+      "loss": 1.031,
+      "step": 6689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.955132134666159e-06,
+      "loss": 1.1956,
+      "step": 6690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.95228285746018e-06,
+      "loss": 0.957,
+      "step": 6691
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94943397319793e-06,
+      "loss": 1.0684,
+      "step": 6692
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.946585482155973e-06,
+      "loss": 1.0559,
+      "step": 6693
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.943737384610837e-06,
+      "loss": 1.0029,
+      "step": 6694
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.940889680839008e-06,
+      "loss": 1.053,
+      "step": 6695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938042371116932e-06,
+      "loss": 1.0483,
+      "step": 6696
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.935195455721024e-06,
+      "loss": 1.0129,
+      "step": 6697
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.932348934927652e-06,
+      "loss": 1.0002,
+      "step": 6698
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9295028090131545e-06,
+      "loss": 0.9138,
+      "step": 6699
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.926657078253826e-06,
+      "loss": 1.1028,
+      "step": 6700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923811742925923e-06,
+      "loss": 1.1067,
+      "step": 6701
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.920966803305664e-06,
+      "loss": 1.0605,
+      "step": 6702
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9181222596692345e-06,
+      "loss": 1.1101,
+      "step": 6703
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.91527811229277e-06,
+      "loss": 1.0737,
+      "step": 6704
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.912434361452379e-06,
+      "loss": 1.0657,
+      "step": 6705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9095910074241215e-06,
+      "loss": 1.1748,
+      "step": 6706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.90674805048403e-06,
+      "loss": 1.1118,
+      "step": 6707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.903905490908091e-06,
+      "loss": 0.9706,
+      "step": 6708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.901063328972253e-06,
+      "loss": 0.9252,
+      "step": 6709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89822156495243e-06,
+      "loss": 1.0754,
+      "step": 6710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8953801991244895e-06,
+      "loss": 1.074,
+      "step": 6711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.892539231764268e-06,
+      "loss": 1.103,
+      "step": 6712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.88969866314756e-06,
+      "loss": 0.9868,
+      "step": 6713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.88685849355012e-06,
+      "loss": 0.9834,
+      "step": 6714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.884018723247671e-06,
+      "loss": 1.0219,
+      "step": 6715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8811793525158864e-06,
+      "loss": 1.135,
+      "step": 6716
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.878340381630412e-06,
+      "loss": 1.145,
+      "step": 6717
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.875501810866842e-06,
+      "loss": 1.1418,
+      "step": 6718
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.872663640500742e-06,
+      "loss": 1.0857,
+      "step": 6719
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.869825870807638e-06,
+      "loss": 1.1428,
+      "step": 6720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866988502063009e-06,
+      "loss": 1.0712,
+      "step": 6721
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.864151534542304e-06,
+      "loss": 1.0657,
+      "step": 6722
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8613149685209305e-06,
+      "loss": 1.0701,
+      "step": 6723
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858478804274257e-06,
+      "loss": 1.074,
+      "step": 6724
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8556430420776115e-06,
+      "loss": 0.8636,
+      "step": 6725
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.85280768220628e-06,
+      "loss": 0.995,
+      "step": 6726
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849972724935518e-06,
+      "loss": 1.0863,
+      "step": 6727
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.847138170540535e-06,
+      "loss": 1.0206,
+      "step": 6728
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.844304019296501e-06,
+      "loss": 1.1912,
+      "step": 6729
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.841470271478555e-06,
+      "loss": 1.1279,
+      "step": 6730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.838636927361787e-06,
+      "loss": 1.0121,
+      "step": 6731
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8358039872212554e-06,
+      "loss": 1.1621,
+      "step": 6732
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.832971451331975e-06,
+      "loss": 1.0697,
+      "step": 6733
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.830139319968917e-06,
+      "loss": 1.1096,
+      "step": 6734
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.827307593407021e-06,
+      "loss": 1.0115,
+      "step": 6735
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.824476271921193e-06,
+      "loss": 0.988,
+      "step": 6736
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.821645355786281e-06,
+      "loss": 1.0096,
+      "step": 6737
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.818814845277111e-06,
+      "loss": 0.9858,
+      "step": 6738
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8159847406684615e-06,
+      "loss": 1.0576,
+      "step": 6739
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.813155042235068e-06,
+      "loss": 1.0662,
+      "step": 6740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.81032575025164e-06,
+      "loss": 1.1041,
+      "step": 6741
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8074968649928305e-06,
+      "loss": 1.0,
+      "step": 6742
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.804668386733266e-06,
+      "loss": 1.1072,
+      "step": 6743
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801840315747534e-06,
+      "loss": 0.9954,
+      "step": 6744
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.799012652310168e-06,
+      "loss": 0.9989,
+      "step": 6745
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79618539669568e-06,
+      "loss": 1.1055,
+      "step": 6746
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79335854917853e-06,
+      "loss": 0.9929,
+      "step": 6747
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.790532110033138e-06,
+      "loss": 1.0522,
+      "step": 6748
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.787706079533898e-06,
+      "loss": 1.0479,
+      "step": 6749
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7848804579551466e-06,
+      "loss": 0.9146,
+      "step": 6750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.782055245571191e-06,
+      "loss": 1.0825,
+      "step": 6751
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.779230442656303e-06,
+      "loss": 1.0771,
+      "step": 6752
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7764060494847e-06,
+      "loss": 1.0251,
+      "step": 6753
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.773582066330577e-06,
+      "loss": 1.0562,
+      "step": 6754
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.770758493468073e-06,
+      "loss": 0.9807,
+      "step": 6755
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.767935331171293e-06,
+      "loss": 0.998,
+      "step": 6756
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7651125797143114e-06,
+      "loss": 1.1272,
+      "step": 6757
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.762290239371147e-06,
+      "loss": 1.126,
+      "step": 6758
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.75946831041579e-06,
+      "loss": 1.0374,
+      "step": 6759
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.756646793122191e-06,
+      "loss": 1.1663,
+      "step": 6760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.753825687764254e-06,
+      "loss": 1.0311,
+      "step": 6761
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751004994615841e-06,
+      "loss": 1.1028,
+      "step": 6762
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.748184713950786e-06,
+      "loss": 0.9663,
+      "step": 6763
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.745364846042869e-06,
+      "loss": 1.1414,
+      "step": 6764
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.742545391165845e-06,
+      "loss": 1.1506,
+      "step": 6765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.739726349593413e-06,
+      "loss": 1.1702,
+      "step": 6766
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7369077215992415e-06,
+      "loss": 1.1057,
+      "step": 6767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.734089507456963e-06,
+      "loss": 1.0477,
+      "step": 6768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.731271707440158e-06,
+      "loss": 1.0991,
+      "step": 6769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.72845432182237e-06,
+      "loss": 1.071,
+      "step": 6770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.725637350877111e-06,
+      "loss": 1.0647,
+      "step": 6771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7228207948778416e-06,
+      "loss": 0.9243,
+      "step": 6772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.720004654097993e-06,
+      "loss": 1.0436,
+      "step": 6773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.71718892881094e-06,
+      "loss": 1.0828,
+      "step": 6774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.714373619290034e-06,
+      "loss": 1.0342,
+      "step": 6775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711558725808583e-06,
+      "loss": 1.0242,
+      "step": 6776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.708744248639848e-06,
+      "loss": 1.0699,
+      "step": 6777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.705930188057045e-06,
+      "loss": 0.9528,
+      "step": 6778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.703116544333367e-06,
+      "loss": 1.094,
+      "step": 6779
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.700303317741949e-06,
+      "loss": 1.0994,
+      "step": 6780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697490508555897e-06,
+      "loss": 1.0343,
+      "step": 6781
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.694678117048275e-06,
+      "loss": 1.126,
+      "step": 6782
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.691866143492101e-06,
+      "loss": 1.1361,
+      "step": 6783
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689054588160352e-06,
+      "loss": 1.0801,
+      "step": 6784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.686243451325973e-06,
+      "loss": 1.0493,
+      "step": 6785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68343273326186e-06,
+      "loss": 1.1816,
+      "step": 6786
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.680622434240874e-06,
+      "loss": 1.0823,
+      "step": 6787
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.67781255453583e-06,
+      "loss": 1.1055,
+      "step": 6788
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675003094419504e-06,
+      "loss": 0.9611,
+      "step": 6789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6721940541646415e-06,
+      "loss": 0.8878,
+      "step": 6790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.669385434043931e-06,
+      "loss": 1.1101,
+      "step": 6791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.666577234330024e-06,
+      "loss": 1.0281,
+      "step": 6792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.663769455295542e-06,
+      "loss": 1.0812,
+      "step": 6793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.660962097213052e-06,
+      "loss": 1.1733,
+      "step": 6794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.658155160355092e-06,
+      "loss": 1.1184,
+      "step": 6795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.655348644994146e-06,
+      "loss": 1.0583,
+      "step": 6796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.65254255140267e-06,
+      "loss": 1.1013,
+      "step": 6797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.649736879853081e-06,
+      "loss": 1.0825,
+      "step": 6798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.646931630617729e-06,
+      "loss": 1.1243,
+      "step": 6799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.644126803968953e-06,
+      "loss": 1.0688,
+      "step": 6800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.641322400179044e-06,
+      "loss": 1.0173,
+      "step": 6801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638518419520234e-06,
+      "loss": 1.1155,
+      "step": 6802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.635714862264743e-06,
+      "loss": 1.0522,
+      "step": 6803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63291172868472e-06,
+      "loss": 1.042,
+      "step": 6804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6301090190523e-06,
+      "loss": 1.1526,
+      "step": 6805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.627306733639557e-06,
+      "loss": 1.0991,
+      "step": 6806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.624504872718528e-06,
+      "loss": 1.0784,
+      "step": 6807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621703436561215e-06,
+      "loss": 1.0189,
+      "step": 6808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.618902425439581e-06,
+      "loss": 1.0627,
+      "step": 6809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.616101839625532e-06,
+      "loss": 1.0909,
+      "step": 6810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.613301679390952e-06,
+      "loss": 1.1204,
+      "step": 6811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610501945007667e-06,
+      "loss": 1.1602,
+      "step": 6812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607702636747477e-06,
+      "loss": 1.103,
+      "step": 6813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.604903754882128e-06,
+      "loss": 0.9854,
+      "step": 6814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602105299683327e-06,
+      "loss": 1.0613,
+      "step": 6815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5993072714227445e-06,
+      "loss": 1.0946,
+      "step": 6816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.596509670372012e-06,
+      "loss": 1.0868,
+      "step": 6817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.593712496802707e-06,
+      "loss": 1.0857,
+      "step": 6818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.59091575098638e-06,
+      "loss": 1.0103,
+      "step": 6819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.58811943319453e-06,
+      "loss": 0.9674,
+      "step": 6820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.585323543698612e-06,
+      "loss": 0.9448,
+      "step": 6821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.582528082770056e-06,
+      "loss": 1.0637,
+      "step": 6822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.579733050680227e-06,
+      "loss": 0.9156,
+      "step": 6823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.576938447700469e-06,
+      "loss": 1.0544,
+      "step": 6824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.574144274102078e-06,
+      "loss": 1.0488,
+      "step": 6825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571350530156298e-06,
+      "loss": 1.1541,
+      "step": 6826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5685572161343514e-06,
+      "loss": 0.9927,
+      "step": 6827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5657643323073975e-06,
+      "loss": 1.0161,
+      "step": 6828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.562971878946563e-06,
+      "loss": 1.017,
+      "step": 6829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.56017985632294e-06,
+      "loss": 1.094,
+      "step": 6830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.557388264707566e-06,
+      "loss": 1.1201,
+      "step": 6831
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.554597104371444e-06,
+      "loss": 0.9921,
+      "step": 6832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55180637558554e-06,
+      "loss": 1.0636,
+      "step": 6833
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.549016078620764e-06,
+      "loss": 0.9644,
+      "step": 6834
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.546226213747998e-06,
+      "loss": 1.053,
+      "step": 6835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.543436781238074e-06,
+      "loss": 0.9738,
+      "step": 6836
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.540647781361778e-06,
+      "loss": 0.91,
+      "step": 6837
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.537859214389871e-06,
+      "loss": 1.021,
+      "step": 6838
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535071080593053e-06,
+      "loss": 0.9822,
+      "step": 6839
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.532283380241992e-06,
+      "loss": 1.0865,
+      "step": 6840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.529496113607316e-06,
+      "loss": 1.0957,
+      "step": 6841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.526709280959604e-06,
+      "loss": 1.0779,
+      "step": 6842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.523922882569393e-06,
+      "loss": 0.9612,
+      "step": 6843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.521136918707186e-06,
+      "loss": 1.0383,
+      "step": 6844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518351389643433e-06,
+      "loss": 1.0681,
+      "step": 6845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5155662956485464e-06,
+      "loss": 1.0315,
+      "step": 6846
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.512781636992908e-06,
+      "loss": 1.0625,
+      "step": 6847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5099974139468325e-06,
+      "loss": 1.0405,
+      "step": 6848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5072136267806184e-06,
+      "loss": 1.1423,
+      "step": 6849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.504430275764505e-06,
+      "loss": 1.0647,
+      "step": 6850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5016473611686885e-06,
+      "loss": 1.0593,
+      "step": 6851
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.498864883263337e-06,
+      "loss": 1.0427,
+      "step": 6852
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.49608284231856e-06,
+      "loss": 1.189,
+      "step": 6853
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.493301238604435e-06,
+      "loss": 1.1024,
+      "step": 6854
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.490520072390999e-06,
+      "loss": 1.0642,
+      "step": 6855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.487739343948234e-06,
+      "loss": 1.0437,
+      "step": 6856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.484959053546095e-06,
+      "loss": 1.0657,
+      "step": 6857
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.482179201454483e-06,
+      "loss": 1.0903,
+      "step": 6858
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.479399787943254e-06,
+      "loss": 1.0269,
+      "step": 6859
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4766208132822375e-06,
+      "loss": 1.1411,
+      "step": 6860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.473842277741201e-06,
+      "loss": 1.0237,
+      "step": 6861
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.471064181589887e-06,
+      "loss": 1.0876,
+      "step": 6862
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4682865250979856e-06,
+      "loss": 1.1222,
+      "step": 6863
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465509308535144e-06,
+      "loss": 1.0757,
+      "step": 6864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.462732532170964e-06,
+      "loss": 1.0813,
+      "step": 6865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.459956196275018e-06,
+      "loss": 1.0959,
+      "step": 6866
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457180301116819e-06,
+      "loss": 1.1274,
+      "step": 6867
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.454404846965853e-06,
+      "loss": 1.0737,
+      "step": 6868
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.451629834091545e-06,
+      "loss": 0.96,
+      "step": 6869
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448855262763293e-06,
+      "loss": 1.0454,
+      "step": 6870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.44608113325045e-06,
+      "loss": 1.2012,
+      "step": 6871
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.44330744582232e-06,
+      "loss": 1.0035,
+      "step": 6872
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.440534200748161e-06,
+      "loss": 1.0391,
+      "step": 6873
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.437761398297201e-06,
+      "loss": 1.135,
+      "step": 6874
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.434989038738613e-06,
+      "loss": 1.2056,
+      "step": 6875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432217122341537e-06,
+      "loss": 1.0912,
+      "step": 6876
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.429445649375056e-06,
+      "loss": 1.0281,
+      "step": 6877
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.426674620108225e-06,
+      "loss": 1.0481,
+      "step": 6878
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423904034810057e-06,
+      "loss": 0.986,
+      "step": 6879
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421133893749498e-06,
+      "loss": 1.0457,
+      "step": 6880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.418364197195476e-06,
+      "loss": 0.9369,
+      "step": 6881
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.415594945416869e-06,
+      "loss": 1.0167,
+      "step": 6882
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.412826138682505e-06,
+      "loss": 1.1343,
+      "step": 6883
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410057777261181e-06,
+      "loss": 0.9902,
+      "step": 6884
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.407289861421634e-06,
+      "loss": 1.1399,
+      "step": 6885
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404522391432578e-06,
+      "loss": 1.1418,
+      "step": 6886
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.401755367562666e-06,
+      "loss": 1.0189,
+      "step": 6887
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.398988790080515e-06,
+      "loss": 1.1279,
+      "step": 6888
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396222659254697e-06,
+      "loss": 1.0286,
+      "step": 6889
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.393456975353751e-06,
+      "loss": 1.1067,
+      "step": 6890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.390691738646153e-06,
+      "loss": 0.9629,
+      "step": 6891
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.387926949400355e-06,
+      "loss": 1.1909,
+      "step": 6892
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.385162607884749e-06,
+      "loss": 1.1001,
+      "step": 6893
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.382398714367698e-06,
+      "loss": 0.9908,
+      "step": 6894
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.379635269117512e-06,
+      "loss": 1.1089,
+      "step": 6895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.376872272402458e-06,
+      "loss": 1.019,
+      "step": 6896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3741097244907625e-06,
+      "loss": 0.9622,
+      "step": 6897
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.371347625650614e-06,
+      "loss": 0.9291,
+      "step": 6898
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.368585976150142e-06,
+      "loss": 1.0542,
+      "step": 6899
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.365824776257451e-06,
+      "loss": 1.0776,
+      "step": 6900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.363064026240586e-06,
+      "loss": 1.0377,
+      "step": 6901
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360303726367554e-06,
+      "loss": 1.0519,
+      "step": 6902
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.357543876906324e-06,
+      "loss": 1.0269,
+      "step": 6903
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.35478447812481e-06,
+      "loss": 1.058,
+      "step": 6904
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352025530290892e-06,
+      "loss": 0.9861,
+      "step": 6905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.349267033672406e-06,
+      "loss": 1.1785,
+      "step": 6906
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3465089885371345e-06,
+      "loss": 1.0583,
+      "step": 6907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.34375139515283e-06,
+      "loss": 1.0942,
+      "step": 6908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340994253787189e-06,
+      "loss": 1.1011,
+      "step": 6909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.338237564707866e-06,
+      "loss": 1.0953,
+      "step": 6910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.335481328182477e-06,
+      "loss": 1.085,
+      "step": 6911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3327255444785974e-06,
+      "loss": 1.2017,
+      "step": 6912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.329970213863744e-06,
+      "loss": 1.1761,
+      "step": 6913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.327215336605407e-06,
+      "loss": 1.1079,
+      "step": 6914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.324460912971014e-06,
+      "loss": 1.1929,
+      "step": 6915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.321706943227969e-06,
+      "loss": 1.0056,
+      "step": 6916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.318953427643615e-06,
+      "loss": 1.1526,
+      "step": 6917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.316200366485256e-06,
+      "loss": 1.1436,
+      "step": 6918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.313447760020156e-06,
+      "loss": 1.0691,
+      "step": 6919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.310695608515535e-06,
+      "loss": 1.0442,
+      "step": 6920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.30794391223856e-06,
+      "loss": 0.8699,
+      "step": 6921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3051926714563685e-06,
+      "loss": 1.1038,
+      "step": 6922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.302441886436038e-06,
+      "loss": 1.045,
+      "step": 6923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299691557444608e-06,
+      "loss": 1.1245,
+      "step": 6924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29694168474908e-06,
+      "loss": 1.0496,
+      "step": 6925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294192268616398e-06,
+      "loss": 0.9624,
+      "step": 6926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.291443309313475e-06,
+      "loss": 1.1326,
+      "step": 6927
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.288694807107178e-06,
+      "loss": 1.0088,
+      "step": 6928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2859467622643154e-06,
+      "loss": 1.0825,
+      "step": 6929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.283199175051672e-06,
+      "loss": 1.0208,
+      "step": 6930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2804520457359735e-06,
+      "loss": 0.9808,
+      "step": 6931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277705374583899e-06,
+      "loss": 1.2031,
+      "step": 6932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2749591618621e-06,
+      "loss": 1.053,
+      "step": 6933
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.272213407837164e-06,
+      "loss": 1.0894,
+      "step": 6934
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269468112775649e-06,
+      "loss": 1.0892,
+      "step": 6935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2667232769440615e-06,
+      "loss": 1.0227,
+      "step": 6936
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.263978900608862e-06,
+      "loss": 1.0583,
+      "step": 6937
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.261234984036472e-06,
+      "loss": 1.0186,
+      "step": 6938
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.258491527493262e-06,
+      "loss": 1.104,
+      "step": 6939
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.255748531245559e-06,
+      "loss": 1.0115,
+      "step": 6940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.253005995559654e-06,
+      "loss": 1.1521,
+      "step": 6941
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25026392070178e-06,
+      "loss": 0.9626,
+      "step": 6942
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2475223069381334e-06,
+      "loss": 0.9417,
+      "step": 6943
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244781154534869e-06,
+      "loss": 1.095,
+      "step": 6944
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.242040463758089e-06,
+      "loss": 1.0759,
+      "step": 6945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239300234873847e-06,
+      "loss": 1.1038,
+      "step": 6946
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.236560468148171e-06,
+      "loss": 1.0963,
+      "step": 6947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.233821163847019e-06,
+      "loss": 1.0674,
+      "step": 6948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23108232223633e-06,
+      "loss": 1.0106,
+      "step": 6949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.22834394358197e-06,
+      "loss": 1.0447,
+      "step": 6950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.225606028149786e-06,
+      "loss": 1.0293,
+      "step": 6951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.222868576205573e-06,
+      "loss": 1.1055,
+      "step": 6952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.22013158801506e-06,
+      "loss": 0.9749,
+      "step": 6953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2173950638439605e-06,
+      "loss": 1.0713,
+      "step": 6954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2146590039579314e-06,
+      "loss": 1.1824,
+      "step": 6955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2119234086225745e-06,
+      "loss": 0.9409,
+      "step": 6956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.209188278103465e-06,
+      "loss": 1.1472,
+      "step": 6957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.206453612666117e-06,
+      "loss": 1.1724,
+      "step": 6958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.203719412576007e-06,
+      "loss": 1.0488,
+      "step": 6959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.200985678098576e-06,
+      "loss": 1.04,
+      "step": 6960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1982524094991916e-06,
+      "loss": 1.0625,
+      "step": 6961
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195519607043201e-06,
+      "loss": 1.0811,
+      "step": 6962
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.192787270995905e-06,
+      "loss": 1.0801,
+      "step": 6963
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190055401622542e-06,
+      "loss": 1.1292,
+      "step": 6964
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.187323999188327e-06,
+      "loss": 1.0386,
+      "step": 6965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.184593063958409e-06,
+      "loss": 0.9871,
+      "step": 6966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.18186259619791e-06,
+      "loss": 1.073,
+      "step": 6967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.179132596171893e-06,
+      "loss": 0.988,
+      "step": 6968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.176403064145379e-06,
+      "loss": 1.0864,
+      "step": 6969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173674000383347e-06,
+      "loss": 1.1831,
+      "step": 6970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.170945405150731e-06,
+      "loss": 0.9421,
+      "step": 6971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.168217278712413e-06,
+      "loss": 1.0845,
+      "step": 6972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.165489621333239e-06,
+      "loss": 1.1392,
+      "step": 6973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.162762433278003e-06,
+      "loss": 1.0552,
+      "step": 6974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.160035714811448e-06,
+      "loss": 1.1277,
+      "step": 6975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.157309466198286e-06,
+      "loss": 0.9955,
+      "step": 6976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.15458368770317e-06,
+      "loss": 0.9941,
+      "step": 6977
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.151858379590713e-06,
+      "loss": 1.0757,
+      "step": 6978
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.149133542125488e-06,
+      "loss": 1.1233,
+      "step": 6979
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14640917557201e-06,
+      "loss": 1.0195,
+      "step": 6980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.143685280194757e-06,
+      "loss": 0.9003,
+      "step": 6981
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.140961856258166e-06,
+      "loss": 0.9333,
+      "step": 6982
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138238904026606e-06,
+      "loss": 1.0618,
+      "step": 6983
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135516423764424e-06,
+      "loss": 1.0916,
+      "step": 6984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.132794415735916e-06,
+      "loss": 1.0146,
+      "step": 6985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130072880205321e-06,
+      "loss": 1.0946,
+      "step": 6986
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.127351817436847e-06,
+      "loss": 1.103,
+      "step": 6987
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.124631227694643e-06,
+      "loss": 1.0631,
+      "step": 6988
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.121911111242823e-06,
+      "loss": 1.0297,
+      "step": 6989
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1191914683454505e-06,
+      "loss": 1.0072,
+      "step": 6990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.116472299266534e-06,
+      "loss": 1.1072,
+      "step": 6991
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113753604270051e-06,
+      "loss": 1.0781,
+      "step": 6992
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.111035383619929e-06,
+      "loss": 1.0474,
+      "step": 6993
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1083176375800424e-06,
+      "loss": 1.0393,
+      "step": 6994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1056003664142285e-06,
+      "loss": 0.9758,
+      "step": 6995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1028835703862676e-06,
+      "loss": 1.0908,
+      "step": 6996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.100167249759909e-06,
+      "loss": 1.1794,
+      "step": 6997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.097451404798843e-06,
+      "loss": 1.0614,
+      "step": 6998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.094736035766714e-06,
+      "loss": 0.9722,
+      "step": 6999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092021142927127e-06,
+      "loss": 1.141,
+      "step": 7000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089306726543643e-06,
+      "loss": 1.098,
+      "step": 7001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.086592786879765e-06,
+      "loss": 1.0592,
+      "step": 7002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0838793241989635e-06,
+      "loss": 0.995,
+      "step": 7003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.081166338764651e-06,
+      "loss": 1.1165,
+      "step": 7004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.078453830840193e-06,
+      "loss": 1.0989,
+      "step": 7005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.075741800688926e-06,
+      "loss": 1.062,
+      "step": 7006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.073030248574116e-06,
+      "loss": 1.0764,
+      "step": 7007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0703191747590015e-06,
+      "loss": 1.0657,
+      "step": 7008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.067608579506771e-06,
+      "loss": 1.0499,
+      "step": 7009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064898463080555e-06,
+      "loss": 1.0243,
+      "step": 7010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0621888257434535e-06,
+      "loss": 1.0581,
+      "step": 7011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.059479667758509e-06,
+      "loss": 1.0481,
+      "step": 7012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.056770989388716e-06,
+      "loss": 1.2219,
+      "step": 7013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.054062790897036e-06,
+      "loss": 0.9999,
+      "step": 7014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.051355072546367e-06,
+      "loss": 1.0088,
+      "step": 7015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.048647834599571e-06,
+      "loss": 1.053,
+      "step": 7016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045941077319467e-06,
+      "loss": 0.9121,
+      "step": 7017
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.043234800968813e-06,
+      "loss": 1.0564,
+      "step": 7018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.040529005810335e-06,
+      "loss": 1.0339,
+      "step": 7019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037823692106703e-06,
+      "loss": 1.1069,
+      "step": 7020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.035118860120538e-06,
+      "loss": 1.0664,
+      "step": 7021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.03241451011443e-06,
+      "loss": 1.009,
+      "step": 7022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0297106423509e-06,
+      "loss": 1.017,
+      "step": 7023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.02700725709244e-06,
+      "loss": 1.0607,
+      "step": 7024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0243043546014915e-06,
+      "loss": 0.9724,
+      "step": 7025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.021601935140443e-06,
+      "loss": 1.1653,
+      "step": 7026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0188999989716354e-06,
+      "loss": 1.0635,
+      "step": 7027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.016198546357377e-06,
+      "loss": 1.0818,
+      "step": 7028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0134975775599095e-06,
+      "loss": 1.0983,
+      "step": 7029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.010797092841443e-06,
+      "loss": 1.0747,
+      "step": 7030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0080970924641304e-06,
+      "loss": 0.9121,
+      "step": 7031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.005397576690083e-06,
+      "loss": 0.9832,
+      "step": 7032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.002698545781374e-06,
+      "loss": 1.1611,
+      "step": 7033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 1.0071,
+      "step": 7034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.997301939607946e-06,
+      "loss": 0.9933,
+      "step": 7035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.99460436486713e-06,
+      "loss": 1.0206,
+      "step": 7036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.991907276039424e-06,
+      "loss": 1.0023,
+      "step": 7037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.989210673386658e-06,
+      "loss": 1.0542,
+      "step": 7038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.986514557170611e-06,
+      "loss": 1.1628,
+      "step": 7039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.983818927653014e-06,
+      "loss": 1.0009,
+      "step": 7040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.981123785095566e-06,
+      "loss": 0.9991,
+      "step": 7041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9784291297598865e-06,
+      "loss": 1.1433,
+      "step": 7042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.975734961907577e-06,
+      "loss": 1.0295,
+      "step": 7043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9730412818001825e-06,
+      "loss": 0.9072,
+      "step": 7044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.970348089699196e-06,
+      "loss": 1.0701,
+      "step": 7045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.967655385866072e-06,
+      "loss": 1.0183,
+      "step": 7046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.964963170562204e-06,
+      "loss": 0.9233,
+      "step": 7047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.962271444048956e-06,
+      "loss": 1.0659,
+      "step": 7048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.959580206587626e-06,
+      "loss": 1.0911,
+      "step": 7049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9568894584394835e-06,
+      "loss": 1.1299,
+      "step": 7050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9541991998657305e-06,
+      "loss": 1.0151,
+      "step": 7051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.951509431127539e-06,
+      "loss": 1.0696,
+      "step": 7052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948820152486021e-06,
+      "loss": 1.0664,
+      "step": 7053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9461313642022476e-06,
+      "loss": 1.0806,
+      "step": 7054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.943443066537249e-06,
+      "loss": 1.1296,
+      "step": 7055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940755259751984e-06,
+      "loss": 1.0688,
+      "step": 7056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.938067944107387e-06,
+      "loss": 1.1902,
+      "step": 7057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9353811198643395e-06,
+      "loss": 1.0735,
+      "step": 7058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.932694787283667e-06,
+      "loss": 1.1832,
+      "step": 7059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.930008946626159e-06,
+      "loss": 0.9922,
+      "step": 7060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9273235981525435e-06,
+      "loss": 0.887,
+      "step": 7061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.924638742123514e-06,
+      "loss": 1.1545,
+      "step": 7062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921954378799716e-06,
+      "loss": 1.0608,
+      "step": 7063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.919270508441728e-06,
+      "loss": 1.017,
+      "step": 7064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.916587131310101e-06,
+      "loss": 1.0562,
+      "step": 7065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.913904247665336e-06,
+      "loss": 1.1287,
+      "step": 7066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.911221857767874e-06,
+      "loss": 0.9897,
+      "step": 7067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9085399618781226e-06,
+      "loss": 0.9557,
+      "step": 7068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.905858560256428e-06,
+      "loss": 1.0513,
+      "step": 7069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903177653163101e-06,
+      "loss": 1.0018,
+      "step": 7070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900497240858398e-06,
+      "loss": 1.0251,
+      "step": 7071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8978173236025184e-06,
+      "loss": 1.0862,
+      "step": 7072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.895137901655631e-06,
+      "loss": 1.0911,
+      "step": 7073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892458975277851e-06,
+      "loss": 1.1558,
+      "step": 7074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8897805447292334e-06,
+      "loss": 1.0305,
+      "step": 7075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887102610269805e-06,
+      "loss": 1.0725,
+      "step": 7076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.884425172159529e-06,
+      "loss": 1.0492,
+      "step": 7077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881748230658322e-06,
+      "loss": 1.0776,
+      "step": 7078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879071786026062e-06,
+      "loss": 1.0669,
+      "step": 7079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.876395838522565e-06,
+      "loss": 1.0007,
+      "step": 7080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.873720388407612e-06,
+      "loss": 1.042,
+      "step": 7081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871045435940932e-06,
+      "loss": 0.9916,
+      "step": 7082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.868370981382198e-06,
+      "loss": 1.1421,
+      "step": 7083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.865697024991045e-06,
+      "loss": 1.1183,
+      "step": 7084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.863023567027053e-06,
+      "loss": 1.092,
+      "step": 7085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.860350607749752e-06,
+      "loss": 0.913,
+      "step": 7086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.857678147418634e-06,
+      "loss": 1.0872,
+      "step": 7087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85500618629313e-06,
+      "loss": 0.9968,
+      "step": 7088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852334724632628e-06,
+      "loss": 1.2034,
+      "step": 7089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.849663762696476e-06,
+      "loss": 1.0078,
+      "step": 7090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.846993300743955e-06,
+      "loss": 1.0046,
+      "step": 7091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844323339034315e-06,
+      "loss": 1.0453,
+      "step": 7092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.84165387782675e-06,
+      "loss": 1.0145,
+      "step": 7093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.838984917380398e-06,
+      "loss": 1.0546,
+      "step": 7094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.836316457954365e-06,
+      "loss": 1.0874,
+      "step": 7095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.833648499807691e-06,
+      "loss": 1.111,
+      "step": 7096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.830981043199382e-06,
+      "loss": 0.9548,
+      "step": 7097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8283140883883895e-06,
+      "loss": 1.0928,
+      "step": 7098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.825647635633614e-06,
+      "loss": 0.9604,
+      "step": 7099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.822981685193903e-06,
+      "loss": 1.1753,
+      "step": 7100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.820316237328072e-06,
+      "loss": 1.1152,
+      "step": 7101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.817651292294868e-06,
+      "loss": 1.0593,
+      "step": 7102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.814986850353005e-06,
+      "loss": 1.1089,
+      "step": 7103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.812322911761135e-06,
+      "loss": 1.0406,
+      "step": 7104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.809659476777869e-06,
+      "loss": 1.0195,
+      "step": 7105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.806996545661775e-06,
+      "loss": 0.9878,
+      "step": 7106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.804334118671358e-06,
+      "loss": 1.042,
+      "step": 7107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801672196065078e-06,
+      "loss": 1.0093,
+      "step": 7108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.799010778101355e-06,
+      "loss": 1.1737,
+      "step": 7109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7963498650385485e-06,
+      "loss": 1.0713,
+      "step": 7110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.79368945713498e-06,
+      "loss": 1.1069,
+      "step": 7111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.791029554648911e-06,
+      "loss": 1.1431,
+      "step": 7112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.788370157838559e-06,
+      "loss": 0.8765,
+      "step": 7113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7857112669621e-06,
+      "loss": 1.0121,
+      "step": 7114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.783052882277648e-06,
+      "loss": 1.0798,
+      "step": 7115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.780395004043269e-06,
+      "loss": 1.0852,
+      "step": 7116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7777376325169935e-06,
+      "loss": 1.0637,
+      "step": 7117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7750807679567835e-06,
+      "loss": 1.1511,
+      "step": 7118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772424410620567e-06,
+      "loss": 1.0129,
+      "step": 7119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7697685607662225e-06,
+      "loss": 0.9683,
+      "step": 7120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.767113218651568e-06,
+      "loss": 1.0894,
+      "step": 7121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764458384534376e-06,
+      "loss": 1.093,
+      "step": 7122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.76180405867238e-06,
+      "loss": 1.0986,
+      "step": 7123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.759150241323247e-06,
+      "loss": 1.0754,
+      "step": 7124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.756496932744613e-06,
+      "loss": 0.9736,
+      "step": 7125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7538441331940485e-06,
+      "loss": 1.1172,
+      "step": 7126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7511918429290835e-06,
+      "loss": 1.0343,
+      "step": 7127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.748540062207202e-06,
+      "loss": 1.1287,
+      "step": 7128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.745888791285829e-06,
+      "loss": 1.1975,
+      "step": 7129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.743238030422339e-06,
+      "loss": 1.0854,
+      "step": 7130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.740587779874072e-06,
+      "loss": 1.1165,
+      "step": 7131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7379380398983e-06,
+      "loss": 1.1023,
+      "step": 7132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.735288810752262e-06,
+      "loss": 1.0485,
+      "step": 7133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732640092693132e-06,
+      "loss": 1.0116,
+      "step": 7134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.729991885978045e-06,
+      "loss": 1.085,
+      "step": 7135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7273441908640905e-06,
+      "loss": 1.1099,
+      "step": 7136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.724697007608288e-06,
+      "loss": 1.106,
+      "step": 7137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.722050336467626e-06,
+      "loss": 1.0522,
+      "step": 7138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.719404177699042e-06,
+      "loss": 1.1223,
+      "step": 7139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.716758531559412e-06,
+      "loss": 1.0105,
+      "step": 7140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.714113398305577e-06,
+      "loss": 1.0593,
+      "step": 7141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.711468778194315e-06,
+      "loss": 1.0024,
+      "step": 7142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.708824671482361e-06,
+      "loss": 1.1411,
+      "step": 7143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.706181078426408e-06,
+      "loss": 1.1299,
+      "step": 7144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.703537999283076e-06,
+      "loss": 1.1523,
+      "step": 7145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7008954343089565e-06,
+      "loss": 1.0618,
+      "step": 7146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6982533837605885e-06,
+      "loss": 1.0676,
+      "step": 7147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695611847894447e-06,
+      "loss": 1.0781,
+      "step": 7148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.692970826966976e-06,
+      "loss": 1.0728,
+      "step": 7149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.690330321234552e-06,
+      "loss": 1.1,
+      "step": 7150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.687690330953517e-06,
+      "loss": 0.9961,
+      "step": 7151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.685050856380153e-06,
+      "loss": 0.988,
+      "step": 7152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.68241189777069e-06,
+      "loss": 1.108,
+      "step": 7153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.679773455381316e-06,
+      "loss": 1.1278,
+      "step": 7154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.67713552946817e-06,
+      "loss": 0.9353,
+      "step": 7155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674498120287328e-06,
+      "loss": 1.0494,
+      "step": 7156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671861228094831e-06,
+      "loss": 1.045,
+      "step": 7157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669224853146659e-06,
+      "loss": 1.048,
+      "step": 7158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666588995698743e-06,
+      "loss": 1.0693,
+      "step": 7159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663953656006975e-06,
+      "loss": 1.0209,
+      "step": 7160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6613188343271776e-06,
+      "loss": 0.9872,
+      "step": 7161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65868453091514e-06,
+      "loss": 1.0717,
+      "step": 7162
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.656050746026596e-06,
+      "loss": 1.0449,
+      "step": 7163
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653417479917223e-06,
+      "loss": 1.1237,
+      "step": 7164
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.650784732842658e-06,
+      "loss": 1.1032,
+      "step": 7165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.648152505058481e-06,
+      "loss": 0.9846,
+      "step": 7166
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.645520796820218e-06,
+      "loss": 1.1436,
+      "step": 7167
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6428896083833566e-06,
+      "loss": 1.1177,
+      "step": 7168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64025894000332e-06,
+      "loss": 1.1095,
+      "step": 7169
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637628791935492e-06,
+      "loss": 1.1577,
+      "step": 7170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.634999164435205e-06,
+      "loss": 1.0303,
+      "step": 7171
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.632370057757731e-06,
+      "loss": 1.1238,
+      "step": 7172
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629741472158304e-06,
+      "loss": 1.0188,
+      "step": 7173
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6271134078920995e-06,
+      "loss": 1.1208,
+      "step": 7174
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.62448586521424e-06,
+      "loss": 1.0742,
+      "step": 7175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.621858844379809e-06,
+      "loss": 1.0381,
+      "step": 7176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619232345643823e-06,
+      "loss": 0.9905,
+      "step": 7177
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.616606369261263e-06,
+      "loss": 1.0659,
+      "step": 7178
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.613980915487056e-06,
+      "loss": 1.1099,
+      "step": 7179
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.611355984576071e-06,
+      "loss": 1.0198,
+      "step": 7180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.60873157678313e-06,
+      "loss": 0.8767,
+      "step": 7181
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.606107692363008e-06,
+      "loss": 0.9606,
+      "step": 7182
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603484331570422e-06,
+      "loss": 1.0386,
+      "step": 7183
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6008614946600425e-06,
+      "loss": 1.0504,
+      "step": 7184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.598239181886497e-06,
+      "loss": 0.9971,
+      "step": 7185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595617393504343e-06,
+      "loss": 1.0693,
+      "step": 7186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592996129768108e-06,
+      "loss": 0.9362,
+      "step": 7187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590375390932254e-06,
+      "loss": 1.0854,
+      "step": 7188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.587755177251192e-06,
+      "loss": 1.0215,
+      "step": 7189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585135488979296e-06,
+      "loss": 1.0413,
+      "step": 7190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.582516326370871e-06,
+      "loss": 1.0812,
+      "step": 7191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579897689680184e-06,
+      "loss": 1.1311,
+      "step": 7192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57727957916145e-06,
+      "loss": 1.0422,
+      "step": 7193
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.574661995068822e-06,
+      "loss": 1.0654,
+      "step": 7194
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.572044937656418e-06,
+      "loss": 1.1206,
+      "step": 7195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569428407178291e-06,
+      "loss": 1.1472,
+      "step": 7196
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.566812403888446e-06,
+      "loss": 0.9835,
+      "step": 7197
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.564196928040846e-06,
+      "loss": 1.1069,
+      "step": 7198
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.561581979889388e-06,
+      "loss": 1.2695,
+      "step": 7199
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.558967559687929e-06,
+      "loss": 1.0437,
+      "step": 7200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.556353667690275e-06,
+      "loss": 0.9214,
+      "step": 7201
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5537403041501745e-06,
+      "loss": 0.9523,
+      "step": 7202
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551127469321324e-06,
+      "loss": 1.1279,
+      "step": 7203
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.548515163457376e-06,
+      "loss": 0.9501,
+      "step": 7204
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.545903386811924e-06,
+      "loss": 1.1279,
+      "step": 7205
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543292139638519e-06,
+      "loss": 1.0479,
+      "step": 7206
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540681422190648e-06,
+      "loss": 1.095,
+      "step": 7207
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.538071234721758e-06,
+      "loss": 0.9596,
+      "step": 7208
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.535461577485245e-06,
+      "loss": 1.1389,
+      "step": 7209
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532852450734444e-06,
+      "loss": 1.0812,
+      "step": 7210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530243854722639e-06,
+      "loss": 1.1091,
+      "step": 7211
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.527635789703076e-06,
+      "loss": 1.0027,
+      "step": 7212
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.525028255928933e-06,
+      "loss": 0.9941,
+      "step": 7213
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5224212536533496e-06,
+      "loss": 1.0254,
+      "step": 7214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5198147831294015e-06,
+      "loss": 1.1362,
+      "step": 7215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.517208844610124e-06,
+      "loss": 1.2061,
+      "step": 7216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5146034383485015e-06,
+      "loss": 1.0642,
+      "step": 7217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.511998564597448e-06,
+      "loss": 1.1185,
+      "step": 7218
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.509394223609846e-06,
+      "loss": 1.0818,
+      "step": 7219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.506790415638523e-06,
+      "loss": 1.0682,
+      "step": 7220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504187140936244e-06,
+      "loss": 1.1282,
+      "step": 7221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.501584399755736e-06,
+      "loss": 0.9678,
+      "step": 7222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4989821923496605e-06,
+      "loss": 0.9873,
+      "step": 7223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.496380518970642e-06,
+      "loss": 1.0276,
+      "step": 7224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.493779379871241e-06,
+      "loss": 1.1318,
+      "step": 7225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491178775303968e-06,
+      "loss": 1.014,
+      "step": 7226
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488578705521287e-06,
+      "loss": 0.9832,
+      "step": 7227
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485979170775611e-06,
+      "loss": 0.9583,
+      "step": 7228
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.483380171319289e-06,
+      "loss": 1.1411,
+      "step": 7229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4807817074046354e-06,
+      "loss": 1.0321,
+      "step": 7230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.478183779283895e-06,
+      "loss": 1.1453,
+      "step": 7231
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475586387209277e-06,
+      "loss": 1.0701,
+      "step": 7232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4729895314329275e-06,
+      "loss": 1.0183,
+      "step": 7233
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4703932122069386e-06,
+      "loss": 0.9895,
+      "step": 7234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.46779742978336e-06,
+      "loss": 0.9929,
+      "step": 7235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.465202184414188e-06,
+      "loss": 1.0527,
+      "step": 7236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.462607476351357e-06,
+      "loss": 1.0333,
+      "step": 7237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.460013305846762e-06,
+      "loss": 0.9149,
+      "step": 7238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.457419673152237e-06,
+      "loss": 1.0466,
+      "step": 7239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454826578519561e-06,
+      "loss": 1.084,
+      "step": 7240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.452234022200474e-06,
+      "loss": 0.9619,
+      "step": 7241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449642004446649e-06,
+      "loss": 0.9119,
+      "step": 7242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.447050525509718e-06,
+      "loss": 1.126,
+      "step": 7243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.444459585641259e-06,
+      "loss": 1.0439,
+      "step": 7244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441869185092787e-06,
+      "loss": 1.1091,
+      "step": 7245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439279324115779e-06,
+      "loss": 1.1631,
+      "step": 7246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.436690002961654e-06,
+      "loss": 1.0549,
+      "step": 7247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.434101221881768e-06,
+      "loss": 1.0862,
+      "step": 7248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.431512981127442e-06,
+      "loss": 0.9725,
+      "step": 7249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428925280949941e-06,
+      "loss": 1.0598,
+      "step": 7250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.426338121600463e-06,
+      "loss": 1.178,
+      "step": 7251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423751503330173e-06,
+      "loss": 1.1384,
+      "step": 7252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.421165426390167e-06,
+      "loss": 1.0161,
+      "step": 7253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.418579891031504e-06,
+      "loss": 1.0818,
+      "step": 7254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4159948975051766e-06,
+      "loss": 1.0901,
+      "step": 7255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.413410446062128e-06,
+      "loss": 0.8524,
+      "step": 7256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.410826536953256e-06,
+      "loss": 1.0941,
+      "step": 7257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408243170429402e-06,
+      "loss": 0.964,
+      "step": 7258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.405660346741347e-06,
+      "loss": 0.8853,
+      "step": 7259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403078066139835e-06,
+      "loss": 1.1152,
+      "step": 7260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.400496328875544e-06,
+      "loss": 1.0962,
+      "step": 7261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3979151351991e-06,
+      "loss": 1.0397,
+      "step": 7262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.395334485361086e-06,
+      "loss": 1.1074,
+      "step": 7263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.392754379612019e-06,
+      "loss": 0.9945,
+      "step": 7264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.390174818202374e-06,
+      "loss": 1.0442,
+      "step": 7265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.387595801382574e-06,
+      "loss": 1.105,
+      "step": 7266
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385017329402975e-06,
+      "loss": 0.9702,
+      "step": 7267
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.382439402513899e-06,
+      "loss": 1.0334,
+      "step": 7268
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379862020965601e-06,
+      "loss": 1.125,
+      "step": 7269
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3772851850082835e-06,
+      "loss": 0.9316,
+      "step": 7270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.374708894892108e-06,
+      "loss": 1.0283,
+      "step": 7271
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372133150867168e-06,
+      "loss": 1.0322,
+      "step": 7272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.369557953183516e-06,
+      "loss": 1.0698,
+      "step": 7273
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.366983302091149e-06,
+      "loss": 1.0627,
+      "step": 7274
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36440919784e-06,
+      "loss": 1.0071,
+      "step": 7275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361835640679969e-06,
+      "loss": 1.1427,
+      "step": 7276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359262630860883e-06,
+      "loss": 1.1265,
+      "step": 7277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356690168632524e-06,
+      "loss": 1.093,
+      "step": 7278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.354118254244627e-06,
+      "loss": 1.042,
+      "step": 7279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.35154688794686e-06,
+      "loss": 1.0332,
+      "step": 7280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.348976069988851e-06,
+      "loss": 0.9875,
+      "step": 7281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.346405800620174e-06,
+      "loss": 1.1284,
+      "step": 7282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.343836080090339e-06,
+      "loss": 1.0891,
+      "step": 7283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.341266908648806e-06,
+      "loss": 1.0967,
+      "step": 7284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.338698286544992e-06,
+      "loss": 0.9469,
+      "step": 7285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.336130214028248e-06,
+      "loss": 1.0841,
+      "step": 7286
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333562691347881e-06,
+      "loss": 0.9083,
+      "step": 7287
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.330995718753136e-06,
+      "loss": 0.9839,
+      "step": 7288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3284292964932115e-06,
+      "loss": 1.0447,
+      "step": 7289
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3258634248172596e-06,
+      "loss": 1.1326,
+      "step": 7290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3232981039743515e-06,
+      "loss": 1.1333,
+      "step": 7291
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.320733334213533e-06,
+      "loss": 1.0449,
+      "step": 7292
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318169115783789e-06,
+      "loss": 1.1165,
+      "step": 7293
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315605448934043e-06,
+      "loss": 1.0254,
+      "step": 7294
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3130423339131745e-06,
+      "loss": 1.0332,
+      "step": 7295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31047977097e-06,
+      "loss": 1.0323,
+      "step": 7296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307917760353291e-06,
+      "loss": 0.9956,
+      "step": 7297
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305356302311769e-06,
+      "loss": 1.0952,
+      "step": 7298
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.30279539709408e-06,
+      "loss": 1.0482,
+      "step": 7299
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.300235044948838e-06,
+      "loss": 1.1226,
+      "step": 7300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297675246124603e-06,
+      "loss": 1.1106,
+      "step": 7301
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.295116000869864e-06,
+      "loss": 1.1217,
+      "step": 7302
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292557309433077e-06,
+      "loss": 1.0991,
+      "step": 7303
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.289999172062626e-06,
+      "loss": 1.0586,
+      "step": 7304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2874415890068565e-06,
+      "loss": 1.1012,
+      "step": 7305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.284884560514048e-06,
+      "loss": 1.0942,
+      "step": 7306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282328086832433e-06,
+      "loss": 1.0664,
+      "step": 7307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279772168210185e-06,
+      "loss": 1.1193,
+      "step": 7308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2772168048954365e-06,
+      "loss": 1.0175,
+      "step": 7309
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.274661997136247e-06,
+      "loss": 1.1416,
+      "step": 7310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272107745180639e-06,
+      "loss": 1.0641,
+      "step": 7311
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2695540492765675e-06,
+      "loss": 1.0731,
+      "step": 7312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.267000909671946e-06,
+      "loss": 1.0955,
+      "step": 7313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.264448326614625e-06,
+      "loss": 0.923,
+      "step": 7314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.261896300352399e-06,
+      "loss": 1.116,
+      "step": 7315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.259344831133019e-06,
+      "loss": 1.0386,
+      "step": 7316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.256793919204178e-06,
+      "loss": 1.0471,
+      "step": 7317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.254243564813507e-06,
+      "loss": 1.0344,
+      "step": 7318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.251693768208591e-06,
+      "loss": 1.1289,
+      "step": 7319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2491445296369685e-06,
+      "loss": 1.0352,
+      "step": 7320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.246595849346097e-06,
+      "loss": 1.0295,
+      "step": 7321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244047727583406e-06,
+      "loss": 1.0586,
+      "step": 7322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241500164596264e-06,
+      "loss": 1.0583,
+      "step": 7323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.238953160631977e-06,
+      "loss": 1.0872,
+      "step": 7324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.236406715937808e-06,
+      "loss": 1.0852,
+      "step": 7325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2338608307609564e-06,
+      "loss": 1.1567,
+      "step": 7326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.231315505348576e-06,
+      "loss": 0.9927,
+      "step": 7327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2287707399477595e-06,
+      "loss": 1.0132,
+      "step": 7328
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.226226534805542e-06,
+      "loss": 1.0256,
+      "step": 7329
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.223682890168914e-06,
+      "loss": 1.1239,
+      "step": 7330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.22113980628481e-06,
+      "loss": 1.1761,
+      "step": 7331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.218597283400101e-06,
+      "loss": 1.0778,
+      "step": 7332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.216055321761617e-06,
+      "loss": 1.0698,
+      "step": 7333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.213513921616118e-06,
+      "loss": 1.1416,
+      "step": 7334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.210973083210325e-06,
+      "loss": 1.0481,
+      "step": 7335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208432806790894e-06,
+      "loss": 0.9319,
+      "step": 7336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.205893092604426e-06,
+      "loss": 1.0649,
+      "step": 7337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203353940897474e-06,
+      "loss": 1.0122,
+      "step": 7338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.200815351916539e-06,
+      "loss": 1.0027,
+      "step": 7339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.198277325908052e-06,
+      "loss": 1.0625,
+      "step": 7340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.195739863118407e-06,
+      "loss": 1.1169,
+      "step": 7341
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.193202963793934e-06,
+      "loss": 1.0732,
+      "step": 7342
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1906666281809025e-06,
+      "loss": 1.0778,
+      "step": 7343
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.188130856525545e-06,
+      "loss": 1.1353,
+      "step": 7344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.185595649074021e-06,
+      "loss": 1.0934,
+      "step": 7345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.183061006072445e-06,
+      "loss": 1.2102,
+      "step": 7346
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.180526927766879e-06,
+      "loss": 1.0842,
+      "step": 7347
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1779934144033195e-06,
+      "loss": 1.0366,
+      "step": 7348
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.175460466227721e-06,
+      "loss": 1.1714,
+      "step": 7349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172928083485974e-06,
+      "loss": 1.067,
+      "step": 7350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.170396266423912e-06,
+      "loss": 0.9612,
+      "step": 7351
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.167865015287326e-06,
+      "loss": 1.1387,
+      "step": 7352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165334330321937e-06,
+      "loss": 1.0591,
+      "step": 7353
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1628042117734236e-06,
+      "loss": 1.0662,
+      "step": 7354
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.160274659887406e-06,
+      "loss": 0.9698,
+      "step": 7355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.157745674909443e-06,
+      "loss": 1.071,
+      "step": 7356
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.155217257085047e-06,
+      "loss": 1.1389,
+      "step": 7357
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.152689406659669e-06,
+      "loss": 1.1506,
+      "step": 7358
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.150162123878705e-06,
+      "loss": 1.1213,
+      "step": 7359
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147635408987504e-06,
+      "loss": 1.1533,
+      "step": 7360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.145109262231347e-06,
+      "loss": 0.9236,
+      "step": 7361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.142583683855471e-06,
+      "loss": 1.0261,
+      "step": 7362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.140058674105058e-06,
+      "loss": 1.1257,
+      "step": 7363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137534233225226e-06,
+      "loss": 1.1547,
+      "step": 7364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.135010361461037e-06,
+      "loss": 1.0911,
+      "step": 7365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.132487059057514e-06,
+      "loss": 1.0499,
+      "step": 7366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.129964326259604e-06,
+      "loss": 1.0391,
+      "step": 7367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1274421633122174e-06,
+      "loss": 1.0908,
+      "step": 7368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.124920570460192e-06,
+      "loss": 1.0957,
+      "step": 7369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.122399547948323e-06,
+      "loss": 1.1157,
+      "step": 7370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.119879096021352e-06,
+      "loss": 1.0471,
+      "step": 7371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.117359214923945e-06,
+      "loss": 1.0586,
+      "step": 7372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.114839904900735e-06,
+      "loss": 1.0564,
+      "step": 7373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1123211661962934e-06,
+      "loss": 1.105,
+      "step": 7374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109802999055127e-06,
+      "loss": 0.9493,
+      "step": 7375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.107285403721703e-06,
+      "loss": 1.1401,
+      "step": 7376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104768380440415e-06,
+      "loss": 0.9973,
+      "step": 7377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.102251929455614e-06,
+      "loss": 1.0891,
+      "step": 7378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.099736051011599e-06,
+      "loss": 1.0815,
+      "step": 7379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.097220745352592e-06,
+      "loss": 1.0879,
+      "step": 7380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.094706012722782e-06,
+      "loss": 1.03,
+      "step": 7381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.092191853366296e-06,
+      "loss": 1.0066,
+      "step": 7382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089678267527196e-06,
+      "loss": 1.0864,
+      "step": 7383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.087165255449502e-06,
+      "loss": 1.0591,
+      "step": 7384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.084652817377167e-06,
+      "loss": 1.2175,
+      "step": 7385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.082140953554099e-06,
+      "loss": 1.062,
+      "step": 7386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.079629664224136e-06,
+      "loss": 1.0198,
+      "step": 7387
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.077118949631078e-06,
+      "loss": 0.9308,
+      "step": 7388
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074608810018651e-06,
+      "loss": 0.9802,
+      "step": 7389
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0720992456305416e-06,
+      "loss": 1.0859,
+      "step": 7390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069590256710366e-06,
+      "loss": 1.0574,
+      "step": 7391
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.067081843501696e-06,
+      "loss": 1.0371,
+      "step": 7392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064574006248048e-06,
+      "loss": 1.0979,
+      "step": 7393
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.062066745192864e-06,
+      "loss": 1.0537,
+      "step": 7394
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0595600605795506e-06,
+      "loss": 1.1581,
+      "step": 7395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.057053952651456e-06,
+      "loss": 1.0417,
+      "step": 7396
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0545484216518606e-06,
+      "loss": 1.1104,
+      "step": 7397
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.052043467824002e-06,
+      "loss": 1.0881,
+      "step": 7398
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0495390914110485e-06,
+      "loss": 1.1174,
+      "step": 7399
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.047035292656125e-06,
+      "loss": 0.9841,
+      "step": 7400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0445320718023e-06,
+      "loss": 1.0059,
+      "step": 7401
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0420294290925675e-06,
+      "loss": 0.9874,
+      "step": 7402
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039527364769887e-06,
+      "loss": 1.1199,
+      "step": 7403
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0370258790771546e-06,
+      "loss": 1.0315,
+      "step": 7404
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0345249722572046e-06,
+      "loss": 1.1829,
+      "step": 7405
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.032024644552824e-06,
+      "loss": 1.0679,
+      "step": 7406
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.029524896206734e-06,
+      "loss": 1.0367,
+      "step": 7407
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.027025727461613e-06,
+      "loss": 1.0596,
+      "step": 7408
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.024527138560069e-06,
+      "loss": 1.093,
+      "step": 7409
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.022029129744658e-06,
+      "loss": 1.0309,
+      "step": 7410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019531701257884e-06,
+      "loss": 1.0366,
+      "step": 7411
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0170348533421945e-06,
+      "loss": 1.047,
+      "step": 7412
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.014538586239973e-06,
+      "loss": 1.1392,
+      "step": 7413
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.012042900193559e-06,
+      "loss": 1.0697,
+      "step": 7414
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009547795445222e-06,
+      "loss": 1.0068,
+      "step": 7415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.007053272237181e-06,
+      "loss": 0.9951,
+      "step": 7416
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.004559330811606e-06,
+      "loss": 1.1365,
+      "step": 7417
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.002065971410594e-06,
+      "loss": 1.2031,
+      "step": 7418
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999573194276199e-06,
+      "loss": 0.9272,
+      "step": 7419
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.997080999650421e-06,
+      "loss": 0.9943,
+      "step": 7420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994589387775186e-06,
+      "loss": 1.0432,
+      "step": 7421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.992098358892384e-06,
+      "loss": 1.0688,
+      "step": 7422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989607913243835e-06,
+      "loss": 1.0186,
+      "step": 7423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9871180510713005e-06,
+      "loss": 1.2834,
+      "step": 7424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984628772616501e-06,
+      "loss": 1.155,
+      "step": 7425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.98214007812108e-06,
+      "loss": 1.0549,
+      "step": 7426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9796519678266425e-06,
+      "loss": 1.0609,
+      "step": 7427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977164441974728e-06,
+      "loss": 1.1382,
+      "step": 7428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.974677500806816e-06,
+      "loss": 1.1232,
+      "step": 7429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972191144564339e-06,
+      "loss": 1.0425,
+      "step": 7430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9697053734886645e-06,
+      "loss": 1.0996,
+      "step": 7431
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.967220187821102e-06,
+      "loss": 1.0402,
+      "step": 7432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964735587802915e-06,
+      "loss": 1.1497,
+      "step": 7433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.962251573675297e-06,
+      "loss": 1.2017,
+      "step": 7434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.959768145679392e-06,
+      "loss": 1.1338,
+      "step": 7435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.957285304056292e-06,
+      "loss": 1.0817,
+      "step": 7436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9548030490470215e-06,
+      "loss": 1.1299,
+      "step": 7437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9523213808925486e-06,
+      "loss": 1.1125,
+      "step": 7438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949840299833796e-06,
+      "loss": 1.0012,
+      "step": 7439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.947359806111613e-06,
+      "loss": 1.0612,
+      "step": 7440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.94487989996681e-06,
+      "loss": 1.0938,
+      "step": 7441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942400581640123e-06,
+      "loss": 0.984,
+      "step": 7442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939921851372243e-06,
+      "loss": 1.0339,
+      "step": 7443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.937443709403801e-06,
+      "loss": 1.0378,
+      "step": 7444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.934966155975369e-06,
+      "loss": 1.1338,
+      "step": 7445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932489191327457e-06,
+      "loss": 0.9402,
+      "step": 7446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.930012815700534e-06,
+      "loss": 1.1355,
+      "step": 7447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.92753702933499e-06,
+      "loss": 1.0782,
+      "step": 7448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925061832471178e-06,
+      "loss": 1.0609,
+      "step": 7449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.922587225349378e-06,
+      "loss": 0.9869,
+      "step": 7450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9201132082098215e-06,
+      "loss": 0.9412,
+      "step": 7451
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.917639781292687e-06,
+      "loss": 1.1606,
+      "step": 7452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.915166944838084e-06,
+      "loss": 1.2039,
+      "step": 7453
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9126946990860655e-06,
+      "loss": 1.0037,
+      "step": 7454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.910223044276643e-06,
+      "loss": 1.1121,
+      "step": 7455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907751980649749e-06,
+      "loss": 1.0725,
+      "step": 7456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.905281508445273e-06,
+      "loss": 1.1007,
+      "step": 7457
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.902811627903048e-06,
+      "loss": 1.0164,
+      "step": 7458
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.900342339262837e-06,
+      "loss": 1.0342,
+      "step": 7459
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8978736427643594e-06,
+      "loss": 1.1079,
+      "step": 7460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895405538647269e-06,
+      "loss": 1.0693,
+      "step": 7461
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.892938027151159e-06,
+      "loss": 1.126,
+      "step": 7462
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890471108515579e-06,
+      "loss": 1.0623,
+      "step": 7463
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.888004782980004e-06,
+      "loss": 1.1313,
+      "step": 7464
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8855390507838615e-06,
+      "loss": 1.0026,
+      "step": 7465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.883073912166527e-06,
+      "loss": 1.0251,
+      "step": 7466
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.880609367367304e-06,
+      "loss": 1.0366,
+      "step": 7467
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.878145416625444e-06,
+      "loss": 1.1168,
+      "step": 7468
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875682060180147e-06,
+      "loss": 1.0657,
+      "step": 7469
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.873219298270545e-06,
+      "loss": 1.1182,
+      "step": 7470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.870757131135724e-06,
+      "loss": 1.1418,
+      "step": 7471
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.868295559014699e-06,
+      "loss": 1.0563,
+      "step": 7472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.865834582146437e-06,
+      "loss": 1.115,
+      "step": 7473
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.863374200769853e-06,
+      "loss": 1.0352,
+      "step": 7474
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8609144151237795e-06,
+      "loss": 1.0844,
+      "step": 7475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858455225447017e-06,
+      "loss": 1.0671,
+      "step": 7476
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855996631978298e-06,
+      "loss": 1.0508,
+      "step": 7477
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.853538634956294e-06,
+      "loss": 1.0959,
+      "step": 7478
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851081234619627e-06,
+      "loss": 1.0057,
+      "step": 7479
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848624431206851e-06,
+      "loss": 1.1138,
+      "step": 7480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846168224956469e-06,
+      "loss": 1.0428,
+      "step": 7481
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.843712616106931e-06,
+      "loss": 1.158,
+      "step": 7482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.841257604896611e-06,
+      "loss": 1.1985,
+      "step": 7483
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.838803191563842e-06,
+      "loss": 1.1096,
+      "step": 7484
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836349376346894e-06,
+      "loss": 0.9589,
+      "step": 7485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.833896159483974e-06,
+      "loss": 1.1135,
+      "step": 7486
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.831443541213241e-06,
+      "loss": 1.0076,
+      "step": 7487
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.828991521772785e-06,
+      "loss": 1.043,
+      "step": 7488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.826540101400647e-06,
+      "loss": 1.125,
+      "step": 7489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.824089280334802e-06,
+      "loss": 1.0475,
+      "step": 7490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.821639058813169e-06,
+      "loss": 1.0808,
+      "step": 7491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.819189437073613e-06,
+      "loss": 0.9673,
+      "step": 7492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816740415353942e-06,
+      "loss": 1.116,
+      "step": 7493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8142919938918945e-06,
+      "loss": 1.1058,
+      "step": 7494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8118441729251643e-06,
+      "loss": 1.2627,
+      "step": 7495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809396952691378e-06,
+      "loss": 1.0898,
+      "step": 7496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806950333428102e-06,
+      "loss": 1.037,
+      "step": 7497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8045043153728565e-06,
+      "loss": 1.1431,
+      "step": 7498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802058898763089e-06,
+      "loss": 0.995,
+      "step": 7499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7996140838361994e-06,
+      "loss": 1.0608,
+      "step": 7500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797169870829527e-06,
+      "loss": 0.9147,
+      "step": 7501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794726259980345e-06,
+      "loss": 1.0084,
+      "step": 7502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79228325152588e-06,
+      "loss": 1.036,
+      "step": 7503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.789840845703291e-06,
+      "loss": 1.0559,
+      "step": 7504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7873990427496776e-06,
+      "loss": 1.0876,
+      "step": 7505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.784957842902094e-06,
+      "loss": 1.1893,
+      "step": 7506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.782517246397516e-06,
+      "loss": 1.0901,
+      "step": 7507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.780077253472878e-06,
+      "loss": 1.1013,
+      "step": 7508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7776378643650515e-06,
+      "loss": 1.1177,
+      "step": 7509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7751990793108395e-06,
+      "loss": 1.0859,
+      "step": 7510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7727608985470032e-06,
+      "loss": 1.1584,
+      "step": 7511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7703233223102307e-06,
+      "loss": 1.1047,
+      "step": 7512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7678863508371534e-06,
+      "loss": 1.1421,
+      "step": 7513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7654499843643555e-06,
+      "loss": 0.9863,
+      "step": 7514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7630142231283452e-06,
+      "loss": 1.126,
+      "step": 7515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7605790673655862e-06,
+      "loss": 1.0747,
+      "step": 7516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.758144517312482e-06,
+      "loss": 1.019,
+      "step": 7517
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.755710573205368e-06,
+      "loss": 1.1423,
+      "step": 7518
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7532772352805236e-06,
+      "loss": 1.1113,
+      "step": 7519
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7508445037741804e-06,
+      "loss": 1.0334,
+      "step": 7520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7484123789224945e-06,
+      "loss": 1.1191,
+      "step": 7521
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7459808609615753e-06,
+      "loss": 1.0857,
+      "step": 7522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.743549950127473e-06,
+      "loss": 1.072,
+      "step": 7523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.741119646656167e-06,
+      "loss": 1.1045,
+      "step": 7524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.738689950783595e-06,
+      "loss": 1.0361,
+      "step": 7525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.736260862745621e-06,
+      "loss": 1.0911,
+      "step": 7526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.733832382778054e-06,
+      "loss": 0.9235,
+      "step": 7527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.731404511116652e-06,
+      "loss": 1.0278,
+      "step": 7528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7289772479971e-06,
+      "loss": 1.1675,
+      "step": 7529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7265505936550363e-06,
+      "loss": 1.0718,
+      "step": 7530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7241245483260367e-06,
+      "loss": 1.1467,
+      "step": 7531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7216991122456125e-06,
+      "loss": 1.1462,
+      "step": 7532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.719274285649225e-06,
+      "loss": 1.1804,
+      "step": 7533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7168500687722663e-06,
+      "loss": 1.0581,
+      "step": 7534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7144264618500735e-06,
+      "loss": 1.0979,
+      "step": 7535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7120034651179315e-06,
+      "loss": 1.0486,
+      "step": 7536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7095810788110508e-06,
+      "loss": 1.1262,
+      "step": 7537
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.707159303164597e-06,
+      "loss": 1.1597,
+      "step": 7538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.704738138413674e-06,
+      "loss": 1.074,
+      "step": 7539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.702317584793319e-06,
+      "loss": 0.9633,
+      "step": 7540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6998976425385114e-06,
+      "loss": 1.064,
+      "step": 7541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.697478311884181e-06,
+      "loss": 1.0752,
+      "step": 7542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6950595930651833e-06,
+      "loss": 1.05,
+      "step": 7543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6926414863163308e-06,
+      "loss": 1.0693,
+      "step": 7544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6902239918723602e-06,
+      "loss": 1.1394,
+      "step": 7545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6878071099679604e-06,
+      "loss": 1.1152,
+      "step": 7546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6853908408377615e-06,
+      "loss": 1.0023,
+      "step": 7547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6829751847163242e-06,
+      "loss": 0.937,
+      "step": 7548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.680560141838153e-06,
+      "loss": 1.0188,
+      "step": 7549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.678145712437703e-06,
+      "loss": 1.0049,
+      "step": 7550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6757318967493526e-06,
+      "loss": 1.1699,
+      "step": 7551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6733186950074397e-06,
+      "loss": 0.9948,
+      "step": 7552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6709061074462226e-06,
+      "loss": 1.1204,
+      "step": 7553
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.668494134299916e-06,
+      "loss": 1.1191,
+      "step": 7554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6660827758026753e-06,
+      "loss": 1.0962,
+      "step": 7555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6636720321885766e-06,
+      "loss": 1.1987,
+      "step": 7556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.661261903691656e-06,
+      "loss": 1.1007,
+      "step": 7557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6588523905458874e-06,
+      "loss": 1.0608,
+      "step": 7558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6564434929851735e-06,
+      "loss": 1.0142,
+      "step": 7559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.654035211243373e-06,
+      "loss": 1.0574,
+      "step": 7560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6516275455542693e-06,
+      "loss": 0.9873,
+      "step": 7561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.649220496151601e-06,
+      "loss": 1.1763,
+      "step": 7562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.646814063269034e-06,
+      "loss": 1.0667,
+      "step": 7563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6444082471401786e-06,
+      "loss": 1.0952,
+      "step": 7564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.642003047998588e-06,
+      "loss": 1.1266,
+      "step": 7565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6395984660777585e-06,
+      "loss": 1.04,
+      "step": 7566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6371945016111143e-06,
+      "loss": 1.002,
+      "step": 7567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6347911548320335e-06,
+      "loss": 1.1252,
+      "step": 7568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6323884259738206e-06,
+      "loss": 1.165,
+      "step": 7569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6299863152697368e-06,
+      "loss": 1.072,
+      "step": 7570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6275848229529676e-06,
+      "loss": 1.0647,
+      "step": 7571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.625183949256643e-06,
+      "loss": 0.9674,
+      "step": 7572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6227836944138372e-06,
+      "loss": 1.1909,
+      "step": 7573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6203840586575657e-06,
+      "loss": 0.9731,
+      "step": 7574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.617985042220773e-06,
+      "loss": 1.0734,
+      "step": 7575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6155866453363575e-06,
+      "loss": 1.1616,
+      "step": 7576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6131888682371475e-06,
+      "loss": 1.051,
+      "step": 7577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.61079171115591e-06,
+      "loss": 1.105,
+      "step": 7578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.608395174325362e-06,
+      "loss": 0.9968,
+      "step": 7579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.605999257978149e-06,
+      "loss": 1.1101,
+      "step": 7580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6036039623468632e-06,
+      "loss": 1.0748,
+      "step": 7581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6012092876640394e-06,
+      "loss": 1.0637,
+      "step": 7582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.598815234162141e-06,
+      "loss": 1.0786,
+      "step": 7583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.596421802073582e-06,
+      "loss": 1.0361,
+      "step": 7584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5940289916307114e-06,
+      "loss": 1.0798,
+      "step": 7585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.591636803065812e-06,
+      "loss": 1.1194,
+      "step": 7586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5892452366111187e-06,
+      "loss": 1.0791,
+      "step": 7587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5868542924988005e-06,
+      "loss": 1.0066,
+      "step": 7588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.58446397096096e-06,
+      "loss": 0.9897,
+      "step": 7589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5820742722296487e-06,
+      "loss": 1.04,
+      "step": 7590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5796851965368506e-06,
+      "loss": 1.1501,
+      "step": 7591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5772967441144956e-06,
+      "loss": 1.0426,
+      "step": 7592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.574908915194447e-06,
+      "loss": 1.1241,
+      "step": 7593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5725217100085075e-06,
+      "loss": 1.1088,
+      "step": 7594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5701351287884258e-06,
+      "loss": 1.0908,
+      "step": 7595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5677491717658875e-06,
+      "loss": 0.9966,
+      "step": 7596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5653638391725097e-06,
+      "loss": 1.1047,
+      "step": 7597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5629791312398643e-06,
+      "loss": 1.0875,
+      "step": 7598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.560595048199449e-06,
+      "loss": 1.1112,
+      "step": 7599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5582115902827006e-06,
+      "loss": 1.0844,
+      "step": 7600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.555828757721009e-06,
+      "loss": 1.084,
+      "step": 7601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5534465507456874e-06,
+      "loss": 1.1628,
+      "step": 7602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.551064969587997e-06,
+      "loss": 1.1875,
+      "step": 7603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5486840144791435e-06,
+      "loss": 0.9198,
+      "step": 7604
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546303685650254e-06,
+      "loss": 1.0088,
+      "step": 7605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543923983332416e-06,
+      "loss": 1.188,
+      "step": 7606
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.54154490775664e-06,
+      "loss": 1.0399,
+      "step": 7607
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5391664591538787e-06,
+      "loss": 1.0569,
+      "step": 7608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5367886377550354e-06,
+      "loss": 0.962,
+      "step": 7609
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.534411443790935e-06,
+      "loss": 1.0159,
+      "step": 7610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5320348774923553e-06,
+      "loss": 1.03,
+      "step": 7611
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5296589390900105e-06,
+      "loss": 0.972,
+      "step": 7612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.527283628814545e-06,
+      "loss": 1.1252,
+      "step": 7613
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.524908946896556e-06,
+      "loss": 0.9833,
+      "step": 7614
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5225348935665693e-06,
+      "loss": 0.9646,
+      "step": 7615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.52016146905505e-06,
+      "loss": 1.0972,
+      "step": 7616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5177886735924105e-06,
+      "loss": 1.0277,
+      "step": 7617
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.515416507408991e-06,
+      "loss": 1.0667,
+      "step": 7618
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.513044970735079e-06,
+      "loss": 1.0862,
+      "step": 7619
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.510674063800903e-06,
+      "loss": 1.0859,
+      "step": 7620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.50830378683662e-06,
+      "loss": 1.0266,
+      "step": 7621
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.505934140072329e-06,
+      "loss": 0.9487,
+      "step": 7622
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5035651237380787e-06,
+      "loss": 1.1577,
+      "step": 7623
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.501196738063839e-06,
+      "loss": 1.1566,
+      "step": 7624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4988289832795353e-06,
+      "loss": 1.0656,
+      "step": 7625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4964618596150168e-06,
+      "loss": 0.8979,
+      "step": 7626
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4940953673000834e-06,
+      "loss": 1.0726,
+      "step": 7627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4917295065644717e-06,
+      "loss": 0.9922,
+      "step": 7628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.489364277637852e-06,
+      "loss": 1.1465,
+      "step": 7629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.486999680749832e-06,
+      "loss": 1.0498,
+      "step": 7630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.484635716129967e-06,
+      "loss": 1.0635,
+      "step": 7631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.48227238400774e-06,
+      "loss": 1.0294,
+      "step": 7632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4799096846125847e-06,
+      "loss": 1.0173,
+      "step": 7633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.477547618173861e-06,
+      "loss": 1.031,
+      "step": 7634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4751861849208756e-06,
+      "loss": 0.9919,
+      "step": 7635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4728253850828785e-06,
+      "loss": 0.9767,
+      "step": 7636
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4704652188890385e-06,
+      "loss": 0.9614,
+      "step": 7637
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.46810568656848e-06,
+      "loss": 0.9717,
+      "step": 7638
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.465746788350267e-06,
+      "loss": 0.968,
+      "step": 7639
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4633885244633893e-06,
+      "loss": 1.0786,
+      "step": 7640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4610308951367864e-06,
+      "loss": 1.1435,
+      "step": 7641
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4586739005993285e-06,
+      "loss": 1.0469,
+      "step": 7642
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4563175410798323e-06,
+      "loss": 1.1024,
+      "step": 7643
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4539618168070456e-06,
+      "loss": 1.0941,
+      "step": 7644
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.451606728009653e-06,
+      "loss": 1.147,
+      "step": 7645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.449252274916286e-06,
+      "loss": 1.0259,
+      "step": 7646
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.446898457755512e-06,
+      "loss": 1.0325,
+      "step": 7647
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4445452767558284e-06,
+      "loss": 1.106,
+      "step": 7648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.442192732145684e-06,
+      "loss": 1.0779,
+      "step": 7649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.439840824153451e-06,
+      "loss": 0.9547,
+      "step": 7650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.437489553007455e-06,
+      "loss": 1.1511,
+      "step": 7651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4351389189359497e-06,
+      "loss": 1.1455,
+      "step": 7652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.432788922167125e-06,
+      "loss": 0.9835,
+      "step": 7653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.430439562929119e-06,
+      "loss": 0.9656,
+      "step": 7654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4280908414500023e-06,
+      "loss": 1.0724,
+      "step": 7655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4257427579577805e-06,
+      "loss": 1.0842,
+      "step": 7656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.423395312680402e-06,
+      "loss": 1.083,
+      "step": 7657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4210485058457586e-06,
+      "loss": 0.939,
+      "step": 7658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.418702337681662e-06,
+      "loss": 1.051,
+      "step": 7659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.416356808415877e-06,
+      "loss": 1.1934,
+      "step": 7660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4140119182761066e-06,
+      "loss": 1.0393,
+      "step": 7661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.411667667489982e-06,
+      "loss": 0.9226,
+      "step": 7662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4093240562850847e-06,
+      "loss": 0.9902,
+      "step": 7663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4069810848889197e-06,
+      "loss": 1.042,
+      "step": 7664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.404638753528945e-06,
+      "loss": 1.0825,
+      "step": 7665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.402297062432547e-06,
+      "loss": 1.165,
+      "step": 7666
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3999560118270458e-06,
+      "loss": 1.1165,
+      "step": 7667
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.397615601939711e-06,
+      "loss": 0.9913,
+      "step": 7668
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.395275832997749e-06,
+      "loss": 0.9948,
+      "step": 7669
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3929367052282904e-06,
+      "loss": 1.0431,
+      "step": 7670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3905982188584195e-06,
+      "loss": 1.0547,
+      "step": 7671
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3882603741151476e-06,
+      "loss": 1.0032,
+      "step": 7672
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3859231712254314e-06,
+      "loss": 0.9822,
+      "step": 7673
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3835866104161596e-06,
+      "loss": 1.0671,
+      "step": 7674
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3812506919141574e-06,
+      "loss": 1.1129,
+      "step": 7675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3789154159461935e-06,
+      "loss": 1.0486,
+      "step": 7676
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.376580782738975e-06,
+      "loss": 1.1416,
+      "step": 7677
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.374246792519137e-06,
+      "loss": 1.022,
+      "step": 7678
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.371913445513263e-06,
+      "loss": 1.0223,
+      "step": 7679
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3695807419478686e-06,
+      "loss": 0.8606,
+      "step": 7680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.367248682049403e-06,
+      "loss": 1.1189,
+      "step": 7681
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.364917266044264e-06,
+      "loss": 0.9586,
+      "step": 7682
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3625864941587748e-06,
+      "loss": 0.9897,
+      "step": 7683
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3602563666192045e-06,
+      "loss": 1.0743,
+      "step": 7684
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.357926883651761e-06,
+      "loss": 1.0825,
+      "step": 7685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3555980454825764e-06,
+      "loss": 0.9611,
+      "step": 7686
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.353269852337739e-06,
+      "loss": 1.2273,
+      "step": 7687
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3509423044432597e-06,
+      "loss": 1.0469,
+      "step": 7688
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3486154020250893e-06,
+      "loss": 1.0454,
+      "step": 7689
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.346289145309125e-06,
+      "loss": 1.052,
+      "step": 7690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3439635345211884e-06,
+      "loss": 0.899,
+      "step": 7691
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.341638569887048e-06,
+      "loss": 1.0903,
+      "step": 7692
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3393142516324084e-06,
+      "loss": 1.0466,
+      "step": 7693
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.336990579982905e-06,
+      "loss": 1.0486,
+      "step": 7694
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3346675551641204e-06,
+      "loss": 1.1228,
+      "step": 7695
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3323451774015658e-06,
+      "loss": 0.9763,
+      "step": 7696
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3300234469206884e-06,
+      "loss": 1.0471,
+      "step": 7697
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.327702363946885e-06,
+      "loss": 0.9542,
+      "step": 7698
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3253819287054735e-06,
+      "loss": 0.9144,
+      "step": 7699
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3230621414217214e-06,
+      "loss": 1.0022,
+      "step": 7700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.320743002320831e-06,
+      "loss": 1.0898,
+      "step": 7701
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3184245116279355e-06,
+      "loss": 1.0959,
+      "step": 7702
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.316106669568108e-06,
+      "loss": 1.0837,
+      "step": 7703
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3137894763663636e-06,
+      "loss": 1.1758,
+      "step": 7704
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.311472932247646e-06,
+      "loss": 1.0051,
+      "step": 7705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3091570374368453e-06,
+      "loss": 1.1069,
+      "step": 7706
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3068417921587783e-06,
+      "loss": 1.1118,
+      "step": 7707
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3045271966382066e-06,
+      "loss": 1.0671,
+      "step": 7708
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3022132510998315e-06,
+      "loss": 1.0924,
+      "step": 7709
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2998999557682766e-06,
+      "loss": 1.1575,
+      "step": 7710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2975873108681147e-06,
+      "loss": 1.0547,
+      "step": 7711
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.295275316623856e-06,
+      "loss": 1.1146,
+      "step": 7712
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2929639732599393e-06,
+      "loss": 1.0728,
+      "step": 7713
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.290653281000751e-06,
+      "loss": 0.9819,
+      "step": 7714
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2883432400705994e-06,
+      "loss": 1.0704,
+      "step": 7715
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.286033850693744e-06,
+      "loss": 0.9937,
+      "step": 7716
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2837251130943814e-06,
+      "loss": 1.0767,
+      "step": 7717
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2814170274966252e-06,
+      "loss": 1.0913,
+      "step": 7718
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2791095941245455e-06,
+      "loss": 0.9434,
+      "step": 7719
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.276802813202148e-06,
+      "loss": 0.989,
+      "step": 7720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.274496684953363e-06,
+      "loss": 1.127,
+      "step": 7721
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.27219120960207e-06,
+      "loss": 1.1001,
+      "step": 7722
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2698863873720733e-06,
+      "loss": 1.056,
+      "step": 7723
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2675822184871286e-06,
+      "loss": 1.1311,
+      "step": 7724
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.26527870317091e-06,
+      "loss": 1.115,
+      "step": 7725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.262975841647047e-06,
+      "loss": 0.885,
+      "step": 7726
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2606736341390886e-06,
+      "loss": 1.1287,
+      "step": 7727
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2583720808705355e-06,
+      "loss": 0.9962,
+      "step": 7728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2560711820648094e-06,
+      "loss": 1.1135,
+      "step": 7729
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.253770937945282e-06,
+      "loss": 1.0505,
+      "step": 7730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2514713487352622e-06,
+      "loss": 1.0411,
+      "step": 7731
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.249172414657975e-06,
+      "loss": 1.0,
+      "step": 7732
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.246874135936603e-06,
+      "loss": 0.9958,
+      "step": 7733
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.244576512794263e-06,
+      "loss": 1.0718,
+      "step": 7734
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2422795454539945e-06,
+      "loss": 1.3101,
+      "step": 7735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.239983234138789e-06,
+      "loss": 1.0906,
+      "step": 7736
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2376875790715623e-06,
+      "loss": 0.9946,
+      "step": 7737
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.235392580475175e-06,
+      "loss": 1.0237,
+      "step": 7738
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.233098238572425e-06,
+      "loss": 1.064,
+      "step": 7739
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.230804553586032e-06,
+      "loss": 1.127,
+      "step": 7740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.228511525738667e-06,
+      "loss": 0.9905,
+      "step": 7741
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2262191552529353e-06,
+      "loss": 1.072,
+      "step": 7742
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2239274423513687e-06,
+      "loss": 1.0493,
+      "step": 7743
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.221636387256449e-06,
+      "loss": 1.0261,
+      "step": 7744
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.21934599019058e-06,
+      "loss": 1.0851,
+      "step": 7745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.217056251376116e-06,
+      "loss": 0.9993,
+      "step": 7746
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.214767171035336e-06,
+      "loss": 1.0603,
+      "step": 7747
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2124787493904543e-06,
+      "loss": 0.9927,
+      "step": 7748
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2101909866636316e-06,
+      "loss": 1.134,
+      "step": 7749
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2079038830769606e-06,
+      "loss": 1.0706,
+      "step": 7750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.205617438852463e-06,
+      "loss": 0.9979,
+      "step": 7751
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2033316542121086e-06,
+      "loss": 1.0044,
+      "step": 7752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2010465293777924e-06,
+      "loss": 1.1128,
+      "step": 7753
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1987620645713455e-06,
+      "loss": 1.1454,
+      "step": 7754
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1964782600145484e-06,
+      "loss": 0.9939,
+      "step": 7755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1941951159290973e-06,
+      "loss": 1.0449,
+      "step": 7756
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.191912632536641e-06,
+      "loss": 1.0229,
+      "step": 7757
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1896308100587613e-06,
+      "loss": 1.0334,
+      "step": 7758
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1873496487169643e-06,
+      "loss": 1.094,
+      "step": 7759
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185069148732709e-06,
+      "loss": 1.0989,
+      "step": 7760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1827893103273768e-06,
+      "loss": 1.0269,
+      "step": 7761
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1805101337222867e-06,
+      "loss": 1.0095,
+      "step": 7762
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.178231619138703e-06,
+      "loss": 1.1224,
+      "step": 7763
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1759537667978122e-06,
+      "loss": 1.063,
+      "step": 7764
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1736765769207468e-06,
+      "loss": 0.9265,
+      "step": 7765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1714000497285736e-06,
+      "loss": 1.0046,
+      "step": 7766
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1691241854422882e-06,
+      "loss": 1.1093,
+      "step": 7767
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1668489842828333e-06,
+      "loss": 1.1233,
+      "step": 7768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.164574446471076e-06,
+      "loss": 1.0083,
+      "step": 7769
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.16230057222782e-06,
+      "loss": 0.9912,
+      "step": 7770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1600273617738165e-06,
+      "loss": 1.1335,
+      "step": 7771
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1577548153297365e-06,
+      "loss": 1.0901,
+      "step": 7772
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1554829331161963e-06,
+      "loss": 1.016,
+      "step": 7773
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.15321171535375e-06,
+      "loss": 1.262,
+      "step": 7774
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1509411622628748e-06,
+      "loss": 1.0591,
+      "step": 7775
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.148671274063999e-06,
+      "loss": 1.101,
+      "step": 7776
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.146402050977474e-06,
+      "loss": 1.0142,
+      "step": 7777
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.144133493223589e-06,
+      "loss": 1.1477,
+      "step": 7778
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141865601022577e-06,
+      "loss": 1.1777,
+      "step": 7779
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1395983745945924e-06,
+      "loss": 1.0085,
+      "step": 7780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1373318141597377e-06,
+      "loss": 1.1606,
+      "step": 7781
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1350659199380475e-06,
+      "loss": 1.0354,
+      "step": 7782
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.132800692149488e-06,
+      "loss": 0.9888,
+      "step": 7783
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.130536131013958e-06,
+      "loss": 1.074,
+      "step": 7784
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.128272236751304e-06,
+      "loss": 1.1631,
+      "step": 7785
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1260090095812935e-06,
+      "loss": 1.0918,
+      "step": 7786
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1237464497236415e-06,
+      "loss": 1.1531,
+      "step": 7787
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.121484557397986e-06,
+      "loss": 0.9329,
+      "step": 7788
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1192233328239107e-06,
+      "loss": 1.0796,
+      "step": 7789
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.116962776220933e-06,
+      "loss": 1.0582,
+      "step": 7790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1147028878084993e-06,
+      "loss": 1.0859,
+      "step": 7791
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.112443667805993e-06,
+      "loss": 0.9843,
+      "step": 7792
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.11018511643274e-06,
+      "loss": 1.1199,
+      "step": 7793
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.107927233907988e-06,
+      "loss": 1.2002,
+      "step": 7794
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1056700204509327e-06,
+      "loss": 1.085,
+      "step": 7795
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.103413476280702e-06,
+      "loss": 1.0439,
+      "step": 7796
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1011576016163493e-06,
+      "loss": 1.0613,
+      "step": 7797
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.098902396676876e-06,
+      "loss": 0.9432,
+      "step": 7798
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0966478616812123e-06,
+      "loss": 1.0215,
+      "step": 7799
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0943939968482173e-06,
+      "loss": 1.1423,
+      "step": 7800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.092140802396698e-06,
+      "loss": 1.0745,
+      "step": 7801
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0898882785453866e-06,
+      "loss": 1.026,
+      "step": 7802
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.087636425512952e-06,
+      "loss": 0.9912,
+      "step": 7803
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.085385243518005e-06,
+      "loss": 1.1523,
+      "step": 7804
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0831347327790827e-06,
+      "loss": 1.082,
+      "step": 7805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.080884893514656e-06,
+      "loss": 1.0898,
+      "step": 7806
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.078635725943141e-06,
+      "loss": 0.9932,
+      "step": 7807
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0763872302828746e-06,
+      "loss": 1.071,
+      "step": 7808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0741394067521444e-06,
+      "loss": 1.1104,
+      "step": 7809
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0718922555691555e-06,
+      "loss": 1.1836,
+      "step": 7810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0696457769520627e-06,
+      "loss": 1.0168,
+      "step": 7811
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0673999711189527e-06,
+      "loss": 1.2244,
+      "step": 7812
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0651548382878326e-06,
+      "loss": 1.1121,
+      "step": 7813
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.06291037867666e-06,
+      "loss": 1.0859,
+      "step": 7814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0606665925033274e-06,
+      "loss": 1.064,
+      "step": 7815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0584234799856494e-06,
+      "loss": 1.0315,
+      "step": 7816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0561810413413896e-06,
+      "loss": 1.0554,
+      "step": 7817
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0539392767882305e-06,
+      "loss": 1.1079,
+      "step": 7818
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0516981865438035e-06,
+      "loss": 1.0432,
+      "step": 7819
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0494577708256747e-06,
+      "loss": 1.0142,
+      "step": 7820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0472180298513256e-06,
+      "loss": 0.9896,
+      "step": 7821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.044978963838191e-06,
+      "loss": 1.1335,
+      "step": 7822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0427405730036395e-06,
+      "loss": 1.1376,
+      "step": 7823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.040502857564963e-06,
+      "loss": 1.1179,
+      "step": 7824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0382658177393985e-06,
+      "loss": 1.1689,
+      "step": 7825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0360294537441094e-06,
+      "loss": 1.0884,
+      "step": 7826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0337937657962015e-06,
+      "loss": 1.062,
+      "step": 7827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.031558754112708e-06,
+      "loss": 1.0361,
+      "step": 7828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0293244189105954e-06,
+      "loss": 1.0525,
+      "step": 7829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.027090760406771e-06,
+      "loss": 0.9941,
+      "step": 7830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.024857778818079e-06,
+      "loss": 1.1707,
+      "step": 7831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.022625474361285e-06,
+      "loss": 1.0023,
+      "step": 7832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0203938472531012e-06,
+      "loss": 1.1487,
+      "step": 7833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0181628977101687e-06,
+      "loss": 1.2251,
+      "step": 7834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0159326259490575e-06,
+      "loss": 0.916,
+      "step": 7835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.013703032186286e-06,
+      "loss": 0.9081,
+      "step": 7836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0114741166382897e-06,
+      "loss": 1.1057,
+      "step": 7837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.009245879521453e-06,
+      "loss": 1.0269,
+      "step": 7838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0070183210520896e-06,
+      "loss": 1.1711,
+      "step": 7839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0047914414464408e-06,
+      "loss": 1.1365,
+      "step": 7840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.002565240920693e-06,
+      "loss": 1.0439,
+      "step": 7841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.000339719690958e-06,
+      "loss": 1.1121,
+      "step": 7842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9981148779732817e-06,
+      "loss": 1.0662,
+      "step": 7843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9958907159836524e-06,
+      "loss": 1.0857,
+      "step": 7844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9936672339379825e-06,
+      "loss": 1.2297,
+      "step": 7845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.991444432052124e-06,
+      "loss": 1.0161,
+      "step": 7846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9892223105418672e-06,
+      "loss": 1.0876,
+      "step": 7847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.987000869622921e-06,
+      "loss": 1.0723,
+      "step": 7848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9847801095109484e-06,
+      "loss": 0.9926,
+      "step": 7849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.982560030421532e-06,
+      "loss": 1.1072,
+      "step": 7850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9803406325701867e-06,
+      "loss": 1.1017,
+      "step": 7851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9781219161723752e-06,
+      "loss": 0.9786,
+      "step": 7852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9759038814434795e-06,
+      "loss": 0.9673,
+      "step": 7853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9736865285988238e-06,
+      "loss": 0.9747,
+      "step": 7854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9714698578536683e-06,
+      "loss": 1.063,
+      "step": 7855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9692538694231988e-06,
+      "loss": 0.9177,
+      "step": 7856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9670385635225363e-06,
+      "loss": 1.1395,
+      "step": 7857
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9648239403667432e-06,
+      "loss": 1.0891,
+      "step": 7858
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9626100001708047e-06,
+      "loss": 1.1655,
+      "step": 7859
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9603967431496485e-06,
+      "loss": 0.9293,
+      "step": 7860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9581841695181355e-06,
+      "loss": 1.0613,
+      "step": 7861
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9559722794910526e-06,
+      "loss": 1.075,
+      "step": 7862
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9537610732831302e-06,
+      "loss": 1.0559,
+      "step": 7863
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9515505511090247e-06,
+      "loss": 1.0536,
+      "step": 7864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9493407131833264e-06,
+      "loss": 1.1289,
+      "step": 7865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9471315597205675e-06,
+      "loss": 1.054,
+      "step": 7866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9449230909352024e-06,
+      "loss": 1.1602,
+      "step": 7867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9427153070416257e-06,
+      "loss": 0.991,
+      "step": 7868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9405082082541694e-06,
+      "loss": 1.0624,
+      "step": 7869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.938301794787086e-06,
+      "loss": 0.9877,
+      "step": 7870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9360960668545767e-06,
+      "loss": 1.1489,
+      "step": 7871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9338910246707663e-06,
+      "loss": 1.0569,
+      "step": 7872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9316866684497115e-06,
+      "loss": 1.0779,
+      "step": 7873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.929482998405413e-06,
+      "loss": 1.0764,
+      "step": 7874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.927280014751792e-06,
+      "loss": 1.0042,
+      "step": 7875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9250777177027135e-06,
+      "loss": 1.1338,
+      "step": 7876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.922876107471974e-06,
+      "loss": 1.0645,
+      "step": 7877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9206751842732973e-06,
+      "loss": 1.1086,
+      "step": 7878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.918474948320342e-06,
+      "loss": 0.9918,
+      "step": 7879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.916275399826709e-06,
+      "loss": 1.0743,
+      "step": 7880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.914076539005919e-06,
+      "loss": 1.0928,
+      "step": 7881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.911878366071439e-06,
+      "loss": 1.0551,
+      "step": 7882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.909680881236656e-06,
+      "loss": 1.0518,
+      "step": 7883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.907484084714902e-06,
+      "loss": 0.863,
+      "step": 7884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9052879767194397e-06,
+      "loss": 1.1055,
+      "step": 7885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.903092557463458e-06,
+      "loss": 0.9751,
+      "step": 7886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9008978271600828e-06,
+      "loss": 1.0936,
+      "step": 7887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8987037860223776e-06,
+      "loss": 1.0928,
+      "step": 7888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.896510434263331e-06,
+      "loss": 1.0166,
+      "step": 7889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.894317772095875e-06,
+      "loss": 1.0315,
+      "step": 7890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.89212579973286e-06,
+      "loss": 1.0112,
+      "step": 7891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.889934517387084e-06,
+      "loss": 1.042,
+      "step": 7892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.887743925271277e-06,
+      "loss": 1.0531,
+      "step": 7893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8855540235980838e-06,
+      "loss": 1.0239,
+      "step": 7894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.883364812580103e-06,
+      "loss": 1.0854,
+      "step": 7895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8811762924298616e-06,
+      "loss": 1.0088,
+      "step": 7896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8789884633598086e-06,
+      "loss": 1.0358,
+      "step": 7897
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8768013255823413e-06,
+      "loss": 0.9712,
+      "step": 7898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874614879309775e-06,
+      "loss": 1.1045,
+      "step": 7899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8724291247543735e-06,
+      "loss": 1.0404,
+      "step": 7900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8702440621283213e-06,
+      "loss": 1.0571,
+      "step": 7901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8680596916437354e-06,
+      "loss": 1.0748,
+      "step": 7902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8658760135126727e-06,
+      "loss": 1.0718,
+      "step": 7903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.863693027947125e-06,
+      "loss": 0.9884,
+      "step": 7904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8615107351590056e-06,
+      "loss": 1.0234,
+      "step": 7905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8593291353601705e-06,
+      "loss": 1.2163,
+      "step": 7906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.857148228762401e-06,
+      "loss": 1.082,
+      "step": 7907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8549680155774206e-06,
+      "loss": 1.0713,
+      "step": 7908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.852788496016875e-06,
+      "loss": 1.1897,
+      "step": 7909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.850609670292347e-06,
+      "loss": 1.0317,
+      "step": 7910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.848431538615353e-06,
+      "loss": 1.0269,
+      "step": 7911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8462541011973456e-06,
+      "loss": 0.996,
+      "step": 7912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8440773582496996e-06,
+      "loss": 0.979,
+      "step": 7913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8419013099837333e-06,
+      "loss": 0.9084,
+      "step": 7914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8397259566106915e-06,
+      "loss": 0.9913,
+      "step": 7915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8375512983417497e-06,
+      "loss": 0.9978,
+      "step": 7916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.835377335388024e-06,
+      "loss": 0.9934,
+      "step": 7917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8332040679605543e-06,
+      "loss": 1.0658,
+      "step": 7918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8310314962703166e-06,
+      "loss": 0.9739,
+      "step": 7919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8288596205282248e-06,
+      "loss": 0.9088,
+      "step": 7920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8266884409451134e-06,
+      "loss": 1.0903,
+      "step": 7921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8245179577317607e-06,
+      "loss": 1.0696,
+      "step": 7922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.822348171098872e-06,
+      "loss": 1.04,
+      "step": 7923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8201790812570795e-06,
+      "loss": 0.9552,
+      "step": 7924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.81801068841696e-06,
+      "loss": 1.0885,
+      "step": 7925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8158429927890174e-06,
+      "loss": 1.0854,
+      "step": 7926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8136759945836813e-06,
+      "loss": 1.1372,
+      "step": 7927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.811509694011325e-06,
+      "loss": 0.9684,
+      "step": 7928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8093440912822436e-06,
+      "loss": 1.0867,
+      "step": 7929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8071791866066743e-06,
+      "loss": 1.0432,
+      "step": 7930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8050149801947777e-06,
+      "loss": 1.1216,
+      "step": 7931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8028514722566493e-06,
+      "loss": 1.0544,
+      "step": 7932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8006886630023177e-06,
+      "loss": 1.1365,
+      "step": 7933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7985265526417504e-06,
+      "loss": 1.0204,
+      "step": 7934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7963651413848326e-06,
+      "loss": 1.1985,
+      "step": 7935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.794204429441395e-06,
+      "loss": 1.0757,
+      "step": 7936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7920444170211925e-06,
+      "loss": 0.941,
+      "step": 7937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7898851043339126e-06,
+      "loss": 1.2932,
+      "step": 7938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.787726491589181e-06,
+      "loss": 1.0496,
+      "step": 7939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.785568578996546e-06,
+      "loss": 1.026,
+      "step": 7940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.783411366765495e-06,
+      "loss": 1.0457,
+      "step": 7941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7812548551054507e-06,
+      "loss": 1.0037,
+      "step": 7942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.779099044225756e-06,
+      "loss": 1.0833,
+      "step": 7943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7769439343356973e-06,
+      "loss": 1.0908,
+      "step": 7944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.774789525644486e-06,
+      "loss": 1.0117,
+      "step": 7945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7726358183612634e-06,
+      "loss": 1.0792,
+      "step": 7946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7704828126951145e-06,
+      "loss": 1.1172,
+      "step": 7947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7683305088550416e-06,
+      "loss": 1.0759,
+      "step": 7948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.766178907049989e-06,
+      "loss": 1.1459,
+      "step": 7949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7640280074888314e-06,
+      "loss": 0.9983,
+      "step": 7950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.761877810380369e-06,
+      "loss": 1.0823,
+      "step": 7951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7597283159333443e-06,
+      "loss": 1.0828,
+      "step": 7952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7575795243564216e-06,
+      "loss": 1.0989,
+      "step": 7953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.755431435858199e-06,
+      "loss": 1.0852,
+      "step": 7954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7532840506472136e-06,
+      "loss": 1.0066,
+      "step": 7955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7511373689319233e-06,
+      "loss": 1.0066,
+      "step": 7956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7489913909207254e-06,
+      "loss": 1.0273,
+      "step": 7957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.746846116821953e-06,
+      "loss": 1.1379,
+      "step": 7958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7447015468438575e-06,
+      "loss": 1.0919,
+      "step": 7959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.74255768119463e-06,
+      "loss": 1.0581,
+      "step": 7960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.740414520082395e-06,
+      "loss": 1.0104,
+      "step": 7961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7382720637152025e-06,
+      "loss": 1.1104,
+      "step": 7962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.736130312301042e-06,
+      "loss": 1.167,
+      "step": 7963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.733989266047825e-06,
+      "loss": 0.9819,
+      "step": 7964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.731848925163403e-06,
+      "loss": 1.1707,
+      "step": 7965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7297092898555588e-06,
+      "loss": 0.9883,
+      "step": 7966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7275703603319992e-06,
+      "loss": 1.0781,
+      "step": 7967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7254321368003643e-06,
+      "loss": 0.9393,
+      "step": 7968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.723294619468235e-06,
+      "loss": 1.0261,
+      "step": 7969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7211578085431113e-06,
+      "loss": 1.0232,
+      "step": 7970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.719021704232436e-06,
+      "loss": 0.8733,
+      "step": 7971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7168863067435703e-06,
+      "loss": 1.1248,
+      "step": 7972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7147516162838184e-06,
+      "loss": 1.1162,
+      "step": 7973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.712617633060416e-06,
+      "loss": 1.0168,
+      "step": 7974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.710484357280515e-06,
+      "loss": 1.033,
+      "step": 7975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.708351789151216e-06,
+      "loss": 0.9297,
+      "step": 7976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7062199288795445e-06,
+      "loss": 0.8949,
+      "step": 7977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.704088776672453e-06,
+      "loss": 1.0125,
+      "step": 7978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.701958332736835e-06,
+      "loss": 0.917,
+      "step": 7979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6998285972795025e-06,
+      "loss": 1.0508,
+      "step": 7980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.697699570507213e-06,
+      "loss": 1.0286,
+      "step": 7981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.695571252626643e-06,
+      "loss": 1.0199,
+      "step": 7982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6934436438444045e-06,
+      "loss": 1.0569,
+      "step": 7983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.691316744367042e-06,
+      "loss": 1.0237,
+      "step": 7984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6891905544010344e-06,
+      "loss": 1.1167,
+      "step": 7985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6870650741527817e-06,
+      "loss": 0.8579,
+      "step": 7986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.684940303828627e-06,
+      "loss": 1.1653,
+      "step": 7987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6828162436348314e-06,
+      "loss": 0.9862,
+      "step": 7988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.680692893777602e-06,
+      "loss": 1.0154,
+      "step": 7989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6785702544630644e-06,
+      "loss": 1.1388,
+      "step": 7990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6764483258972783e-06,
+      "loss": 1.0427,
+      "step": 7991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6743271082862375e-06,
+      "loss": 1.0735,
+      "step": 7992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6722066018358696e-06,
+      "loss": 1.1799,
+      "step": 7993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6700868067520215e-06,
+      "loss": 1.1831,
+      "step": 7994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.667967723240482e-06,
+      "loss": 1.1537,
+      "step": 7995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6658493515069737e-06,
+      "loss": 1.0825,
+      "step": 7996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.663731691757132e-06,
+      "loss": 1.0765,
+      "step": 7997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.661614744196539e-06,
+      "loss": 1.0288,
+      "step": 7998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.659498509030708e-06,
+      "loss": 1.0454,
+      "step": 7999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.657382986465071e-06,
+      "loss": 0.9476,
+      "step": 8000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.655268176705007e-06,
+      "loss": 1.0505,
+      "step": 8001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6531540799558087e-06,
+      "loss": 1.1047,
+      "step": 8002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6510406964227155e-06,
+      "loss": 1.0784,
+      "step": 8003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6489280263108865e-06,
+      "loss": 1.0591,
+      "step": 8004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6468160698254132e-06,
+      "loss": 0.9595,
+      "step": 8005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6447048271713206e-06,
+      "loss": 1.0564,
+      "step": 8006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6425942985535693e-06,
+      "loss": 0.9628,
+      "step": 8007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6404844841770362e-06,
+      "loss": 1.0483,
+      "step": 8008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6383753842465464e-06,
+      "loss": 0.9518,
+      "step": 8009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.636266998966839e-06,
+      "loss": 1.0382,
+      "step": 8010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6341593285425983e-06,
+      "loss": 1.1294,
+      "step": 8011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.632052373178429e-06,
+      "loss": 1.0688,
+      "step": 8012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6299461330788655e-06,
+      "loss": 1.0149,
+      "step": 8013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6278406084483832e-06,
+      "loss": 0.9879,
+      "step": 8014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6257357994913834e-06,
+      "loss": 1.0431,
+      "step": 8015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6236317064121897e-06,
+      "loss": 0.9708,
+      "step": 8016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6215283294150694e-06,
+      "loss": 0.9653,
+      "step": 8017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.619425668704212e-06,
+      "loss": 1.1511,
+      "step": 8018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6173237244837346e-06,
+      "loss": 1.0107,
+      "step": 8019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6152224969576955e-06,
+      "loss": 1.1501,
+      "step": 8020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6131219863300726e-06,
+      "loss": 1.1746,
+      "step": 8021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6110221928047818e-06,
+      "loss": 1.0647,
+      "step": 8022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.60892311658567e-06,
+      "loss": 1.1392,
+      "step": 8023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6068247578765026e-06,
+      "loss": 1.0619,
+      "step": 8024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.604727116880993e-06,
+      "loss": 1.0918,
+      "step": 8025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.60263019380277e-06,
+      "loss": 1.1091,
+      "step": 8026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6005339888453973e-06,
+      "loss": 0.9126,
+      "step": 8027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598438502212375e-06,
+      "loss": 1.0562,
+      "step": 8028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5963437341071218e-06,
+      "loss": 1.1602,
+      "step": 8029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.594249684732998e-06,
+      "loss": 1.042,
+      "step": 8030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5921563542932913e-06,
+      "loss": 1.1089,
+      "step": 8031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5900637429912136e-06,
+      "loss": 1.0388,
+      "step": 8032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5879718510299135e-06,
+      "loss": 1.1039,
+      "step": 8033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.585880678612468e-06,
+      "loss": 1.0032,
+      "step": 8034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.583790225941879e-06,
+      "loss": 1.1707,
+      "step": 8035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.581700493221089e-06,
+      "loss": 0.9779,
+      "step": 8036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5796114806529595e-06,
+      "loss": 0.9749,
+      "step": 8037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5775231884402896e-06,
+      "loss": 1.0996,
+      "step": 8038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.57543561678581e-06,
+      "loss": 1.0563,
+      "step": 8039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.573348765892174e-06,
+      "loss": 1.1062,
+      "step": 8040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5712626359619664e-06,
+      "loss": 1.1387,
+      "step": 8041
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.56917722719771e-06,
+      "loss": 1.0842,
+      "step": 8042
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5670925398018455e-06,
+      "loss": 1.0442,
+      "step": 8043
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.565008573976755e-06,
+      "loss": 0.993,
+      "step": 8044
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5629253299247414e-06,
+      "loss": 0.9702,
+      "step": 8045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.560842807848044e-06,
+      "loss": 1.0605,
+      "step": 8046
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.558761007948836e-06,
+      "loss": 1.0562,
+      "step": 8047
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5566799304292e-06,
+      "loss": 0.9359,
+      "step": 8048
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55459957549117e-06,
+      "loss": 1.1427,
+      "step": 8049
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5525199433367055e-06,
+      "loss": 1.0428,
+      "step": 8050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5504410341676864e-06,
+      "loss": 1.0972,
+      "step": 8051
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5483628481859355e-06,
+      "loss": 1.0946,
+      "step": 8052
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5462853855931923e-06,
+      "loss": 1.0524,
+      "step": 8053
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.544208646591134e-06,
+      "loss": 1.0579,
+      "step": 8054
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5421326313813745e-06,
+      "loss": 1.1006,
+      "step": 8055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5400573401654347e-06,
+      "loss": 1.0823,
+      "step": 8056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.537982773144787e-06,
+      "loss": 1.014,
+      "step": 8057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5359089305208284e-06,
+      "loss": 1.0881,
+      "step": 8058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.533835812494877e-06,
+      "loss": 1.1221,
+      "step": 8059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.531763419268192e-06,
+      "loss": 1.116,
+      "step": 8060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5296917510419518e-06,
+      "loss": 1.0713,
+      "step": 8061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.527620808017276e-06,
+      "loss": 1.0381,
+      "step": 8062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5255505903951995e-06,
+      "loss": 1.0714,
+      "step": 8063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.523481098376702e-06,
+      "loss": 1.1323,
+      "step": 8064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5214123321626794e-06,
+      "loss": 0.9673,
+      "step": 8065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.519344291953969e-06,
+      "loss": 0.9661,
+      "step": 8066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.517276977951325e-06,
+      "loss": 1.1763,
+      "step": 8067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5152103903554424e-06,
+      "loss": 1.0552,
+      "step": 8068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.513144529366942e-06,
+      "loss": 1.0342,
+      "step": 8069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.511079395186372e-06,
+      "loss": 1.0468,
+      "step": 8070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5090149880142066e-06,
+      "loss": 0.868,
+      "step": 8071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5069513080508612e-06,
+      "loss": 1.1311,
+      "step": 8072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5048883554966672e-06,
+      "loss": 1.0103,
+      "step": 8073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5028261305518976e-06,
+      "loss": 1.1284,
+      "step": 8074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5007646334167425e-06,
+      "loss": 1.2124,
+      "step": 8075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.498703864291331e-06,
+      "loss": 1.0177,
+      "step": 8076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496643823375724e-06,
+      "loss": 1.0712,
+      "step": 8077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.494584510869894e-06,
+      "loss": 1.0371,
+      "step": 8078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.492525926973761e-06,
+      "loss": 0.9554,
+      "step": 8079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.49046807188717e-06,
+      "loss": 1.0786,
+      "step": 8080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.488410945809887e-06,
+      "loss": 1.0565,
+      "step": 8081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4863545489416206e-06,
+      "loss": 1.0767,
+      "step": 8082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4842988814819958e-06,
+      "loss": 1.0181,
+      "step": 8083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4822439436305766e-06,
+      "loss": 1.1292,
+      "step": 8084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.48018973558685e-06,
+      "loss": 1.0605,
+      "step": 8085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.478136257550231e-06,
+      "loss": 0.9023,
+      "step": 8086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4760835097200696e-06,
+      "loss": 1.0547,
+      "step": 8087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4740314922956452e-06,
+      "loss": 1.0735,
+      "step": 8088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4719802054761578e-06,
+      "loss": 1.0769,
+      "step": 8089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4699296494607484e-06,
+      "loss": 0.9568,
+      "step": 8090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4678798244484724e-06,
+      "loss": 1.0986,
+      "step": 8091
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4658307306383313e-06,
+      "loss": 1.0259,
+      "step": 8092
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4637823682292406e-06,
+      "loss": 1.1152,
+      "step": 8093
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4617347374200506e-06,
+      "loss": 0.8812,
+      "step": 8094
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4596878384095435e-06,
+      "loss": 1.0725,
+      "step": 8095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4576416713964293e-06,
+      "loss": 1.0211,
+      "step": 8096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.455596236579342e-06,
+      "loss": 1.1772,
+      "step": 8097
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4535515341568504e-06,
+      "loss": 1.0684,
+      "step": 8098
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.451507564327451e-06,
+      "loss": 0.9078,
+      "step": 8099
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4494643272895623e-06,
+      "loss": 1.0458,
+      "step": 8100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4474218232415434e-06,
+      "loss": 1.125,
+      "step": 8101
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.445380052381673e-06,
+      "loss": 1.0752,
+      "step": 8102
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4433390149081614e-06,
+      "loss": 1.0916,
+      "step": 8103
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.441298711019152e-06,
+      "loss": 0.9625,
+      "step": 8104
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4392591409127086e-06,
+      "loss": 1.0237,
+      "step": 8105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4372203047868336e-06,
+      "loss": 1.0564,
+      "step": 8106
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.43518220283945e-06,
+      "loss": 1.1169,
+      "step": 8107
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4331448352684084e-06,
+      "loss": 0.9945,
+      "step": 8108
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4311082022714993e-06,
+      "loss": 0.96,
+      "step": 8109
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4290723040464272e-06,
+      "loss": 0.9419,
+      "step": 8110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4270371407908377e-06,
+      "loss": 0.984,
+      "step": 8111
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4250027127023014e-06,
+      "loss": 0.9794,
+      "step": 8112
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.422969019978312e-06,
+      "loss": 1.0166,
+      "step": 8113
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4209360628162993e-06,
+      "loss": 0.9998,
+      "step": 8114
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4189038414136177e-06,
+      "loss": 1.0669,
+      "step": 8115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4168723559675487e-06,
+      "loss": 1.0146,
+      "step": 8116
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4148416066753077e-06,
+      "loss": 0.9523,
+      "step": 8117
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.412811593734031e-06,
+      "loss": 1.0216,
+      "step": 8118
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.410782317340792e-06,
+      "loss": 1.0742,
+      "step": 8119
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4087537776925897e-06,
+      "loss": 1.0962,
+      "step": 8120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.406725974986348e-06,
+      "loss": 1.0765,
+      "step": 8121
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.404698909418919e-06,
+      "loss": 1.0315,
+      "step": 8122
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.402672581187091e-06,
+      "loss": 1.0947,
+      "step": 8123
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4006469904875707e-06,
+      "loss": 1.0867,
+      "step": 8124
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.398622137517004e-06,
+      "loss": 1.0038,
+      "step": 8125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.396598022471953e-06,
+      "loss": 1.009,
+      "step": 8126
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3945746455489173e-06,
+      "loss": 0.9935,
+      "step": 8127
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.392552006944325e-06,
+      "loss": 1.0728,
+      "step": 8128
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.390530106854527e-06,
+      "loss": 0.9592,
+      "step": 8129
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3885089454758015e-06,
+      "loss": 1.0935,
+      "step": 8130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3864885230043643e-06,
+      "loss": 1.0837,
+      "step": 8131
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.384468839636349e-06,
+      "loss": 1.0107,
+      "step": 8132
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3824498955678243e-06,
+      "loss": 1.0491,
+      "step": 8133
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3804316909947878e-06,
+      "loss": 1.0586,
+      "step": 8134
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.378414226113156e-06,
+      "loss": 1.0348,
+      "step": 8135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.376397501118788e-06,
+      "loss": 1.1464,
+      "step": 8136
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.374381516207458e-06,
+      "loss": 0.9508,
+      "step": 8137
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.372366271574872e-06,
+      "loss": 0.9767,
+      "step": 8138
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3703517674166697e-06,
+      "loss": 1.0093,
+      "step": 8139
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.368338003928411e-06,
+      "loss": 1.1555,
+      "step": 8140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3663249813055913e-06,
+      "loss": 1.0469,
+      "step": 8141
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3643126997436305e-06,
+      "loss": 1.0715,
+      "step": 8142
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3623011594378753e-06,
+      "loss": 1.147,
+      "step": 8143
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.360290360583599e-06,
+      "loss": 0.9688,
+      "step": 8144
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3582803033760116e-06,
+      "loss": 1.1049,
+      "step": 8145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3562709880102398e-06,
+      "loss": 1.0807,
+      "step": 8146
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3542624146813477e-06,
+      "loss": 1.075,
+      "step": 8147
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352254583584318e-06,
+      "loss": 0.9725,
+      "step": 8148
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3502474949140706e-06,
+      "loss": 1.0541,
+      "step": 8149
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3482411488654555e-06,
+      "loss": 1.0592,
+      "step": 8150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.346235545633231e-06,
+      "loss": 1.1819,
+      "step": 8151
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3442306854121033e-06,
+      "loss": 1.1643,
+      "step": 8152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3422265683967037e-06,
+      "loss": 1.0693,
+      "step": 8153
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.340223194781581e-06,
+      "loss": 1.0773,
+      "step": 8154
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3382205647612242e-06,
+      "loss": 1.1475,
+      "step": 8155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.33621867853004e-06,
+      "loss": 1.1565,
+      "step": 8156
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3342175362823683e-06,
+      "loss": 1.0818,
+      "step": 8157
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3322171382124816e-06,
+      "loss": 1.087,
+      "step": 8158
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3302174845145632e-06,
+      "loss": 0.929,
+      "step": 8159
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3282185753827413e-06,
+      "loss": 1.0159,
+      "step": 8160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3262204110110685e-06,
+      "loss": 1.1536,
+      "step": 8161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3242229915935166e-06,
+      "loss": 1.0854,
+      "step": 8162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.322226317323997e-06,
+      "loss": 1.0574,
+      "step": 8163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.320230388396335e-06,
+      "loss": 0.9554,
+      "step": 8164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3182352050042976e-06,
+      "loss": 1.1477,
+      "step": 8165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.316240767341571e-06,
+      "loss": 1.1323,
+      "step": 8166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3142470756017686e-06,
+      "loss": 1.0785,
+      "step": 8167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.312254129978434e-06,
+      "loss": 1.0554,
+      "step": 8168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3102619306650444e-06,
+      "loss": 1.113,
+      "step": 8169
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.308270477854989e-06,
+      "loss": 1.0454,
+      "step": 8170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3062797717416017e-06,
+      "loss": 1.0424,
+      "step": 8171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3042898125181334e-06,
+      "loss": 0.9683,
+      "step": 8172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.30230060037776e-06,
+      "loss": 0.969,
+      "step": 8173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3003121355135983e-06,
+      "loss": 1.2249,
+      "step": 8174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.298324418118677e-06,
+      "loss": 0.928,
+      "step": 8175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2963374483859623e-06,
+      "loss": 0.991,
+      "step": 8176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.294351226508349e-06,
+      "loss": 1.1638,
+      "step": 8177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.292365752678648e-06,
+      "loss": 0.9811,
+      "step": 8178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.290381027089612e-06,
+      "loss": 1.0884,
+      "step": 8179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2883970499339103e-06,
+      "loss": 1.0798,
+      "step": 8180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2864138214041396e-06,
+      "loss": 1.1053,
+      "step": 8181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.284431341692834e-06,
+      "loss": 1.0559,
+      "step": 8182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2824496109924432e-06,
+      "loss": 1.0542,
+      "step": 8183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2804686294953517e-06,
+      "loss": 0.9071,
+      "step": 8184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.278488397393872e-06,
+      "loss": 1.0232,
+      "step": 8185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2765089148802346e-06,
+      "loss": 1.0815,
+      "step": 8186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.274530182146609e-06,
+      "loss": 1.0656,
+      "step": 8187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2725521993850853e-06,
+      "loss": 0.9281,
+      "step": 8188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.270574966787676e-06,
+      "loss": 1.1089,
+      "step": 8189
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2685984845463347e-06,
+      "loss": 1.0399,
+      "step": 8190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.266622752852927e-06,
+      "loss": 0.9741,
+      "step": 8191
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2646477718992577e-06,
+      "loss": 1.146,
+      "step": 8192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.262673541877054e-06,
+      "loss": 0.9807,
+      "step": 8193
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2607000629779684e-06,
+      "loss": 0.957,
+      "step": 8194
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.258727335393578e-06,
+      "loss": 1.1038,
+      "step": 8195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.256755359315399e-06,
+      "loss": 1.0853,
+      "step": 8196
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2547841349348587e-06,
+      "loss": 1.1702,
+      "step": 8197
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2528136624433227e-06,
+      "loss": 0.9795,
+      "step": 8198
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.250843942032084e-06,
+      "loss": 1.0518,
+      "step": 8199
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.248874973892352e-06,
+      "loss": 0.9834,
+      "step": 8200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.246906758215277e-06,
+      "loss": 0.9943,
+      "step": 8201
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2449392951919248e-06,
+      "loss": 1.0577,
+      "step": 8202
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.242972585013291e-06,
+      "loss": 0.994,
+      "step": 8203
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2410066278703037e-06,
+      "loss": 1.1016,
+      "step": 8204
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2390414239538096e-06,
+      "loss": 1.0784,
+      "step": 8205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2370769734545894e-06,
+      "loss": 1.1584,
+      "step": 8206
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2351132765633488e-06,
+      "loss": 1.0103,
+      "step": 8207
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.233150333470716e-06,
+      "loss": 1.0396,
+      "step": 8208
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2311881443672533e-06,
+      "loss": 1.1147,
+      "step": 8209
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2292267094434438e-06,
+      "loss": 0.9637,
+      "step": 8210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.227266028889696e-06,
+      "loss": 1.0673,
+      "step": 8211
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2253061028963553e-06,
+      "loss": 0.9612,
+      "step": 8212
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2233469316536795e-06,
+      "loss": 1.0615,
+      "step": 8213
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2213885153518655e-06,
+      "loss": 1.0675,
+      "step": 8214
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.219430854181034e-06,
+      "loss": 1.199,
+      "step": 8215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.217473948331229e-06,
+      "loss": 1.0615,
+      "step": 8216
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.215517797992418e-06,
+      "loss": 1.0908,
+      "step": 8217
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2135624033545065e-06,
+      "loss": 1.1499,
+      "step": 8218
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2116077646073153e-06,
+      "loss": 0.8745,
+      "step": 8219
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2096538819406e-06,
+      "loss": 1.0518,
+      "step": 8220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2077007555440356e-06,
+      "loss": 1.0801,
+      "step": 8221
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2057483856072303e-06,
+      "loss": 1.0551,
+      "step": 8222
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2037967723197174e-06,
+      "loss": 1.0883,
+      "step": 8223
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2018459158709536e-06,
+      "loss": 1.0059,
+      "step": 8224
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1998958164503204e-06,
+      "loss": 0.8381,
+      "step": 8225
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.197946474247136e-06,
+      "loss": 1.0771,
+      "step": 8226
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1959978894506305e-06,
+      "loss": 1.1212,
+      "step": 8227
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.194050062249977e-06,
+      "loss": 1.0554,
+      "step": 8228
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1921029928342584e-06,
+      "loss": 1.02,
+      "step": 8229
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.190156681392497e-06,
+      "loss": 1.0272,
+      "step": 8230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.18821112811364e-06,
+      "loss": 1.0122,
+      "step": 8231
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.186266333186546e-06,
+      "loss": 1.1282,
+      "step": 8232
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.18432229680002e-06,
+      "loss": 1.126,
+      "step": 8233
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1823790191427863e-06,
+      "loss": 1.0612,
+      "step": 8234
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1804365004034877e-06,
+      "loss": 1.0289,
+      "step": 8235
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.178494740770706e-06,
+      "loss": 1.0789,
+      "step": 8236
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.176553740432937e-06,
+      "loss": 1.0472,
+      "step": 8237
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1746134995786127e-06,
+      "loss": 0.9945,
+      "step": 8238
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1726740183960916e-06,
+      "loss": 0.9126,
+      "step": 8239
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.170735297073645e-06,
+      "loss": 1.113,
+      "step": 8240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1687973357994852e-06,
+      "loss": 1.0631,
+      "step": 8241
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.166860134761747e-06,
+      "loss": 1.0341,
+      "step": 8242
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.164923694148484e-06,
+      "loss": 1.0876,
+      "step": 8243
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1629880141476876e-06,
+      "loss": 1.0581,
+      "step": 8244
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.161053094947265e-06,
+      "loss": 0.8979,
+      "step": 8245
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1591189367350597e-06,
+      "loss": 1.0051,
+      "step": 8246
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1571855396988306e-06,
+      "loss": 0.9443,
+      "step": 8247
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.155252904026267e-06,
+      "loss": 0.9758,
+      "step": 8248
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.153321029904988e-06,
+      "loss": 1.0857,
+      "step": 8249
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.151389917522536e-06,
+      "loss": 1.0605,
+      "step": 8250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1494595670663764e-06,
+      "loss": 1.1553,
+      "step": 8251
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.147529978723909e-06,
+      "loss": 1.0063,
+      "step": 8252
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1456011526824496e-06,
+      "loss": 0.978,
+      "step": 8253
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1436730891292424e-06,
+      "loss": 1.0089,
+      "step": 8254
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1417457882514657e-06,
+      "loss": 0.8338,
+      "step": 8255
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1398192502362112e-06,
+      "loss": 1.0416,
+      "step": 8256
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1378934752705085e-06,
+      "loss": 0.9238,
+      "step": 8257
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.135968463541307e-06,
+      "loss": 0.9552,
+      "step": 8258
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.134044215235479e-06,
+      "loss": 1.0808,
+      "step": 8259
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1321207305398326e-06,
+      "loss": 1.0176,
+      "step": 8260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.130198009641091e-06,
+      "loss": 1.0432,
+      "step": 8261
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.128276052725907e-06,
+      "loss": 1.0393,
+      "step": 8262
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1263548599808615e-06,
+      "loss": 1.0623,
+      "step": 8263
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1244344315924627e-06,
+      "loss": 0.99,
+      "step": 8264
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.122514767747136e-06,
+      "loss": 1.1523,
+      "step": 8265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1205958686312444e-06,
+      "loss": 1.2034,
+      "step": 8266
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.118677734431065e-06,
+      "loss": 1.1335,
+      "step": 8267
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.116760365332812e-06,
+      "loss": 1.001,
+      "step": 8268
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1148437615226146e-06,
+      "loss": 1.1121,
+      "step": 8269
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.112927923186533e-06,
+      "loss": 1.0964,
+      "step": 8270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1110128505105534e-06,
+      "loss": 1.0376,
+      "step": 8271
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.10909854368059e-06,
+      "loss": 1.0791,
+      "step": 8272
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.107185002882475e-06,
+      "loss": 1.0546,
+      "step": 8273
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1052722283019745e-06,
+      "loss": 1.0229,
+      "step": 8274
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.103360220124776e-06,
+      "loss": 1.0879,
+      "step": 8275
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.101448978536489e-06,
+      "loss": 1.0342,
+      "step": 8276
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.099538503722659e-06,
+      "loss": 1.0967,
+      "step": 8277
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0976287958687445e-06,
+      "loss": 1.0344,
+      "step": 8278
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0957198551601386e-06,
+      "loss": 1.054,
+      "step": 8279
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.093811681782161e-06,
+      "loss": 0.9946,
+      "step": 8280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.091904275920047e-06,
+      "loss": 1.0919,
+      "step": 8281
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.089997637758969e-06,
+      "loss": 1.1073,
+      "step": 8282
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0880917674840173e-06,
+      "loss": 1.0233,
+      "step": 8283
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0861866652802055e-06,
+      "loss": 1.0029,
+      "step": 8284
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0842823313324834e-06,
+      "loss": 0.9908,
+      "step": 8285
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0823787658257132e-06,
+      "loss": 0.97,
+      "step": 8286
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0804759689446928e-06,
+      "loss": 1.0874,
+      "step": 8287
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0785739408741433e-06,
+      "loss": 1.1057,
+      "step": 8288
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0766726817987058e-06,
+      "loss": 1.0193,
+      "step": 8289
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.074772191902955e-06,
+      "loss": 1.1748,
+      "step": 8290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.072872471371383e-06,
+      "loss": 1.032,
+      "step": 8291
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0709735203884085e-06,
+      "loss": 0.9272,
+      "step": 8292
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0690753391383834e-06,
+      "loss": 0.978,
+      "step": 8293
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0671779278055728e-06,
+      "loss": 1.0969,
+      "step": 8294
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0652812865741757e-06,
+      "loss": 1.1072,
+      "step": 8295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0633854156283186e-06,
+      "loss": 1.0723,
+      "step": 8296
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0614903151520447e-06,
+      "loss": 1.1221,
+      "step": 8297
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0595959853293233e-06,
+      "loss": 1.1011,
+      "step": 8298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.057702426344057e-06,
+      "loss": 1.1045,
+      "step": 8299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0558096383800653e-06,
+      "loss": 1.0526,
+      "step": 8300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0539176216210986e-06,
+      "loss": 1.0754,
+      "step": 8301
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0520263762508263e-06,
+      "loss": 0.9055,
+      "step": 8302
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0501359024528476e-06,
+      "loss": 0.8982,
+      "step": 8303
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0482462004106895e-06,
+      "loss": 1.0532,
+      "step": 8304
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0463572703077984e-06,
+      "loss": 1.041,
+      "step": 8305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.044469112327543e-06,
+      "loss": 0.9973,
+      "step": 8306
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0425817266532277e-06,
+      "loss": 1.0283,
+      "step": 8307
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.040695113468071e-06,
+      "loss": 1.054,
+      "step": 8308
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0388092729552258e-06,
+      "loss": 1.043,
+      "step": 8309
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0369242052977613e-06,
+      "loss": 1.0806,
+      "step": 8310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.035039910678677e-06,
+      "loss": 1.0573,
+      "step": 8311
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0331563892809023e-06,
+      "loss": 1.0516,
+      "step": 8312
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.031273641287276e-06,
+      "loss": 1.0774,
+      "step": 8313
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0293916668805747e-06,
+      "loss": 1.1213,
+      "step": 8314
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0275104662434997e-06,
+      "loss": 1.124,
+      "step": 8315
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.025630039558668e-06,
+      "loss": 1.0044,
+      "step": 8316
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.023750387008634e-06,
+      "loss": 1.0238,
+      "step": 8317
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0218715087758657e-06,
+      "loss": 1.1421,
+      "step": 8318
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0199934050427627e-06,
+      "loss": 1.1174,
+      "step": 8319
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0181160759916475e-06,
+      "loss": 1.0227,
+      "step": 8320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.016239521804764e-06,
+      "loss": 1.0632,
+      "step": 8321
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0143637426642857e-06,
+      "loss": 1.0515,
+      "step": 8322
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.012488738752313e-06,
+      "loss": 1.041,
+      "step": 8323
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0106145102508613e-06,
+      "loss": 0.9802,
+      "step": 8324
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.008741057341882e-06,
+      "loss": 1.0142,
+      "step": 8325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.006868380207242e-06,
+      "loss": 1.147,
+      "step": 8326
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.004996479028739e-06,
+      "loss": 0.9849,
+      "step": 8327
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0031253539880935e-06,
+      "loss": 1.1287,
+      "step": 8328
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0012550052669466e-06,
+      "loss": 1.1641,
+      "step": 8329
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999385433046871e-06,
+      "loss": 1.0493,
+      "step": 8330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9975166375093625e-06,
+      "loss": 1.1304,
+      "step": 8331
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.995648618835834e-06,
+      "loss": 1.1372,
+      "step": 8332
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9937813772076353e-06,
+      "loss": 1.0593,
+      "step": 8333
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9919149128060313e-06,
+      "loss": 1.0808,
+      "step": 8334
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9900492258122116e-06,
+      "loss": 1.063,
+      "step": 8335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9881843164072955e-06,
+      "loss": 1.0881,
+      "step": 8336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.986320184772328e-06,
+      "loss": 1.0211,
+      "step": 8337
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9844568310882685e-06,
+      "loss": 1.1184,
+      "step": 8338
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9825942555360134e-06,
+      "loss": 0.9747,
+      "step": 8339
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9807324582963715e-06,
+      "loss": 1.0596,
+      "step": 8340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9788714395500887e-06,
+      "loss": 0.9781,
+      "step": 8341
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977011199477825e-06,
+      "loss": 1.0653,
+      "step": 8342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.975151738260166e-06,
+      "loss": 1.0422,
+      "step": 8343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9732930560776276e-06,
+      "loss": 1.0042,
+      "step": 8344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9714351531106477e-06,
+      "loss": 1.1284,
+      "step": 8345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.969578029539585e-06,
+      "loss": 1.1213,
+      "step": 8346
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9677216855447278e-06,
+      "loss": 1.1426,
+      "step": 8347
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.965866121306281e-06,
+      "loss": 0.9128,
+      "step": 8348
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.964011337004387e-06,
+      "loss": 1.1362,
+      "step": 8349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962157332819098e-06,
+      "loss": 1.1083,
+      "step": 8350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.960304108930395e-06,
+      "loss": 1.0876,
+      "step": 8351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9584516655181896e-06,
+      "loss": 1.0139,
+      "step": 8352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9566000027623134e-06,
+      "loss": 0.9736,
+      "step": 8353
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9547491208425174e-06,
+      "loss": 1.0437,
+      "step": 8354
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9528990199384866e-06,
+      "loss": 1.0511,
+      "step": 8355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9510497002298224e-06,
+      "loss": 1.0864,
+      "step": 8356
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9492011618960505e-06,
+      "loss": 1.0826,
+      "step": 8357
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9473534051166276e-06,
+      "loss": 1.2241,
+      "step": 8358
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9455064300709247e-06,
+      "loss": 0.9274,
+      "step": 8359
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.943660236938245e-06,
+      "loss": 1.0911,
+      "step": 8360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9418148258978165e-06,
+      "loss": 0.9861,
+      "step": 8361
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.939970197128781e-06,
+      "loss": 1.1291,
+      "step": 8362
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9381263508102164e-06,
+      "loss": 1.1621,
+      "step": 8363
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9362832871211177e-06,
+      "loss": 1.0369,
+      "step": 8364
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.934441006240404e-06,
+      "loss": 1.0077,
+      "step": 8365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9325995083469217e-06,
+      "loss": 1.0699,
+      "step": 8366
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9307587936194383e-06,
+      "loss": 1.0601,
+      "step": 8367
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.928918862236646e-06,
+      "loss": 1.2122,
+      "step": 8368
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9270797143771656e-06,
+      "loss": 1.1509,
+      "step": 8369
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9252413502195324e-06,
+      "loss": 0.9868,
+      "step": 8370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.923403769942216e-06,
+      "loss": 1.031,
+      "step": 8371
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.921566973723601e-06,
+      "loss": 1.1375,
+      "step": 8372
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.919730961741998e-06,
+      "loss": 1.0143,
+      "step": 8373
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9178957341756487e-06,
+      "loss": 1.0615,
+      "step": 8374
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9160612912027065e-06,
+      "loss": 1.1445,
+      "step": 8375
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.91422763300126e-06,
+      "loss": 1.0847,
+      "step": 8376
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9123947597493174e-06,
+      "loss": 1.0597,
+      "step": 8377
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910562671624808e-06,
+      "loss": 1.0543,
+      "step": 8378
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9087313688055854e-06,
+      "loss": 1.1672,
+      "step": 8379
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9069008514694321e-06,
+      "loss": 1.1033,
+      "step": 8380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9050711197940474e-06,
+      "loss": 1.0459,
+      "step": 8381
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9032421739570617e-06,
+      "loss": 1.0354,
+      "step": 8382
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9014140141360205e-06,
+      "loss": 1.1583,
+      "step": 8383
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8995866405083996e-06,
+      "loss": 1.0492,
+      "step": 8384
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8977600532516004e-06,
+      "loss": 1.1726,
+      "step": 8385
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8959342525429404e-06,
+      "loss": 0.957,
+      "step": 8386
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8941092385596616e-06,
+      "loss": 1.0862,
+      "step": 8387
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.892285011478938e-06,
+      "loss": 0.9668,
+      "step": 8388
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8904615714778574e-06,
+      "loss": 1.007,
+      "step": 8389
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8886389187334398e-06,
+      "loss": 0.9915,
+      "step": 8390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8868170534226193e-06,
+      "loss": 1.0505,
+      "step": 8391
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8849959757222613e-06,
+      "loss": 1.0219,
+      "step": 8392
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8831756858091566e-06,
+      "loss": 1.1223,
+      "step": 8393
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8813561838600058e-06,
+      "loss": 1.0225,
+      "step": 8394
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8795374700514479e-06,
+      "loss": 1.1016,
+      "step": 8395
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8777195445600404e-06,
+      "loss": 1.1066,
+      "step": 8396
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.87590240756226e-06,
+      "loss": 0.9447,
+      "step": 8397
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8740860592345156e-06,
+      "loss": 1.0322,
+      "step": 8398
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8722704997531282e-06,
+      "loss": 0.8833,
+      "step": 8399
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.870455729294355e-06,
+      "loss": 1.0492,
+      "step": 8400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8686417480343633e-06,
+      "loss": 1.0553,
+      "step": 8401
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8668285561492573e-06,
+      "loss": 1.1104,
+      "step": 8402
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8650161538150513e-06,
+      "loss": 1.0515,
+      "step": 8403
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.863204541207696e-06,
+      "loss": 0.9701,
+      "step": 8404
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8613937185030518e-06,
+      "loss": 1.125,
+      "step": 8405
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8595836858769135e-06,
+      "loss": 1.0999,
+      "step": 8406
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8577744435049972e-06,
+      "loss": 1.135,
+      "step": 8407
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.855965991562938e-06,
+      "loss": 1.2148,
+      "step": 8408
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8541583302262944e-06,
+      "loss": 1.0667,
+      "step": 8409
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.852351459670555e-06,
+      "loss": 0.9993,
+      "step": 8410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8505453800711215e-06,
+      "loss": 1.2878,
+      "step": 8411
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8487400916033293e-06,
+      "loss": 1.1326,
+      "step": 8412
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8469355944424271e-06,
+      "loss": 1.0781,
+      "step": 8413
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8451318887635949e-06,
+      "loss": 1.0829,
+      "step": 8414
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.843328974741936e-06,
+      "loss": 1.1621,
+      "step": 8415
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8415268525524654e-06,
+      "loss": 1.1523,
+      "step": 8416
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8397255223701326e-06,
+      "loss": 1.0898,
+      "step": 8417
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.83792498436981e-06,
+      "loss": 1.044,
+      "step": 8418
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8361252387262852e-06,
+      "loss": 1.0503,
+      "step": 8419
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8343262856142786e-06,
+      "loss": 0.9773,
+      "step": 8420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8325281252084238e-06,
+      "loss": 1.1008,
+      "step": 8421
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8307307576832866e-06,
+      "loss": 1.1516,
+      "step": 8422
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.82893418321335e-06,
+      "loss": 1.0259,
+      "step": 8423
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8271384019730186e-06,
+      "loss": 1.0547,
+      "step": 8424
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8253434141366256e-06,
+      "loss": 1.1096,
+      "step": 8425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8235492198784267e-06,
+      "loss": 1.056,
+      "step": 8426
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8217558193725948e-06,
+      "loss": 1.2075,
+      "step": 8427
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8199632127932331e-06,
+      "loss": 0.9749,
+      "step": 8428
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8181714003143592e-06,
+      "loss": 1.0448,
+      "step": 8429
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8163803821099236e-06,
+      "loss": 1.0889,
+      "step": 8430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8145901583537928e-06,
+      "loss": 0.9828,
+      "step": 8431
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8128007292197536e-06,
+      "loss": 1.1121,
+      "step": 8432
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8110120948815247e-06,
+      "loss": 1.1367,
+      "step": 8433
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8092242555127437e-06,
+      "loss": 1.1021,
+      "step": 8434
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8074372112869653e-06,
+      "loss": 1.1899,
+      "step": 8435
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8056509623776775e-06,
+      "loss": 0.9937,
+      "step": 8436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8038655089582836e-06,
+      "loss": 1.0578,
+      "step": 8437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8020808512021071e-06,
+      "loss": 1.1248,
+      "step": 8438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8002969892824062e-06,
+      "loss": 1.1638,
+      "step": 8439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.798513923372348e-06,
+      "loss": 1.1713,
+      "step": 8440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7967316536450309e-06,
+      "loss": 0.93,
+      "step": 8441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.794950180273478e-06,
+      "loss": 1.0437,
+      "step": 8442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.793169503430624e-06,
+      "loss": 1.0491,
+      "step": 8443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7913896232893391e-06,
+      "loss": 1.1086,
+      "step": 8444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7896105400224073e-06,
+      "loss": 1.1687,
+      "step": 8445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7878322538025373e-06,
+      "loss": 1.04,
+      "step": 8446
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.786054764802364e-06,
+      "loss": 1.0049,
+      "step": 8447
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7842780731944398e-06,
+      "loss": 0.9116,
+      "step": 8448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.782502179151242e-06,
+      "loss": 1.0166,
+      "step": 8449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7807270828451739e-06,
+      "loss": 1.093,
+      "step": 8450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7789527844485544e-06,
+      "loss": 0.9509,
+      "step": 8451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.777179284133632e-06,
+      "loss": 1.1433,
+      "step": 8452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.775406582072573e-06,
+      "loss": 1.1526,
+      "step": 8453
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7736346784374646e-06,
+      "loss": 1.0216,
+      "step": 8454
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7718635734003243e-06,
+      "loss": 1.0616,
+      "step": 8455
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7700932671330816e-06,
+      "loss": 1.0007,
+      "step": 8456
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7683237598075976e-06,
+      "loss": 0.9707,
+      "step": 8457
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7665550515956542e-06,
+      "loss": 1.0981,
+      "step": 8458
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7647871426689523e-06,
+      "loss": 1.046,
+      "step": 8459
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.763020033199112e-06,
+      "loss": 1.1519,
+      "step": 8460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7612537233576875e-06,
+      "loss": 1.1031,
+      "step": 8461
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7594882133161427e-06,
+      "loss": 1.0876,
+      "step": 8462
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.757723503245875e-06,
+      "loss": 1.1011,
+      "step": 8463
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7559595933181929e-06,
+      "loss": 1.1978,
+      "step": 8464
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7541964837043358e-06,
+      "loss": 1.067,
+      "step": 8465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7524341745754657e-06,
+      "loss": 1.2134,
+      "step": 8466
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7506726661026608e-06,
+      "loss": 1.1271,
+      "step": 8467
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7489119584569214e-06,
+      "loss": 1.0833,
+      "step": 8468
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.74715205180918e-06,
+      "loss": 1.094,
+      "step": 8469
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7453929463302777e-06,
+      "loss": 1.0638,
+      "step": 8470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7436346421909877e-06,
+      "loss": 1.1624,
+      "step": 8471
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.741877139562006e-06,
+      "loss": 1.198,
+      "step": 8472
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7401204386139414e-06,
+      "loss": 0.9302,
+      "step": 8473
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7383645395173355e-06,
+      "loss": 1.0842,
+      "step": 8474
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.736609442442645e-06,
+      "loss": 1.0439,
+      "step": 8475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7348551475602483e-06,
+      "loss": 1.0691,
+      "step": 8476
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7331016550404534e-06,
+      "loss": 0.9584,
+      "step": 8477
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.731348965053481e-06,
+      "loss": 1.0361,
+      "step": 8478
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7295970777694816e-06,
+      "loss": 1.1335,
+      "step": 8479
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7278459933585256e-06,
+      "loss": 1.075,
+      "step": 8480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.726095711990604e-06,
+      "loss": 1.0183,
+      "step": 8481
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7243462338356254e-06,
+      "loss": 1.0841,
+      "step": 8482
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7225975590634326e-06,
+      "loss": 1.1497,
+      "step": 8483
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7208496878437774e-06,
+      "loss": 1.1445,
+      "step": 8484
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7191026203463445e-06,
+      "loss": 1.0502,
+      "step": 8485
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7173563567407302e-06,
+      "loss": 1.1208,
+      "step": 8486
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7156108971964602e-06,
+      "loss": 1.0503,
+      "step": 8487
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7138662418829866e-06,
+      "loss": 0.9261,
+      "step": 8488
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7121223909696649e-06,
+      "loss": 0.991,
+      "step": 8489
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7103793446257898e-06,
+      "loss": 1.0918,
+      "step": 8490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7086371030205762e-06,
+      "loss": 1.1177,
+      "step": 8491
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7068956663231518e-06,
+      "loss": 1.1199,
+      "step": 8492
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7051550347025759e-06,
+      "loss": 1.0859,
+      "step": 8493
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70341520832782e-06,
+      "loss": 1.1211,
+      "step": 8494
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7016761873677866e-06,
+      "loss": 1.0566,
+      "step": 8495
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6999379719913e-06,
+      "loss": 1.0199,
+      "step": 8496
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6982005623670927e-06,
+      "loss": 0.9553,
+      "step": 8497
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6964639586638332e-06,
+      "loss": 1.0266,
+      "step": 8498
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.694728161050111e-06,
+      "loss": 1.1304,
+      "step": 8499
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.692993169694428e-06,
+      "loss": 1.0696,
+      "step": 8500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6912589847652183e-06,
+      "loss": 1.0764,
+      "step": 8501
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6895256064308285e-06,
+      "loss": 1.0305,
+      "step": 8502
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.687793034859536e-06,
+      "loss": 1.066,
+      "step": 8503
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6860612702195312e-06,
+      "loss": 1.031,
+      "step": 8504
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6843303126789301e-06,
+      "loss": 1.1023,
+      "step": 8505
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6826001624057709e-06,
+      "loss": 1.2327,
+      "step": 8506
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6808708195680156e-06,
+      "loss": 0.9751,
+      "step": 8507
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6791422843335415e-06,
+      "loss": 0.9894,
+      "step": 8508
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.677414556870155e-06,
+      "loss": 0.9797,
+      "step": 8509
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6756876373455776e-06,
+      "loss": 1.095,
+      "step": 8510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6739615259274533e-06,
+      "loss": 1.1478,
+      "step": 8511
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6722362227833534e-06,
+      "loss": 0.9988,
+      "step": 8512
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6705117280807626e-06,
+      "loss": 1.0548,
+      "step": 8513
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6687880419870917e-06,
+      "loss": 1.1815,
+      "step": 8514
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6670651646696777e-06,
+      "loss": 1.1387,
+      "step": 8515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6653430962957661e-06,
+      "loss": 1.0657,
+      "step": 8516
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6636218370325396e-06,
+      "loss": 1.1107,
+      "step": 8517
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.661901387047089e-06,
+      "loss": 1.1333,
+      "step": 8518
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6601817465064317e-06,
+      "loss": 1.0917,
+      "step": 8519
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.65846291557751e-06,
+      "loss": 1.093,
+      "step": 8520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6567448944271803e-06,
+      "loss": 0.9147,
+      "step": 8521
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6550276832222256e-06,
+      "loss": 1.1006,
+      "step": 8522
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.653311282129354e-06,
+      "loss": 1.0067,
+      "step": 8523
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6515956913151832e-06,
+      "loss": 1.0863,
+      "step": 8524
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6498809109462644e-06,
+      "loss": 1.1821,
+      "step": 8525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6481669411890632e-06,
+      "loss": 1.1121,
+      "step": 8526
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6464537822099647e-06,
+      "loss": 1.1504,
+      "step": 8527
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6447414341752832e-06,
+      "loss": 1.0586,
+      "step": 8528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.643029897251246e-06,
+      "loss": 1.1152,
+      "step": 8529
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6413191716040078e-06,
+      "loss": 0.9587,
+      "step": 8530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6396092573996437e-06,
+      "loss": 1.0981,
+      "step": 8531
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6379001548041474e-06,
+      "loss": 1.1853,
+      "step": 8532
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6361918639834306e-06,
+      "loss": 1.0598,
+      "step": 8533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6344843851033376e-06,
+      "loss": 1.142,
+      "step": 8534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.63277771832962e-06,
+      "loss": 0.9893,
+      "step": 8535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.631071863827962e-06,
+      "loss": 1.1786,
+      "step": 8536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6293668217639647e-06,
+      "loss": 1.1284,
+      "step": 8537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6276625923031452e-06,
+      "loss": 1.1184,
+      "step": 8538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6259591756109528e-06,
+      "loss": 1.0349,
+      "step": 8539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.624256571852748e-06,
+      "loss": 1.0715,
+      "step": 8540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6225547811938135e-06,
+      "loss": 1.0784,
+      "step": 8541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6208538037993616e-06,
+      "loss": 1.0077,
+      "step": 8542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6191536398345131e-06,
+      "loss": 1.1233,
+      "step": 8543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6174542894643185e-06,
+      "loss": 1.0757,
+      "step": 8544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6157557528537515e-06,
+      "loss": 1.032,
+      "step": 8545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6140580301676955e-06,
+      "loss": 1.1841,
+      "step": 8546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6123611215709679e-06,
+      "loss": 1.0786,
+      "step": 8547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6106650272282986e-06,
+      "loss": 1.0249,
+      "step": 8548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6089697473043375e-06,
+      "loss": 1.0349,
+      "step": 8549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.607275281963664e-06,
+      "loss": 1.1008,
+      "step": 8550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6055816313707684e-06,
+      "loss": 0.9894,
+      "step": 8551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6038887956900694e-06,
+      "loss": 1.0579,
+      "step": 8552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6021967750859058e-06,
+      "loss": 1.1746,
+      "step": 8553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6005055697225313e-06,
+      "loss": 1.1345,
+      "step": 8554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5988151797641283e-06,
+      "loss": 1.0586,
+      "step": 8555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.597125605374794e-06,
+      "loss": 1.031,
+      "step": 8556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5954368467185466e-06,
+      "loss": 1.0251,
+      "step": 8557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5937489039593312e-06,
+      "loss": 0.9944,
+      "step": 8558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.592061777261007e-06,
+      "loss": 1.0571,
+      "step": 8559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5903754667873571e-06,
+      "loss": 1.1089,
+      "step": 8560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5886899727020878e-06,
+      "loss": 0.9611,
+      "step": 8561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587005295168821e-06,
+      "loss": 1.0605,
+      "step": 8562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.585321434351098e-06,
+      "loss": 1.0896,
+      "step": 8563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.583638390412392e-06,
+      "loss": 0.9928,
+      "step": 8564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5819561635160819e-06,
+      "loss": 1.0687,
+      "step": 8565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5802747538254803e-06,
+      "loss": 1.1428,
+      "step": 8566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5785941615038102e-06,
+      "loss": 1.0181,
+      "step": 8567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5769143867142235e-06,
+      "loss": 1.1592,
+      "step": 8568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5752354296197914e-06,
+      "loss": 1.02,
+      "step": 8569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5735572903834962e-06,
+      "loss": 0.9701,
+      "step": 8570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5718799691682506e-06,
+      "loss": 1.0959,
+      "step": 8571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5702034661368904e-06,
+      "loss": 1.0386,
+      "step": 8572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5685277814521593e-06,
+      "loss": 1.0884,
+      "step": 8573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5668529152767364e-06,
+      "loss": 1.0464,
+      "step": 8574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5651788677732083e-06,
+      "loss": 0.9745,
+      "step": 8575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5635056391040904e-06,
+      "loss": 0.9722,
+      "step": 8576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5618332294318216e-06,
+      "loss": 1.1453,
+      "step": 8577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5601616389187445e-06,
+      "loss": 1.0803,
+      "step": 8578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5584908677271405e-06,
+      "loss": 1.0286,
+      "step": 8579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.556820916019206e-06,
+      "loss": 1.0388,
+      "step": 8580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5551517839570507e-06,
+      "loss": 0.9325,
+      "step": 8581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5534834717027158e-06,
+      "loss": 0.9568,
+      "step": 8582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.551815979418153e-06,
+      "loss": 0.9075,
+      "step": 8583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.550149307265244e-06,
+      "loss": 1.0452,
+      "step": 8584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5484834554057827e-06,
+      "loss": 1.0583,
+      "step": 8585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5468184240014839e-06,
+      "loss": 1.0426,
+      "step": 8586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5451542132139886e-06,
+      "loss": 1.1133,
+      "step": 8587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5434908232048551e-06,
+      "loss": 1.1609,
+      "step": 8588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.54182825413556e-06,
+      "loss": 0.9927,
+      "step": 8589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.540166506167504e-06,
+      "loss": 0.9941,
+      "step": 8590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5385055794620063e-06,
+      "loss": 1.1592,
+      "step": 8591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5368454741803018e-06,
+      "loss": 1.0208,
+      "step": 8592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5351861904835542e-06,
+      "loss": 1.1128,
+      "step": 8593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.533527728532841e-06,
+      "loss": 1.0471,
+      "step": 8594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5318700884891613e-06,
+      "loss": 0.9673,
+      "step": 8595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.530213270513441e-06,
+      "loss": 1.1086,
+      "step": 8596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5285572747665134e-06,
+      "loss": 1.0229,
+      "step": 8597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5269021014091434e-06,
+      "loss": 0.8976,
+      "step": 8598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5252477506020113e-06,
+      "loss": 1.0056,
+      "step": 8599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5235942225057154e-06,
+      "loss": 1.1187,
+      "step": 8600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5219415172807795e-06,
+      "loss": 1.0671,
+      "step": 8601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5202896350876407e-06,
+      "loss": 1.0703,
+      "step": 8602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5186385760866639e-06,
+      "loss": 1.1089,
+      "step": 8603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5169883404381313e-06,
+      "loss": 1.0049,
+      "step": 8604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.51533892830224e-06,
+      "loss": 1.0181,
+      "step": 8605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5136903398391168e-06,
+      "loss": 1.0354,
+      "step": 8606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5120425752087985e-06,
+      "loss": 1.1006,
+      "step": 8607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.510395634571248e-06,
+      "loss": 1.1348,
+      "step": 8608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5087495180863454e-06,
+      "loss": 1.0807,
+      "step": 8609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5071042259138957e-06,
+      "loss": 0.9392,
+      "step": 8610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5054597582136176e-06,
+      "loss": 1.1128,
+      "step": 8611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5038161151451548e-06,
+      "loss": 1.0723,
+      "step": 8612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.502173296868067e-06,
+      "loss": 0.9779,
+      "step": 8613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5005313035418334e-06,
+      "loss": 0.9958,
+      "step": 8614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.498890135325859e-06,
+      "loss": 1.0168,
+      "step": 8615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4972497923794627e-06,
+      "loss": 1.0127,
+      "step": 8616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.495610274861885e-06,
+      "loss": 1.1321,
+      "step": 8617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4939715829322898e-06,
+      "loss": 1.0833,
+      "step": 8618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4923337167497542e-06,
+      "loss": 1.019,
+      "step": 8619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4906966764732832e-06,
+      "loss": 1.0361,
+      "step": 8620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4890604622617933e-06,
+      "loss": 1.084,
+      "step": 8621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4874250742741247e-06,
+      "loss": 1.0647,
+      "step": 8622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4857905126690397e-06,
+      "loss": 0.9738,
+      "step": 8623
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4841567776052145e-06,
+      "loss": 1.0061,
+      "step": 8624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4825238692412514e-06,
+      "loss": 1.073,
+      "step": 8625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.480891787735672e-06,
+      "loss": 1.0187,
+      "step": 8626
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4792605332469089e-06,
+      "loss": 1.0105,
+      "step": 8627
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4776301059333275e-06,
+      "loss": 1.0415,
+      "step": 8628
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4760005059532034e-06,
+      "loss": 1.0171,
+      "step": 8629
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4743717334647313e-06,
+      "loss": 1.0464,
+      "step": 8630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4727437886260343e-06,
+      "loss": 0.9431,
+      "step": 8631
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4711166715951453e-06,
+      "loss": 1.0764,
+      "step": 8632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4694903825300234e-06,
+      "loss": 1.0486,
+      "step": 8633
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4678649215885476e-06,
+      "loss": 0.949,
+      "step": 8634
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4662402889285131e-06,
+      "loss": 1.0162,
+      "step": 8635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4646164847076317e-06,
+      "loss": 1.1023,
+      "step": 8636
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4629935090835434e-06,
+      "loss": 0.9537,
+      "step": 8637
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4613713622138004e-06,
+      "loss": 1.0171,
+      "step": 8638
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4597500442558798e-06,
+      "loss": 0.9713,
+      "step": 8639
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4581295553671725e-06,
+      "loss": 1.1509,
+      "step": 8640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.456509895704994e-06,
+      "loss": 0.9819,
+      "step": 8641
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.454891065426579e-06,
+      "loss": 1.1152,
+      "step": 8642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4532730646890791e-06,
+      "loss": 1.0723,
+      "step": 8643
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4516558936495629e-06,
+      "loss": 1.0015,
+      "step": 8644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4500395524650268e-06,
+      "loss": 1.0261,
+      "step": 8645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4484240412923778e-06,
+      "loss": 1.1494,
+      "step": 8646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4468093602884502e-06,
+      "loss": 0.9514,
+      "step": 8647
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4451955096099902e-06,
+      "loss": 1.1003,
+      "step": 8648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4435824894136673e-06,
+      "loss": 1.0275,
+      "step": 8649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.441970299856076e-06,
+      "loss": 1.092,
+      "step": 8650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4403589410937158e-06,
+      "loss": 1.0931,
+      "step": 8651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4387484132830164e-06,
+      "loss": 1.0244,
+      "step": 8652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4371387165803285e-06,
+      "loss": 1.0798,
+      "step": 8653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4355298511419124e-06,
+      "loss": 1.1543,
+      "step": 8654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4339218171239588e-06,
+      "loss": 1.137,
+      "step": 8655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4323146146825661e-06,
+      "loss": 1.0298,
+      "step": 8656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4307082439737652e-06,
+      "loss": 1.0881,
+      "step": 8657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4291027051534944e-06,
+      "loss": 0.9727,
+      "step": 8658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.427497998377615e-06,
+      "loss": 1.1096,
+      "step": 8659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4258941238019097e-06,
+      "loss": 1.075,
+      "step": 8660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4242910815820832e-06,
+      "loss": 1.0632,
+      "step": 8661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4226888718737485e-06,
+      "loss": 1.147,
+      "step": 8662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4210874948324515e-06,
+      "loss": 0.9409,
+      "step": 8663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4194869506136455e-06,
+      "loss": 0.9387,
+      "step": 8664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4178872393727116e-06,
+      "loss": 0.9883,
+      "step": 8665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4162883612649448e-06,
+      "loss": 1.0812,
+      "step": 8666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4146903164455571e-06,
+      "loss": 0.9882,
+      "step": 8667
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4130931050696873e-06,
+      "loss": 1.0476,
+      "step": 8668
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4114967272923907e-06,
+      "loss": 1.1287,
+      "step": 8669
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.409901183268636e-06,
+      "loss": 1.0188,
+      "step": 8670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4083064731533202e-06,
+      "loss": 1.1841,
+      "step": 8671
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4067125971012508e-06,
+      "loss": 1.1484,
+      "step": 8672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4051195552671559e-06,
+      "loss": 1.0278,
+      "step": 8673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4035273478056878e-06,
+      "loss": 1.0415,
+      "step": 8674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4019359748714168e-06,
+      "loss": 1.0076,
+      "step": 8675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.400345436618824e-06,
+      "loss": 1.1304,
+      "step": 8676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3987557332023215e-06,
+      "loss": 0.9551,
+      "step": 8677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3971668647762293e-06,
+      "loss": 1.0459,
+      "step": 8678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.395578831494795e-06,
+      "loss": 0.9728,
+      "step": 8679
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3939916335121807e-06,
+      "loss": 1.0962,
+      "step": 8680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3924052709824653e-06,
+      "loss": 1.1787,
+      "step": 8681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3908197440596516e-06,
+      "loss": 1.1011,
+      "step": 8682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3892350528976617e-06,
+      "loss": 0.9846,
+      "step": 8683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3876511976503292e-06,
+      "loss": 1.0996,
+      "step": 8684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.386068178471416e-06,
+      "loss": 1.0647,
+      "step": 8685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3844859955145928e-06,
+      "loss": 1.0531,
+      "step": 8686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.382904648933462e-06,
+      "loss": 1.0372,
+      "step": 8687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3813241388815312e-06,
+      "loss": 1.0712,
+      "step": 8688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3797444655122338e-06,
+      "loss": 1.1567,
+      "step": 8689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3781656289789213e-06,
+      "loss": 1.1168,
+      "step": 8690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.376587629434868e-06,
+      "loss": 1.0564,
+      "step": 8691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3750104670332566e-06,
+      "loss": 1.1082,
+      "step": 8692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.373434141927199e-06,
+      "loss": 1.1106,
+      "step": 8693
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.37185865426972e-06,
+      "loss": 1.0217,
+      "step": 8694
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3702840042137621e-06,
+      "loss": 1.1089,
+      "step": 8695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3687101919121947e-06,
+      "loss": 1.1338,
+      "step": 8696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3671372175177934e-06,
+      "loss": 1.0337,
+      "step": 8697
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3655650811832632e-06,
+      "loss": 1.0635,
+      "step": 8698
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3639937830612248e-06,
+      "loss": 1.0818,
+      "step": 8699
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3624233233042128e-06,
+      "loss": 0.9379,
+      "step": 8700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3608537020646872e-06,
+      "loss": 1.0453,
+      "step": 8701
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3592849194950241e-06,
+      "loss": 1.0162,
+      "step": 8702
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3577169757475117e-06,
+      "loss": 1.0166,
+      "step": 8703
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3561498709743703e-06,
+      "loss": 1.0518,
+      "step": 8704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.354583605327725e-06,
+      "loss": 0.9058,
+      "step": 8705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3530181789596274e-06,
+      "loss": 1.1091,
+      "step": 8706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3514535920220494e-06,
+      "loss": 1.0603,
+      "step": 8707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3498898446668718e-06,
+      "loss": 1.1379,
+      "step": 8708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.348326937045905e-06,
+      "loss": 1.0128,
+      "step": 8709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3467648693108703e-06,
+      "loss": 1.1323,
+      "step": 8710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3452036416134084e-06,
+      "loss": 1.0291,
+      "step": 8711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3436432541050837e-06,
+      "loss": 1.0896,
+      "step": 8712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3420837069373705e-06,
+      "loss": 1.0146,
+      "step": 8713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3405250002616688e-06,
+      "loss": 1.0251,
+      "step": 8714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3389671342292975e-06,
+      "loss": 1.031,
+      "step": 8715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3374101089914882e-06,
+      "loss": 1.0598,
+      "step": 8716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3358539246993906e-06,
+      "loss": 0.9912,
+      "step": 8717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3342985815040821e-06,
+      "loss": 1.0953,
+      "step": 8718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.332744079556546e-06,
+      "loss": 1.1296,
+      "step": 8719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3311904190076942e-06,
+      "loss": 0.8962,
+      "step": 8720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.329637600008349e-06,
+      "loss": 1.0874,
+      "step": 8721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3280856227092575e-06,
+      "loss": 0.9353,
+      "step": 8722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3265344872610841e-06,
+      "loss": 0.9266,
+      "step": 8723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.324984193814407e-06,
+      "loss": 0.9983,
+      "step": 8724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3234347425197237e-06,
+      "loss": 0.9834,
+      "step": 8725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.321886133527457e-06,
+      "loss": 1.0906,
+      "step": 8726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3203383669879355e-06,
+      "loss": 1.0493,
+      "step": 8727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3187914430514193e-06,
+      "loss": 1.0463,
+      "step": 8728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3172453618680758e-06,
+      "loss": 1.0928,
+      "step": 8729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3157001235879985e-06,
+      "loss": 1.0872,
+      "step": 8730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3141557283611983e-06,
+      "loss": 1.071,
+      "step": 8731
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3126121763375944e-06,
+      "loss": 1.124,
+      "step": 8732
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3110694676670343e-06,
+      "loss": 0.9929,
+      "step": 8733
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3095276024992854e-06,
+      "loss": 1.1243,
+      "step": 8734
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.307986580984022e-06,
+      "loss": 1.0457,
+      "step": 8735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3064464032708491e-06,
+      "loss": 1.0012,
+      "step": 8736
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.304907069509279e-06,
+      "loss": 0.9229,
+      "step": 8737
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3033685798487517e-06,
+      "loss": 1.0187,
+      "step": 8738
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.301830934438616e-06,
+      "loss": 1.1379,
+      "step": 8739
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.300294133428146e-06,
+      "loss": 1.0305,
+      "step": 8740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2987581769665293e-06,
+      "loss": 1.0703,
+      "step": 8741
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.297223065202875e-06,
+      "loss": 1.2168,
+      "step": 8742
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2956887982862065e-06,
+      "loss": 1.1973,
+      "step": 8743
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.294155376365468e-06,
+      "loss": 1.0488,
+      "step": 8744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2926227995895234e-06,
+      "loss": 1.1072,
+      "step": 8745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2910910681071499e-06,
+      "loss": 1.1716,
+      "step": 8746
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2895601820670423e-06,
+      "loss": 0.9916,
+      "step": 8747
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2880301416178199e-06,
+      "loss": 1.093,
+      "step": 8748
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2865009469080114e-06,
+      "loss": 1.0133,
+      "step": 8749
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2849725980860716e-06,
+      "loss": 1.1396,
+      "step": 8750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2834450953003675e-06,
+      "loss": 1.1189,
+      "step": 8751
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2819184386991845e-06,
+      "loss": 1.0099,
+      "step": 8752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2803926284307345e-06,
+      "loss": 1.099,
+      "step": 8753
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2788676646431296e-06,
+      "loss": 1.0598,
+      "step": 8754
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2773435474844143e-06,
+      "loss": 1.0461,
+      "step": 8755
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2758202771025495e-06,
+      "loss": 1.1414,
+      "step": 8756
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2742978536454065e-06,
+      "loss": 0.9709,
+      "step": 8757
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2727762772607832e-06,
+      "loss": 1.0918,
+      "step": 8758
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2712555480963861e-06,
+      "loss": 1.0996,
+      "step": 8759
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2697356662998495e-06,
+      "loss": 1.0127,
+      "step": 8760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2682166320187183e-06,
+      "loss": 0.8263,
+      "step": 8761
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.266698445400454e-06,
+      "loss": 1.0881,
+      "step": 8762
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2651811065924414e-06,
+      "loss": 1.0737,
+      "step": 8763
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.263664615741982e-06,
+      "loss": 1.0919,
+      "step": 8764
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.262148972996291e-06,
+      "loss": 1.0977,
+      "step": 8765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2606341785025067e-06,
+      "loss": 1.0905,
+      "step": 8766
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2591202324076779e-06,
+      "loss": 1.0454,
+      "step": 8767
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2576071348587803e-06,
+      "loss": 1.0684,
+      "step": 8768
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2560948860026989e-06,
+      "loss": 0.9792,
+      "step": 8769
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2545834859862383e-06,
+      "loss": 1.1145,
+      "step": 8770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2530729349561244e-06,
+      "loss": 1.1677,
+      "step": 8771
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2515632330589987e-06,
+      "loss": 1.033,
+      "step": 8772
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2500543804414179e-06,
+      "loss": 1.0703,
+      "step": 8773
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2485463772498607e-06,
+      "loss": 1.178,
+      "step": 8774
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2470392236307193e-06,
+      "loss": 1.1077,
+      "step": 8775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2455329197303023e-06,
+      "loss": 0.9399,
+      "step": 8776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2440274656948436e-06,
+      "loss": 0.9424,
+      "step": 8777
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2425228616704854e-06,
+      "loss": 1.0117,
+      "step": 8778
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2410191078032918e-06,
+      "loss": 1.0065,
+      "step": 8779
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2395162042392483e-06,
+      "loss": 1.0823,
+      "step": 8780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2380141511242483e-06,
+      "loss": 1.05,
+      "step": 8781
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2365129486041116e-06,
+      "loss": 1.1694,
+      "step": 8782
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2350125968245707e-06,
+      "loss": 1.1001,
+      "step": 8783
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2335130959312747e-06,
+      "loss": 1.073,
+      "step": 8784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2320144460697948e-06,
+      "loss": 1.0894,
+      "step": 8785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2305166473856122e-06,
+      "loss": 1.067,
+      "step": 8786
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2290197000241344e-06,
+      "loss": 0.9956,
+      "step": 8787
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2275236041306815e-06,
+      "loss": 0.9929,
+      "step": 8788
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2260283598504885e-06,
+      "loss": 1.0596,
+      "step": 8789
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2245339673287148e-06,
+      "loss": 0.9307,
+      "step": 8790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2230404267104301e-06,
+      "loss": 1.0369,
+      "step": 8791
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2215477381406215e-06,
+      "loss": 1.0015,
+      "step": 8792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2200559017642022e-06,
+      "loss": 1.0085,
+      "step": 8793
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2185649177259906e-06,
+      "loss": 1.0912,
+      "step": 8794
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2170747861707298e-06,
+      "loss": 1.0684,
+      "step": 8795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2155855072430833e-06,
+      "loss": 1.0381,
+      "step": 8796
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2140970810876218e-06,
+      "loss": 1.0525,
+      "step": 8797
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2126095078488398e-06,
+      "loss": 0.9928,
+      "step": 8798
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2111227876711484e-06,
+      "loss": 1.0531,
+      "step": 8799
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2096369206988735e-06,
+      "loss": 0.9868,
+      "step": 8800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2081519070762637e-06,
+      "loss": 1.0708,
+      "step": 8801
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2066677469474764e-06,
+      "loss": 1.0664,
+      "step": 8802
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2051844404565927e-06,
+      "loss": 0.9822,
+      "step": 8803
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2037019877476109e-06,
+      "loss": 1.0479,
+      "step": 8804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2022203889644425e-06,
+      "loss": 0.8835,
+      "step": 8805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2007396442509155e-06,
+      "loss": 1.1456,
+      "step": 8806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1992597537507822e-06,
+      "loss": 0.9913,
+      "step": 8807
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1977807176077028e-06,
+      "loss": 1.1272,
+      "step": 8808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1963025359652602e-06,
+      "loss": 0.9644,
+      "step": 8809
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1948252089669564e-06,
+      "loss": 1.0881,
+      "step": 8810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1933487367562025e-06,
+      "loss": 0.8854,
+      "step": 8811
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1918731194763355e-06,
+      "loss": 0.9802,
+      "step": 8812
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1903983572706024e-06,
+      "loss": 1.0159,
+      "step": 8813
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1889244502821695e-06,
+      "loss": 1.0715,
+      "step": 8814
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1874513986541224e-06,
+      "loss": 1.0923,
+      "step": 8815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1859792025294603e-06,
+      "loss": 1.0969,
+      "step": 8816
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1845078620511008e-06,
+      "loss": 0.8938,
+      "step": 8817
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.183037377361882e-06,
+      "loss": 1.1272,
+      "step": 8818
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1815677486045517e-06,
+      "loss": 1.0239,
+      "step": 8819
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1800989759217784e-06,
+      "loss": 1.125,
+      "step": 8820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.17863105945615e-06,
+      "loss": 1.1311,
+      "step": 8821
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.177163999350165e-06,
+      "loss": 1.115,
+      "step": 8822
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1756977957462457e-06,
+      "loss": 1.0898,
+      "step": 8823
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1742324487867251e-06,
+      "loss": 1.0818,
+      "step": 8824
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1727679586138586e-06,
+      "loss": 1.0059,
+      "step": 8825
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1713043253698186e-06,
+      "loss": 1.156,
+      "step": 8826
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.169841549196683e-06,
+      "loss": 1.042,
+      "step": 8827
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1683796302364604e-06,
+      "loss": 0.9972,
+      "step": 8828
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1669185686310713e-06,
+      "loss": 1.0574,
+      "step": 8829
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1654583645223494e-06,
+      "loss": 1.0669,
+      "step": 8830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1639990180520521e-06,
+      "loss": 1.1201,
+      "step": 8831
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1625405293618453e-06,
+      "loss": 0.9879,
+      "step": 8832
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1610828985933186e-06,
+      "loss": 0.9802,
+      "step": 8833
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1596261258879792e-06,
+      "loss": 0.9744,
+      "step": 8834
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1581702113872396e-06,
+      "loss": 1.0642,
+      "step": 8835
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1567151552324408e-06,
+      "loss": 1.032,
+      "step": 8836
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1552609575648387e-06,
+      "loss": 1.0923,
+      "step": 8837
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1538076185255987e-06,
+      "loss": 1.0808,
+      "step": 8838
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.152355138255814e-06,
+      "loss": 1.0754,
+      "step": 8839
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1509035168964822e-06,
+      "loss": 1.061,
+      "step": 8840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1494527545885281e-06,
+      "loss": 1.0879,
+      "step": 8841
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.148002851472788e-06,
+      "loss": 1.0947,
+      "step": 8842
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1465538076900119e-06,
+      "loss": 1.084,
+      "step": 8843
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1451056233808711e-06,
+      "loss": 0.9015,
+      "step": 8844
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1436582986859568e-06,
+      "loss": 1.1108,
+      "step": 8845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1422118337457667e-06,
+      "loss": 1.1113,
+      "step": 8846
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1407662287007237e-06,
+      "loss": 1.0176,
+      "step": 8847
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1393214836911615e-06,
+      "loss": 1.0823,
+      "step": 8848
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1378775988573365e-06,
+      "loss": 1.0688,
+      "step": 8849
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1364345743394157e-06,
+      "loss": 1.0074,
+      "step": 8850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1349924102774822e-06,
+      "loss": 1.1333,
+      "step": 8851
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1335511068115424e-06,
+      "loss": 1.0317,
+      "step": 8852
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.132110664081515e-06,
+      "loss": 1.0413,
+      "step": 8853
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1306710822272305e-06,
+      "loss": 1.0648,
+      "step": 8854
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.129232361388446e-06,
+      "loss": 1.0587,
+      "step": 8855
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1277945017048274e-06,
+      "loss": 1.1685,
+      "step": 8856
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1263575033159569e-06,
+      "loss": 1.0396,
+      "step": 8857
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.124921366361338e-06,
+      "loss": 1.0566,
+      "step": 8858
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1234860909803856e-06,
+      "loss": 1.0522,
+      "step": 8859
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1220516773124336e-06,
+      "loss": 1.1228,
+      "step": 8860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1206181254967352e-06,
+      "loss": 0.9827,
+      "step": 8861
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1191854356724518e-06,
+      "loss": 1.0754,
+      "step": 8862
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.11775360797867e-06,
+      "loss": 1.1414,
+      "step": 8863
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.116322642554386e-06,
+      "loss": 0.9952,
+      "step": 8864
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1148925395385135e-06,
+      "loss": 0.9604,
+      "step": 8865
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.113463299069888e-06,
+      "loss": 1.1357,
+      "step": 8866
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1120349212872516e-06,
+      "loss": 1.0498,
+      "step": 8867
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1106074063292726e-06,
+      "loss": 1.0398,
+      "step": 8868
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1091807543345322e-06,
+      "loss": 1.1302,
+      "step": 8869
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.107754965441522e-06,
+      "loss": 1.1313,
+      "step": 8870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.106330039788659e-06,
+      "loss": 0.9926,
+      "step": 8871
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.10490597751427e-06,
+      "loss": 1.0504,
+      "step": 8872
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1034827787565972e-06,
+      "loss": 1.0903,
+      "step": 8873
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1020604436538052e-06,
+      "loss": 1.0737,
+      "step": 8874
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1006389723439714e-06,
+      "loss": 1.0376,
+      "step": 8875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0992183649650868e-06,
+      "loss": 1.0117,
+      "step": 8876
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.097798621655064e-06,
+      "loss": 0.8921,
+      "step": 8877
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0963797425517276e-06,
+      "loss": 1.1492,
+      "step": 8878
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.094961727792816e-06,
+      "loss": 0.9724,
+      "step": 8879
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0935445775159926e-06,
+      "loss": 0.9692,
+      "step": 8880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.092128291858825e-06,
+      "loss": 1.1289,
+      "step": 8881
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0907128709588077e-06,
+      "loss": 0.9623,
+      "step": 8882
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0892983149533476e-06,
+      "loss": 0.9233,
+      "step": 8883
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0878846239797624e-06,
+      "loss": 0.9587,
+      "step": 8884
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0864717981752948e-06,
+      "loss": 1.0399,
+      "step": 8885
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.085059837677097e-06,
+      "loss": 1.0677,
+      "step": 8886
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0836487426222364e-06,
+      "loss": 1.1077,
+      "step": 8887
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0822385131477042e-06,
+      "loss": 1.1609,
+      "step": 8888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0808291493903978e-06,
+      "loss": 1.0217,
+      "step": 8889
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0794206514871363e-06,
+      "loss": 0.9993,
+      "step": 8890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0780130195746574e-06,
+      "loss": 1.1575,
+      "step": 8891
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0766062537896072e-06,
+      "loss": 1.1256,
+      "step": 8892
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.075200354268553e-06,
+      "loss": 1.0603,
+      "step": 8893
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0737953211479767e-06,
+      "loss": 1.0244,
+      "step": 8894
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0723911545642728e-06,
+      "loss": 1.0334,
+      "step": 8895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0709878546537589e-06,
+      "loss": 1.051,
+      "step": 8896
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0695854215526613e-06,
+      "loss": 1.0112,
+      "step": 8897
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.068183855397127e-06,
+      "loss": 1.0605,
+      "step": 8898
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0667831563232178e-06,
+      "loss": 1.092,
+      "step": 8899
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0653833244669109e-06,
+      "loss": 1.0266,
+      "step": 8900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0639843599640942e-06,
+      "loss": 1.1006,
+      "step": 8901
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0625862629505812e-06,
+      "loss": 0.9998,
+      "step": 8902
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0611890335620934e-06,
+      "loss": 0.989,
+      "step": 8903
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0597926719342744e-06,
+      "loss": 1.1177,
+      "step": 8904
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0583971782026747e-06,
+      "loss": 0.9651,
+      "step": 8905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0570025525027682e-06,
+      "loss": 1.0315,
+      "step": 8906
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0556087949699468e-06,
+      "loss": 0.9872,
+      "step": 8907
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0542159057395063e-06,
+      "loss": 1.0658,
+      "step": 8908
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0528238849466677e-06,
+      "loss": 1.0701,
+      "step": 8909
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0514327327265683e-06,
+      "loss": 1.1393,
+      "step": 8910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0500424492142547e-06,
+      "loss": 1.1207,
+      "step": 8911
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0486530345446954e-06,
+      "loss": 1.0094,
+      "step": 8912
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0472644888527683e-06,
+      "loss": 0.9583,
+      "step": 8913
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.045876812273272e-06,
+      "loss": 1.0021,
+      "step": 8914
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0444900049409245e-06,
+      "loss": 1.1782,
+      "step": 8915
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0431040669903458e-06,
+      "loss": 0.9064,
+      "step": 8916
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0417189985560826e-06,
+      "loss": 1.0054,
+      "step": 8917
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0403347997725977e-06,
+      "loss": 1.0967,
+      "step": 8918
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.038951470774261e-06,
+      "loss": 1.0767,
+      "step": 8919
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0375690116953674e-06,
+      "loss": 1.2012,
+      "step": 8920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0361874226701196e-06,
+      "loss": 1.0339,
+      "step": 8921
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0348067038326436e-06,
+      "loss": 1.0996,
+      "step": 8922
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0334268553169735e-06,
+      "loss": 1.0217,
+      "step": 8923
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.032047877257062e-06,
+      "loss": 1.0703,
+      "step": 8924
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0306697697867761e-06,
+      "loss": 1.1265,
+      "step": 8925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0292925330399052e-06,
+      "loss": 1.0259,
+      "step": 8926
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0279161671501415e-06,
+      "loss": 1.0431,
+      "step": 8927
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.026540672251105e-06,
+      "loss": 1.1196,
+      "step": 8928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0251660484763237e-06,
+      "loss": 0.9398,
+      "step": 8929
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0237922959592417e-06,
+      "loss": 1.0745,
+      "step": 8930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0224194148332224e-06,
+      "loss": 1.0271,
+      "step": 8931
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.02104740523154e-06,
+      "loss": 1.1057,
+      "step": 8932
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0196762672873872e-06,
+      "loss": 0.9789,
+      "step": 8933
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.018306001133873e-06,
+      "loss": 1.0211,
+      "step": 8934
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0169366069040164e-06,
+      "loss": 0.9993,
+      "step": 8935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0155680847307602e-06,
+      "loss": 1.1055,
+      "step": 8936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0142004347469535e-06,
+      "loss": 1.0543,
+      "step": 8937
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0128336570853636e-06,
+      "loss": 0.9885,
+      "step": 8938
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0114677518786798e-06,
+      "loss": 1.075,
+      "step": 8939
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0101027192594948e-06,
+      "loss": 1.0146,
+      "step": 8940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0087385593603272e-06,
+      "loss": 1.1479,
+      "step": 8941
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0073752723136076e-06,
+      "loss": 1.033,
+      "step": 8942
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.006012858251677e-06,
+      "loss": 1.051,
+      "step": 8943
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0046513173068006e-06,
+      "loss": 1.0325,
+      "step": 8944
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0032906496111504e-06,
+      "loss": 0.9331,
+      "step": 8945
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0019308552968165e-06,
+      "loss": 1.0417,
+      "step": 8946
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0005719344958053e-06,
+      "loss": 1.0245,
+      "step": 8947
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.992138873400414e-07,
+      "loss": 1.0387,
+      "step": 8948
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.97856713961357e-07,
+      "loss": 1.051,
+      "step": 8949
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.965004144915069e-07,
+      "loss": 1.177,
+      "step": 8950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.951449890621555e-07,
+      "loss": 0.9541,
+      "step": 8951
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.937904378048846e-07,
+      "loss": 1.1108,
+      "step": 8952
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.92436760851192e-07,
+      "loss": 0.916,
+      "step": 8953
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.910839583324883e-07,
+      "loss": 1.0498,
+      "step": 8954
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.897320303801018e-07,
+      "loss": 1.0106,
+      "step": 8955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.883809771252762e-07,
+      "loss": 1.0166,
+      "step": 8956
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.870307986991657e-07,
+      "loss": 1.0535,
+      "step": 8957
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.856814952328464e-07,
+      "loss": 0.908,
+      "step": 8958
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.843330668573048e-07,
+      "loss": 1.1472,
+      "step": 8959
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.829855137034395e-07,
+      "loss": 1.1301,
+      "step": 8960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.816388359020734e-07,
+      "loss": 1.1287,
+      "step": 8961
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.802930335839346e-07,
+      "loss": 1.0037,
+      "step": 8962
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.789481068796737e-07,
+      "loss": 1.0491,
+      "step": 8963
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.77604055919854e-07,
+      "loss": 1.0798,
+      "step": 8964
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.762608808349494e-07,
+      "loss": 1.1089,
+      "step": 8965
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.749185817553564e-07,
+      "loss": 1.082,
+      "step": 8966
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.735771588113818e-07,
+      "loss": 1.1001,
+      "step": 8967
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.722366121332449e-07,
+      "loss": 0.9092,
+      "step": 8968
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.708969418510861e-07,
+      "loss": 1.1116,
+      "step": 8969
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.695581480949566e-07,
+      "loss": 1.0078,
+      "step": 8970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.68220230994823e-07,
+      "loss": 1.1257,
+      "step": 8971
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.668831906805709e-07,
+      "loss": 1.1025,
+      "step": 8972
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.655470272819946e-07,
+      "loss": 1.0321,
+      "step": 8973
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.642117409288032e-07,
+      "loss": 0.9143,
+      "step": 8974
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.628773317506302e-07,
+      "loss": 1.0344,
+      "step": 8975
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.615437998770105e-07,
+      "loss": 1.0635,
+      "step": 8976
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.602111454374053e-07,
+      "loss": 1.0955,
+      "step": 8977
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.58879368561182e-07,
+      "loss": 1.0466,
+      "step": 8978
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.575484693776294e-07,
+      "loss": 1.0209,
+      "step": 8979
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.562184480159486e-07,
+      "loss": 1.0474,
+      "step": 8980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.548893046052553e-07,
+      "loss": 1.0295,
+      "step": 8981
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.535610392745764e-07,
+      "loss": 1.0405,
+      "step": 8982
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.522336521528608e-07,
+      "loss": 1.2395,
+      "step": 8983
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.509071433689654e-07,
+      "loss": 1.0549,
+      "step": 8984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.495815130516695e-07,
+      "loss": 1.0989,
+      "step": 8985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.482567613296567e-07,
+      "loss": 1.0728,
+      "step": 8986
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.469328883315343e-07,
+      "loss": 0.9937,
+      "step": 8987
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.456098941858238e-07,
+      "loss": 1.0195,
+      "step": 8988
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.442877790209526e-07,
+      "loss": 0.8147,
+      "step": 8989
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.429665429652712e-07,
+      "loss": 1.0889,
+      "step": 8990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.41646186147045e-07,
+      "loss": 1.093,
+      "step": 8991
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.403267086944468e-07,
+      "loss": 1.0913,
+      "step": 8992
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.390081107355731e-07,
+      "loss": 1.1167,
+      "step": 8993
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.376903923984271e-07,
+      "loss": 1.0122,
+      "step": 8994
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.363735538109331e-07,
+      "loss": 1.0593,
+      "step": 8995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.350575951009255e-07,
+      "loss": 1.0,
+      "step": 8996
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.337425163961533e-07,
+      "loss": 0.957,
+      "step": 8997
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.324283178242832e-07,
+      "loss": 0.9503,
+      "step": 8998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.311149995128954e-07,
+      "loss": 1.0151,
+      "step": 8999
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.298025615894824e-07,
+      "loss": 0.9714,
+      "step": 9000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.284910041814543e-07,
+      "loss": 1.125,
+      "step": 9001
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.271803274161329e-07,
+      "loss": 1.0962,
+      "step": 9002
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.258705314207572e-07,
+      "loss": 1.0833,
+      "step": 9003
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.245616163224802e-07,
+      "loss": 1.0247,
+      "step": 9004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.232535822483645e-07,
+      "loss": 1.0232,
+      "step": 9005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.219464293253944e-07,
+      "loss": 1.1021,
+      "step": 9006
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.206401576804669e-07,
+      "loss": 1.0039,
+      "step": 9007
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.193347674403886e-07,
+      "loss": 0.9604,
+      "step": 9008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.180302587318868e-07,
+      "loss": 1.1399,
+      "step": 9009
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.167266316815992e-07,
+      "loss": 1.1643,
+      "step": 9010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.15423886416078e-07,
+      "loss": 1.0146,
+      "step": 9011
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.14122023061792e-07,
+      "loss": 1.1467,
+      "step": 9012
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.128210417451245e-07,
+      "loss": 1.053,
+      "step": 9013
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.115209425923699e-07,
+      "loss": 1.0984,
+      "step": 9014
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.102217257297419e-07,
+      "loss": 1.1101,
+      "step": 9015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.089233912833617e-07,
+      "loss": 1.1461,
+      "step": 9016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.076259393792708e-07,
+      "loss": 1.0596,
+      "step": 9017
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.063293701434273e-07,
+      "loss": 1.1357,
+      "step": 9018
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.050336837016915e-07,
+      "loss": 1.064,
+      "step": 9019
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.037388801798485e-07,
+      "loss": 1.0906,
+      "step": 9020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.024449597036001e-07,
+      "loss": 1.0688,
+      "step": 9021
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.011519223985499e-07,
+      "loss": 1.0286,
+      "step": 9022
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.998597683902299e-07,
+      "loss": 1.0669,
+      "step": 9023
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.985684978040743e-07,
+      "loss": 1.1016,
+      "step": 9024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.972781107654405e-07,
+      "loss": 1.0508,
+      "step": 9025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.95988607399596e-07,
+      "loss": 1.0774,
+      "step": 9026
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.946999878317197e-07,
+      "loss": 1.0646,
+      "step": 9027
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.934122521869104e-07,
+      "loss": 1.0947,
+      "step": 9028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.921254005901813e-07,
+      "loss": 1.0889,
+      "step": 9029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.908394331664516e-07,
+      "loss": 1.0769,
+      "step": 9030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.895543500405668e-07,
+      "loss": 1.0994,
+      "step": 9031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.88270151337276e-07,
+      "loss": 0.9878,
+      "step": 9032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.869868371812451e-07,
+      "loss": 1.0403,
+      "step": 9033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.857044076970589e-07,
+      "loss": 0.9564,
+      "step": 9034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.8442286300921e-07,
+      "loss": 0.9878,
+      "step": 9035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.8314220324211e-07,
+      "loss": 1.0808,
+      "step": 9036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.818624285200827e-07,
+      "loss": 1.0674,
+      "step": 9037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.805835389673644e-07,
+      "loss": 1.0818,
+      "step": 9038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.793055347081103e-07,
+      "loss": 1.04,
+      "step": 9039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.780284158663833e-07,
+      "loss": 1.0845,
+      "step": 9040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.767521825661618e-07,
+      "loss": 1.0519,
+      "step": 9041
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.754768349313436e-07,
+      "loss": 0.9866,
+      "step": 9042
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.742023730857341e-07,
+      "loss": 1.0173,
+      "step": 9043
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.729287971530554e-07,
+      "loss": 0.9863,
+      "step": 9044
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.716561072569474e-07,
+      "loss": 1.051,
+      "step": 9045
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.703843035209547e-07,
+      "loss": 1.0344,
+      "step": 9046
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.691133860685452e-07,
+      "loss": 0.9752,
+      "step": 9047
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.678433550230959e-07,
+      "loss": 1.1882,
+      "step": 9048
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.665742105078967e-07,
+      "loss": 1.0603,
+      "step": 9049
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.65305952646156e-07,
+      "loss": 0.9871,
+      "step": 9050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.640385815609909e-07,
+      "loss": 1.0027,
+      "step": 9051
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.627720973754372e-07,
+      "loss": 1.0664,
+      "step": 9052
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.615065002124434e-07,
+      "loss": 1.0686,
+      "step": 9053
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.602417901948701e-07,
+      "loss": 1.1125,
+      "step": 9054
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.58977967445489e-07,
+      "loss": 0.9526,
+      "step": 9055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.577150320869953e-07,
+      "loss": 1.0383,
+      "step": 9056
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.564529842419866e-07,
+      "loss": 1.0217,
+      "step": 9057
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.551918240329848e-07,
+      "loss": 0.9269,
+      "step": 9058
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.539315515824165e-07,
+      "loss": 1.0413,
+      "step": 9059
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.52672167012627e-07,
+      "loss": 1.0969,
+      "step": 9060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.514136704458786e-07,
+      "loss": 1.0403,
+      "step": 9061
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.501560620043403e-07,
+      "loss": 0.9783,
+      "step": 9062
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.488993418100955e-07,
+      "loss": 1.0464,
+      "step": 9063
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.476435099851488e-07,
+      "loss": 1.114,
+      "step": 9064
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.463885666514105e-07,
+      "loss": 0.9548,
+      "step": 9065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.451345119307097e-07,
+      "loss": 1.1672,
+      "step": 9066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.438813459447848e-07,
+      "loss": 1.1074,
+      "step": 9067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.426290688152927e-07,
+      "loss": 1.1492,
+      "step": 9068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.413776806638041e-07,
+      "loss": 0.9351,
+      "step": 9069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.401271816117951e-07,
+      "loss": 1.1448,
+      "step": 9070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.388775717806641e-07,
+      "loss": 1.1167,
+      "step": 9071
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.37628851291723e-07,
+      "loss": 1.0784,
+      "step": 9072
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.363810202661915e-07,
+      "loss": 1.0822,
+      "step": 9073
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.351340788252094e-07,
+      "loss": 0.9837,
+      "step": 9074
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.338880270898242e-07,
+      "loss": 1.1045,
+      "step": 9075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.326428651810026e-07,
+      "loss": 1.0122,
+      "step": 9076
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.3139859321962e-07,
+      "loss": 1.0945,
+      "step": 9077
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.301552113264688e-07,
+      "loss": 1.0364,
+      "step": 9078
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.289127196222535e-07,
+      "loss": 1.093,
+      "step": 9079
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.276711182275932e-07,
+      "loss": 1.1438,
+      "step": 9080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.26430407263018e-07,
+      "loss": 1.1011,
+      "step": 9081
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.251905868489751e-07,
+      "loss": 1.1313,
+      "step": 9082
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.239516571058248e-07,
+      "loss": 0.9932,
+      "step": 9083
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.227136181538387e-07,
+      "loss": 0.9016,
+      "step": 9084
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.214764701132017e-07,
+      "loss": 1.0352,
+      "step": 9085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.202402131040143e-07,
+      "loss": 1.0483,
+      "step": 9086
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.190048472462897e-07,
+      "loss": 1.0408,
+      "step": 9087
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.17770372659955e-07,
+      "loss": 1.0516,
+      "step": 9088
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.165367894648502e-07,
+      "loss": 1.1265,
+      "step": 9089
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.15304097780727e-07,
+      "loss": 1.048,
+      "step": 9090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.140722977272586e-07,
+      "loss": 0.9965,
+      "step": 9091
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.128413894240173e-07,
+      "loss": 0.9973,
+      "step": 9092
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.116113729905006e-07,
+      "loss": 1.0452,
+      "step": 9093
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.103822485461188e-07,
+      "loss": 1.0262,
+      "step": 9094
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.091540162101874e-07,
+      "loss": 1.0325,
+      "step": 9095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.079266761019455e-07,
+      "loss": 1.1958,
+      "step": 9096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.067002283405356e-07,
+      "loss": 1.0979,
+      "step": 9097
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.054746730450247e-07,
+      "loss": 1.0066,
+      "step": 9098
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.042500103343831e-07,
+      "loss": 1.0916,
+      "step": 9099
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.030262403274969e-07,
+      "loss": 0.9968,
+      "step": 9100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.018033631431699e-07,
+      "loss": 0.9871,
+      "step": 9101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.005813789001171e-07,
+      "loss": 1.0237,
+      "step": 9102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.993602877169626e-07,
+      "loss": 1.1157,
+      "step": 9103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.981400897122526e-07,
+      "loss": 1.0547,
+      "step": 9104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.969207850044359e-07,
+      "loss": 1.0299,
+      "step": 9105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.957023737118841e-07,
+      "loss": 1.0492,
+      "step": 9106
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.944848559528762e-07,
+      "loss": 0.9777,
+      "step": 9107
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.932682318456042e-07,
+      "loss": 1.1919,
+      "step": 9108
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.920525015081771e-07,
+      "loss": 1.0898,
+      "step": 9109
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.90837665058618e-07,
+      "loss": 1.1895,
+      "step": 9110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.896237226148573e-07,
+      "loss": 1.0981,
+      "step": 9111
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.88410674294744e-07,
+      "loss": 1.0491,
+      "step": 9112
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.87198520216037e-07,
+      "loss": 0.9729,
+      "step": 9113
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.85987260496408e-07,
+      "loss": 1.0979,
+      "step": 9114
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.847768952534473e-07,
+      "loss": 1.0803,
+      "step": 9115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.83567424604651e-07,
+      "loss": 1.1091,
+      "step": 9116
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.823588486674338e-07,
+      "loss": 1.0461,
+      "step": 9117
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.811511675591221e-07,
+      "loss": 1.0369,
+      "step": 9118
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.799443813969532e-07,
+      "loss": 1.0996,
+      "step": 9119
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.787384902980821e-07,
+      "loss": 1.1223,
+      "step": 9120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.775334943795732e-07,
+      "loss": 1.0215,
+      "step": 9121
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.763293937584015e-07,
+      "loss": 1.056,
+      "step": 9122
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.751261885514638e-07,
+      "loss": 1.0675,
+      "step": 9123
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.739238788755599e-07,
+      "loss": 1.0605,
+      "step": 9124
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.727224648474097e-07,
+      "loss": 0.9232,
+      "step": 9125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.715219465836454e-07,
+      "loss": 0.9888,
+      "step": 9126
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.703223242008073e-07,
+      "loss": 1.0249,
+      "step": 9127
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.691235978153555e-07,
+      "loss": 1.0466,
+      "step": 9128
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.679257675436591e-07,
+      "loss": 1.0955,
+      "step": 9129
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.667288335019984e-07,
+      "loss": 1.0571,
+      "step": 9130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.655327958065728e-07,
+      "loss": 1.084,
+      "step": 9131
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.643376545734871e-07,
+      "loss": 1.1731,
+      "step": 9132
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.631434099187662e-07,
+      "loss": 0.9885,
+      "step": 9133
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.619500619583442e-07,
+      "loss": 1.1511,
+      "step": 9134
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.607576108080694e-07,
+      "loss": 0.9661,
+      "step": 9135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.595660565836993e-07,
+      "loss": 1.0182,
+      "step": 9136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.583753994009113e-07,
+      "loss": 1.1709,
+      "step": 9137
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.571856393752886e-07,
+      "loss": 1.0908,
+      "step": 9138
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.559967766223331e-07,
+      "loss": 1.0403,
+      "step": 9139
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.548088112574548e-07,
+      "loss": 1.2012,
+      "step": 9140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.536217433959802e-07,
+      "loss": 1.1514,
+      "step": 9141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.524355731531485e-07,
+      "loss": 1.0752,
+      "step": 9142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.512503006441096e-07,
+      "loss": 1.0718,
+      "step": 9143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.500659259839249e-07,
+      "loss": 1.031,
+      "step": 9144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.488824492875746e-07,
+      "loss": 1.0652,
+      "step": 9145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.476998706699445e-07,
+      "loss": 1.0977,
+      "step": 9146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.465181902458395e-07,
+      "loss": 1.0117,
+      "step": 9147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.453374081299747e-07,
+      "loss": 1.0886,
+      "step": 9148
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.441575244369759e-07,
+      "loss": 1.0127,
+      "step": 9149
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.429785392813871e-07,
+      "loss": 1.1545,
+      "step": 9150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.418004527776579e-07,
+      "loss": 1.0601,
+      "step": 9151
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.406232650401557e-07,
+      "loss": 0.9535,
+      "step": 9152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.394469761831602e-07,
+      "loss": 1.121,
+      "step": 9153
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.382715863208612e-07,
+      "loss": 0.843,
+      "step": 9154
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.37097095567364e-07,
+      "loss": 1.0315,
+      "step": 9155
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.359235040366875e-07,
+      "loss": 1.0483,
+      "step": 9156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.347508118427604e-07,
+      "loss": 1.1431,
+      "step": 9157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.335790190994218e-07,
+      "loss": 0.921,
+      "step": 9158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.324081259204319e-07,
+      "loss": 1.0547,
+      "step": 9159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.312381324194551e-07,
+      "loss": 1.0771,
+      "step": 9160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.30069038710074e-07,
+      "loss": 1.1041,
+      "step": 9161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.289008449057788e-07,
+      "loss": 1.068,
+      "step": 9162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.277335511199779e-07,
+      "loss": 1.087,
+      "step": 9163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.265671574659904e-07,
+      "loss": 1.0168,
+      "step": 9164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.25401664057046e-07,
+      "loss": 1.0029,
+      "step": 9165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.242370710062862e-07,
+      "loss": 1.1515,
+      "step": 9166
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.230733784267708e-07,
+      "loss": 1.0701,
+      "step": 9167
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.219105864314646e-07,
+      "loss": 1.0391,
+      "step": 9168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.207486951332543e-07,
+      "loss": 1.0938,
+      "step": 9169
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.195877046449285e-07,
+      "loss": 1.066,
+      "step": 9170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.184276150791969e-07,
+      "loss": 1.0378,
+      "step": 9171
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.172684265486796e-07,
+      "loss": 0.9293,
+      "step": 9172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.161101391659043e-07,
+      "loss": 1.0435,
+      "step": 9173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.149527530433153e-07,
+      "loss": 1.1157,
+      "step": 9174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.137962682932742e-07,
+      "loss": 1.0935,
+      "step": 9175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.126406850280432e-07,
+      "loss": 1.0881,
+      "step": 9176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.114860033598103e-07,
+      "loss": 1.1473,
+      "step": 9177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.103322234006649e-07,
+      "loss": 0.9752,
+      "step": 9178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.091793452626162e-07,
+      "loss": 1.0645,
+      "step": 9179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.080273690575823e-07,
+      "loss": 1.0391,
+      "step": 9180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.068762948973928e-07,
+      "loss": 1.129,
+      "step": 9181
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.057261228937939e-07,
+      "loss": 1.1008,
+      "step": 9182
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.045768531584418e-07,
+      "loss": 1.071,
+      "step": 9183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.034284858029027e-07,
+      "loss": 1.1602,
+      "step": 9184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.022810209386611e-07,
+      "loss": 1.1141,
+      "step": 9185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.011344586771073e-07,
+      "loss": 1.1274,
+      "step": 9186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.999887991295506e-07,
+      "loss": 1.0652,
+      "step": 9187
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.988440424072063e-07,
+      "loss": 1.1561,
+      "step": 9188
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.977001886212043e-07,
+      "loss": 1.0757,
+      "step": 9189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.965572378825891e-07,
+      "loss": 1.0917,
+      "step": 9190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.954151903023176e-07,
+      "loss": 1.1245,
+      "step": 9191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.942740459912534e-07,
+      "loss": 1.0681,
+      "step": 9192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.931338050601799e-07,
+      "loss": 1.072,
+      "step": 9193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.919944676197876e-07,
+      "loss": 0.9579,
+      "step": 9194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.908560337806802e-07,
+      "loss": 1.1115,
+      "step": 9195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.89718503653376e-07,
+      "loss": 1.1001,
+      "step": 9196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.885818773483033e-07,
+      "loss": 0.9597,
+      "step": 9197
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.874461549758015e-07,
+      "loss": 1.0562,
+      "step": 9198
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.863113366461294e-07,
+      "loss": 1.1129,
+      "step": 9199
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.851774224694463e-07,
+      "loss": 1.0212,
+      "step": 9200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.840444125558354e-07,
+      "loss": 0.9561,
+      "step": 9201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.829123070152843e-07,
+      "loss": 1.0684,
+      "step": 9202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.817811059576939e-07,
+      "loss": 1.0315,
+      "step": 9203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.806508094928833e-07,
+      "loss": 1.1458,
+      "step": 9204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.795214177305743e-07,
+      "loss": 1.1125,
+      "step": 9205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.783929307804082e-07,
+      "loss": 1.0237,
+      "step": 9206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.772653487519377e-07,
+      "loss": 1.0739,
+      "step": 9207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.761386717546226e-07,
+      "loss": 1.0718,
+      "step": 9208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.750128998978422e-07,
+      "loss": 1.1626,
+      "step": 9209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.738880332908815e-07,
+      "loss": 0.9734,
+      "step": 9210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.727640720429395e-07,
+      "loss": 1.0859,
+      "step": 9211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.716410162631281e-07,
+      "loss": 0.9523,
+      "step": 9212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.705188660604745e-07,
+      "loss": 1.2224,
+      "step": 9213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.693976215439102e-07,
+      "loss": 1.0513,
+      "step": 9214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.682772828222872e-07,
+      "loss": 1.063,
+      "step": 9215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.67157850004363e-07,
+      "loss": 1.0842,
+      "step": 9216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.660393231988094e-07,
+      "loss": 1.0574,
+      "step": 9217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.64921702514213e-07,
+      "loss": 0.9958,
+      "step": 9218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.638049880590658e-07,
+      "loss": 1.1174,
+      "step": 9219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.626891799417801e-07,
+      "loss": 1.0593,
+      "step": 9220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.615742782706769e-07,
+      "loss": 0.9617,
+      "step": 9221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.604602831539841e-07,
+      "loss": 1.0422,
+      "step": 9222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.593471946998497e-07,
+      "loss": 0.955,
+      "step": 9223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.582350130163295e-07,
+      "loss": 1.0972,
+      "step": 9224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.571237382113882e-07,
+      "loss": 0.9783,
+      "step": 9225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.560133703929106e-07,
+      "loss": 1.0967,
+      "step": 9226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.54903909668686e-07,
+      "loss": 0.9612,
+      "step": 9227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.537953561464183e-07,
+      "loss": 1.0833,
+      "step": 9228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.52687709933727e-07,
+      "loss": 1.0869,
+      "step": 9229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.515809711381349e-07,
+      "loss": 1.0039,
+      "step": 9230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.504751398670872e-07,
+      "loss": 1.1068,
+      "step": 9231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.493702162279325e-07,
+      "loss": 1.1318,
+      "step": 9232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.482662003279327e-07,
+      "loss": 1.0321,
+      "step": 9233
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.471630922742678e-07,
+      "loss": 1.0637,
+      "step": 9234
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.460608921740208e-07,
+      "loss": 1.0029,
+      "step": 9235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.449596001341929e-07,
+      "loss": 1.0513,
+      "step": 9236
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.438592162616963e-07,
+      "loss": 1.1267,
+      "step": 9237
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.427597406633534e-07,
+      "loss": 0.835,
+      "step": 9238
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.416611734458977e-07,
+      "loss": 0.9813,
+      "step": 9239
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.405635147159772e-07,
+      "loss": 1.0925,
+      "step": 9240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.394667645801478e-07,
+      "loss": 1.0825,
+      "step": 9241
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.383709231448832e-07,
+      "loss": 0.9409,
+      "step": 9242
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.372759905165615e-07,
+      "loss": 0.9958,
+      "step": 9243
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.36181966801479e-07,
+      "loss": 1.0803,
+      "step": 9244
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.350888521058429e-07,
+      "loss": 1.1199,
+      "step": 9245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.339966465357661e-07,
+      "loss": 1.0176,
+      "step": 9246
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.329053501972793e-07,
+      "loss": 1.0901,
+      "step": 9247
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.318149631963244e-07,
+      "loss": 1.0081,
+      "step": 9248
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.307254856387524e-07,
+      "loss": 0.9672,
+      "step": 9249
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.296369176303297e-07,
+      "loss": 1.0388,
+      "step": 9250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.285492592767284e-07,
+      "loss": 1.0107,
+      "step": 9251
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.274625106835375e-07,
+      "loss": 1.1365,
+      "step": 9252
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.263766719562614e-07,
+      "loss": 1.053,
+      "step": 9253
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.252917432003025e-07,
+      "loss": 1.0299,
+      "step": 9254
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.242077245209877e-07,
+      "loss": 1.0869,
+      "step": 9255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.231246160235516e-07,
+      "loss": 0.9738,
+      "step": 9256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.22042417813139e-07,
+      "loss": 1.0464,
+      "step": 9257
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.209611299948082e-07,
+      "loss": 0.9587,
+      "step": 9258
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.198807526735273e-07,
+      "loss": 0.9907,
+      "step": 9259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.188012859541781e-07,
+      "loss": 0.9944,
+      "step": 9260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.177227299415534e-07,
+      "loss": 1.0728,
+      "step": 9261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.166450847403538e-07,
+      "loss": 1.0688,
+      "step": 9262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.155683504551968e-07,
+      "loss": 0.9773,
+      "step": 9263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.144925271906122e-07,
+      "loss": 1.0188,
+      "step": 9264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.13417615051034e-07,
+      "loss": 1.1108,
+      "step": 9265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.123436141408168e-07,
+      "loss": 0.9951,
+      "step": 9266
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.112705245642192e-07,
+      "loss": 1.0383,
+      "step": 9267
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.101983464254135e-07,
+      "loss": 0.9548,
+      "step": 9268
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.091270798284887e-07,
+      "loss": 1.208,
+      "step": 9269
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.080567248774372e-07,
+      "loss": 1.0186,
+      "step": 9270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.06987281676168e-07,
+      "loss": 1.0728,
+      "step": 9271
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.059187503285013e-07,
+      "loss": 1.1396,
+      "step": 9272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.048511309381666e-07,
+      "loss": 1.0598,
+      "step": 9273
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.037844236088076e-07,
+      "loss": 1.1215,
+      "step": 9274
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.027186284439768e-07,
+      "loss": 1.0621,
+      "step": 9275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.016537455471383e-07,
+      "loss": 0.9817,
+      "step": 9276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.005897750216705e-07,
+      "loss": 1.0981,
+      "step": 9277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.995267169708608e-07,
+      "loss": 0.9052,
+      "step": 9278
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.984645714979076e-07,
+      "loss": 1.1045,
+      "step": 9279
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.974033387059242e-07,
+      "loss": 1.0898,
+      "step": 9280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.963430186979291e-07,
+      "loss": 1.0835,
+      "step": 9281
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.952836115768612e-07,
+      "loss": 1.0613,
+      "step": 9282
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.942251174455604e-07,
+      "loss": 0.9231,
+      "step": 9283
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.931675364067846e-07,
+      "loss": 1.0732,
+      "step": 9284
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.921108685632026e-07,
+      "loss": 0.9448,
+      "step": 9285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.910551140173926e-07,
+      "loss": 1.0598,
+      "step": 9286
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.900002728718446e-07,
+      "loss": 1.1836,
+      "step": 9287
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.889463452289623e-07,
+      "loss": 0.9902,
+      "step": 9288
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.878933311910572e-07,
+      "loss": 1.0764,
+      "step": 9289
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.868412308603533e-07,
+      "loss": 1.2114,
+      "step": 9290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.857900443389864e-07,
+      "loss": 0.9613,
+      "step": 9291
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.847397717290037e-07,
+      "loss": 1.0994,
+      "step": 9292
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.836904131323628e-07,
+      "loss": 1.0533,
+      "step": 9293
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.826419686509354e-07,
+      "loss": 0.959,
+      "step": 9294
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.815944383864991e-07,
+      "loss": 1.0027,
+      "step": 9295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.805478224407501e-07,
+      "loss": 1.0216,
+      "step": 9296
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.795021209152874e-07,
+      "loss": 0.9928,
+      "step": 9297
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.784573339116273e-07,
+      "loss": 1.1436,
+      "step": 9298
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.774134615311955e-07,
+      "loss": 1.0782,
+      "step": 9299
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.763705038753276e-07,
+      "loss": 1.1404,
+      "step": 9300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.753284610452725e-07,
+      "loss": 0.9081,
+      "step": 9301
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.742873331421917e-07,
+      "loss": 1.1115,
+      "step": 9302
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.73247120267153e-07,
+      "loss": 1.1377,
+      "step": 9303
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.722078225211391e-07,
+      "loss": 1.1199,
+      "step": 9304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.711694400050427e-07,
+      "loss": 1.052,
+      "step": 9305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.701319728196675e-07,
+      "loss": 1.1029,
+      "step": 9306
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.690954210657285e-07,
+      "loss": 1.0693,
+      "step": 9307
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.680597848438518e-07,
+      "loss": 1.0969,
+      "step": 9308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.670250642545739e-07,
+      "loss": 1.1199,
+      "step": 9309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.659912593983463e-07,
+      "loss": 0.8779,
+      "step": 9310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.649583703755257e-07,
+      "loss": 1.0691,
+      "step": 9311
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.639263972863851e-07,
+      "loss": 1.0433,
+      "step": 9312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.628953402311043e-07,
+      "loss": 1.0176,
+      "step": 9313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.618651993097757e-07,
+      "loss": 1.0712,
+      "step": 9314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.608359746224057e-07,
+      "loss": 1.0112,
+      "step": 9315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.598076662689067e-07,
+      "loss": 1.0035,
+      "step": 9316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.587802743491044e-07,
+      "loss": 1.0814,
+      "step": 9317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.577537989627402e-07,
+      "loss": 1.2607,
+      "step": 9318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.567282402094575e-07,
+      "loss": 0.9518,
+      "step": 9319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.55703598188817e-07,
+      "loss": 0.9775,
+      "step": 9320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.54679873000289e-07,
+      "loss": 1.0393,
+      "step": 9321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.536570647432527e-07,
+      "loss": 1.0482,
+      "step": 9322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.526351735170032e-07,
+      "loss": 1.0273,
+      "step": 9323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.516141994207414e-07,
+      "loss": 1.0803,
+      "step": 9324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.50594142553581e-07,
+      "loss": 1.0894,
+      "step": 9325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.495750030145508e-07,
+      "loss": 1.0249,
+      "step": 9326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.485567809025805e-07,
+      "loss": 1.0232,
+      "step": 9327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.47539476316521e-07,
+      "loss": 0.9534,
+      "step": 9328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.465230893551299e-07,
+      "loss": 1.0278,
+      "step": 9329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.455076201170739e-07,
+      "loss": 1.1257,
+      "step": 9330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.444930687009353e-07,
+      "loss": 1.2163,
+      "step": 9331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.43479435205202e-07,
+      "loss": 1.1785,
+      "step": 9332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.424667197282751e-07,
+      "loss": 1.0543,
+      "step": 9333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.414549223684729e-07,
+      "loss": 1.0649,
+      "step": 9334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.404440432240099e-07,
+      "loss": 0.9373,
+      "step": 9335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.394340823930244e-07,
+      "loss": 0.8601,
+      "step": 9336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.384250399735624e-07,
+      "loss": 1.0535,
+      "step": 9337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.374169160635767e-07,
+      "loss": 1.0898,
+      "step": 9338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.364097107609356e-07,
+      "loss": 0.9796,
+      "step": 9339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.354034241634154e-07,
+      "loss": 1.0068,
+      "step": 9340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.343980563687068e-07,
+      "loss": 1.0258,
+      "step": 9341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.333936074744062e-07,
+      "loss": 1.104,
+      "step": 9342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.323900775780221e-07,
+      "loss": 1.0315,
+      "step": 9343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.313874667769781e-07,
+      "loss": 1.1458,
+      "step": 9344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.303857751686048e-07,
+      "loss": 0.9897,
+      "step": 9345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.293850028501424e-07,
+      "loss": 1.0889,
+      "step": 9346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.283851499187475e-07,
+      "loss": 1.0767,
+      "step": 9347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.273862164714805e-07,
+      "loss": 1.1096,
+      "step": 9348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.263882026053157e-07,
+      "loss": 1.0967,
+      "step": 9349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.253911084171381e-07,
+      "loss": 1.0693,
+      "step": 9350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.243949340037469e-07,
+      "loss": 1.083,
+      "step": 9351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.233996794618445e-07,
+      "loss": 0.9614,
+      "step": 9352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.224053448880506e-07,
+      "loss": 0.9857,
+      "step": 9353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.214119303788901e-07,
+      "loss": 1.0305,
+      "step": 9354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.204194360308046e-07,
+      "loss": 1.0479,
+      "step": 9355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.19427861940145e-07,
+      "loss": 1.062,
+      "step": 9356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.184372082031653e-07,
+      "loss": 1.093,
+      "step": 9357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.174474749160396e-07,
+      "loss": 1.0818,
+      "step": 9358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.164586621748502e-07,
+      "loss": 1.0579,
+      "step": 9359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.154707700755857e-07,
+      "loss": 1.1375,
+      "step": 9360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.144837987141526e-07,
+      "loss": 1.0647,
+      "step": 9361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.13497748186359e-07,
+      "loss": 1.0522,
+      "step": 9362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.125126185879336e-07,
+      "loss": 1.1072,
+      "step": 9363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.115284100145079e-07,
+      "loss": 1.0508,
+      "step": 9364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.105451225616265e-07,
+      "loss": 1.0343,
+      "step": 9365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.095627563247452e-07,
+      "loss": 0.9822,
+      "step": 9366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.08581311399231e-07,
+      "loss": 1.2219,
+      "step": 9367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.076007878803601e-07,
+      "loss": 0.9052,
+      "step": 9368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.066211858633196e-07,
+      "loss": 1.1414,
+      "step": 9369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.056425054432068e-07,
+      "loss": 0.9558,
+      "step": 9370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.046647467150289e-07,
+      "loss": 0.99,
+      "step": 9371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.03687909773708e-07,
+      "loss": 0.9395,
+      "step": 9372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.027119947140691e-07,
+      "loss": 1.0869,
+      "step": 9373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.017370016308542e-07,
+      "loss": 1.0022,
+      "step": 9374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.007629306187145e-07,
+      "loss": 1.0044,
+      "step": 9375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.997897817722075e-07,
+      "loss": 1.0942,
+      "step": 9376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.988175551858088e-07,
+      "loss": 1.0481,
+      "step": 9377
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.978462509538972e-07,
+      "loss": 1.0464,
+      "step": 9378
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.968758691707643e-07,
+      "loss": 1.0859,
+      "step": 9379
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.959064099306155e-07,
+      "loss": 1.0325,
+      "step": 9380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9493787332756e-07,
+      "loss": 1.0408,
+      "step": 9381
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.93970259455624e-07,
+      "loss": 1.082,
+      "step": 9382
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.93003568408742e-07,
+      "loss": 0.9463,
+      "step": 9383
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.920378002807558e-07,
+      "loss": 0.9484,
+      "step": 9384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.910729551654214e-07,
+      "loss": 1.0132,
+      "step": 9385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.901090331564051e-07,
+      "loss": 1.0952,
+      "step": 9386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.891460343472798e-07,
+      "loss": 0.9757,
+      "step": 9387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.88183958831534e-07,
+      "loss": 1.0037,
+      "step": 9388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.872228067025608e-07,
+      "loss": 1.1118,
+      "step": 9389
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8626257805367e-07,
+      "loss": 1.0356,
+      "step": 9390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853032729780771e-07,
+      "loss": 0.9987,
+      "step": 9391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.843448915689097e-07,
+      "loss": 1.0806,
+      "step": 9392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.833874339192046e-07,
+      "loss": 1.0974,
+      "step": 9393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.824309001219107e-07,
+      "loss": 0.8832,
+      "step": 9394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.814752902698861e-07,
+      "loss": 1.0493,
+      "step": 9395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.805206044558985e-07,
+      "loss": 1.0381,
+      "step": 9396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.795668427726275e-07,
+      "loss": 0.9475,
+      "step": 9397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.78614005312662e-07,
+      "loss": 1.0798,
+      "step": 9398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.776620921685027e-07,
+      "loss": 1.0593,
+      "step": 9399
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7671110343255777e-07,
+      "loss": 1.0227,
+      "step": 9400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.757610391971468e-07,
+      "loss": 1.1367,
+      "step": 9401
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7481189955450256e-07,
+      "loss": 1.1953,
+      "step": 9402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7386368459676145e-07,
+      "loss": 1.0474,
+      "step": 9403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.729163944159776e-07,
+      "loss": 1.0059,
+      "step": 9404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7197002910411073e-07,
+      "loss": 1.1636,
+      "step": 9405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7102458875303067e-07,
+      "loss": 1.1083,
+      "step": 9406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.70080073454523e-07,
+      "loss": 1.0345,
+      "step": 9407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6913648330027427e-07,
+      "loss": 1.1177,
+      "step": 9408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.681938183818879e-07,
+      "loss": 0.9989,
+      "step": 9409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6725207879087744e-07,
+      "loss": 1.0352,
+      "step": 9410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6631126461866183e-07,
+      "loss": 1.0974,
+      "step": 9411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.65371375956577e-07,
+      "loss": 1.1497,
+      "step": 9412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6443241289586106e-07,
+      "loss": 1.0715,
+      "step": 9413
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6349437552767105e-07,
+      "loss": 1.0239,
+      "step": 9414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.625572639430664e-07,
+      "loss": 1.0439,
+      "step": 9415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6162107823302217e-07,
+      "loss": 1.085,
+      "step": 9416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.606858184884178e-07,
+      "loss": 1.0793,
+      "step": 9417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5975148480005174e-07,
+      "loss": 1.0481,
+      "step": 9418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.588180772586215e-07,
+      "loss": 1.0198,
+      "step": 9419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5788559595474233e-07,
+      "loss": 1.0251,
+      "step": 9420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.569540409789397e-07,
+      "loss": 1.1106,
+      "step": 9421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.560234124216456e-07,
+      "loss": 0.9176,
+      "step": 9422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.550937103732012e-07,
+      "loss": 1.0952,
+      "step": 9423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5416493492386306e-07,
+      "loss": 0.9988,
+      "step": 9424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5323708616379247e-07,
+      "loss": 1.0321,
+      "step": 9425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.523101641830663e-07,
+      "loss": 1.0062,
+      "step": 9426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.513841690716636e-07,
+      "loss": 1.1265,
+      "step": 9427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.504591009194803e-07,
+      "loss": 1.0957,
+      "step": 9428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4953495981632236e-07,
+      "loss": 1.0513,
+      "step": 9429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.48611745851899e-07,
+      "loss": 1.0542,
+      "step": 9430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.476894591158365e-07,
+      "loss": 1.0708,
+      "step": 9431
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4676809969766865e-07,
+      "loss": 1.0507,
+      "step": 9432
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.458476676868373e-07,
+      "loss": 0.9642,
+      "step": 9433
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.449281631726976e-07,
+      "loss": 0.9564,
+      "step": 9434
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.440095862445115e-07,
+      "loss": 1.1721,
+      "step": 9435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4309193699145547e-07,
+      "loss": 1.0216,
+      "step": 9436
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4217521550261046e-07,
+      "loss": 1.0447,
+      "step": 9437
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.412594218669697e-07,
+      "loss": 1.0409,
+      "step": 9438
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.403445561734365e-07,
+      "loss": 0.9855,
+      "step": 9439
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.394306185108266e-07,
+      "loss": 1.0911,
+      "step": 9440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3851760896786e-07,
+      "loss": 1.0219,
+      "step": 9441
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.376055276331714e-07,
+      "loss": 1.0845,
+      "step": 9442
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.366943745953023e-07,
+      "loss": 1.1224,
+      "step": 9443
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3578414994270846e-07,
+      "loss": 0.9856,
+      "step": 9444
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.348748537637504e-07,
+      "loss": 1.0349,
+      "step": 9445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3396648614669856e-07,
+      "loss": 1.132,
+      "step": 9446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3305904717973803e-07,
+      "loss": 0.962,
+      "step": 9447
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3215253695096273e-07,
+      "loss": 1.0818,
+      "step": 9448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3124695554837005e-07,
+      "loss": 1.0516,
+      "step": 9449
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3034230305987635e-07,
+      "loss": 1.0125,
+      "step": 9450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.294385795733014e-07,
+      "loss": 1.0825,
+      "step": 9451
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2853578517637604e-07,
+      "loss": 1.0439,
+      "step": 9452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.276339199567425e-07,
+      "loss": 1.0026,
+      "step": 9453
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2673298400195075e-07,
+      "loss": 1.0181,
+      "step": 9454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2583297739946407e-07,
+      "loss": 1.1057,
+      "step": 9455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2493390023665147e-07,
+      "loss": 1.1467,
+      "step": 9456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.240357526007932e-07,
+      "loss": 1.0708,
+      "step": 9457
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2313853457908174e-07,
+      "loss": 0.9456,
+      "step": 9458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.222422462586151e-07,
+      "loss": 0.9875,
+      "step": 9459
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.213468877264026e-07,
+      "loss": 1.0403,
+      "step": 9460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.204524590693659e-07,
+      "loss": 1.0348,
+      "step": 9461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.195589603743311e-07,
+      "loss": 1.0695,
+      "step": 9462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.186663917280398e-07,
+      "loss": 1.1458,
+      "step": 9463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1777475321713943e-07,
+      "loss": 0.9962,
+      "step": 9464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.168840449281897e-07,
+      "loss": 0.9747,
+      "step": 9465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.159942669476569e-07,
+      "loss": 1.012,
+      "step": 9466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1510541936192084e-07,
+      "loss": 1.093,
+      "step": 9467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.142175022572659e-07,
+      "loss": 1.0232,
+      "step": 9468
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.13330515719893e-07,
+      "loss": 0.9519,
+      "step": 9469
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1244445983590456e-07,
+      "loss": 0.974,
+      "step": 9470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1155933469131937e-07,
+      "loss": 0.9431,
+      "step": 9471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1067514037206544e-07,
+      "loss": 0.9575,
+      "step": 9472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.097918769639764e-07,
+      "loss": 1.0471,
+      "step": 9473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0890954455279574e-07,
+      "loss": 0.9929,
+      "step": 9474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.080281432241817e-07,
+      "loss": 1.041,
+      "step": 9475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0714767306369584e-07,
+      "loss": 1.0168,
+      "step": 9476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0626813415681534e-07,
+      "loss": 1.1113,
+      "step": 9477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.053895265889218e-07,
+      "loss": 1.0203,
+      "step": 9478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0451185044530825e-07,
+      "loss": 1.0591,
+      "step": 9479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.036351058111809e-07,
+      "loss": 0.9894,
+      "step": 9480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0275929277164946e-07,
+      "loss": 1.1746,
+      "step": 9481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0188441141173484e-07,
+      "loss": 1.1539,
+      "step": 9482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0101046181637236e-07,
+      "loss": 1.0852,
+      "step": 9483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.001374440703998e-07,
+      "loss": 1.0364,
+      "step": 9484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9926535825856925e-07,
+      "loss": 1.1023,
+      "step": 9485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.98394204465542e-07,
+      "loss": 1.0781,
+      "step": 9486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.975239827758848e-07,
+      "loss": 1.048,
+      "step": 9487
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9665469327408003e-07,
+      "loss": 1.0651,
+      "step": 9488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.957863360445158e-07,
+      "loss": 0.9456,
+      "step": 9489
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.949189111714891e-07,
+      "loss": 0.9413,
+      "step": 9490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.940524187392092e-07,
+      "loss": 1.0439,
+      "step": 9491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.931868588317911e-07,
+      "loss": 1.0598,
+      "step": 9492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.92322231533262e-07,
+      "loss": 1.063,
+      "step": 9493
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.914585369275614e-07,
+      "loss": 1.1267,
+      "step": 9494
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.905957750985323e-07,
+      "loss": 1.0477,
+      "step": 9495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8973394612992877e-07,
+      "loss": 1.0974,
+      "step": 9496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.888730501054172e-07,
+      "loss": 1.0325,
+      "step": 9497
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.880130871085708e-07,
+      "loss": 1.1193,
+      "step": 9498
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.871540572228738e-07,
+      "loss": 1.1475,
+      "step": 9499
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8629596053171627e-07,
+      "loss": 1.0073,
+      "step": 9500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8543879711840257e-07,
+      "loss": 1.0505,
+      "step": 9501
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8458256706614625e-07,
+      "loss": 0.9711,
+      "step": 9502
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8372727045806526e-07,
+      "loss": 1.0759,
+      "step": 9503
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8287290737718975e-07,
+      "loss": 1.0619,
+      "step": 9504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.820194779064612e-07,
+      "loss": 1.0471,
+      "step": 9505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.811669821287278e-07,
+      "loss": 1.0847,
+      "step": 9506
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.803154201267489e-07,
+      "loss": 1.0518,
+      "step": 9507
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7946479198319063e-07,
+      "loss": 0.9814,
+      "step": 9508
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.786150977806302e-07,
+      "loss": 1.1138,
+      "step": 9509
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7776633760155944e-07,
+      "loss": 1.1133,
+      "step": 9510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7691851152836577e-07,
+      "loss": 1.0525,
+      "step": 9511
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.760716196433589e-07,
+      "loss": 1.0505,
+      "step": 9512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.752256620287542e-07,
+      "loss": 1.0469,
+      "step": 9513
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7438063876667375e-07,
+      "loss": 1.0439,
+      "step": 9514
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7353654993915077e-07,
+      "loss": 1.0088,
+      "step": 9515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7269339562812646e-07,
+      "loss": 1.068,
+      "step": 9516
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.718511759154553e-07,
+      "loss": 1.1116,
+      "step": 9517
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.710098908828963e-07,
+      "loss": 1.0732,
+      "step": 9518
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7016954061211975e-07,
+      "loss": 0.9641,
+      "step": 9519
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.693301251847037e-07,
+      "loss": 1.0803,
+      "step": 9520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.684916446821407e-07,
+      "loss": 1.0575,
+      "step": 9521
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.676540991858235e-07,
+      "loss": 0.9861,
+      "step": 9522
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.668174887770648e-07,
+      "loss": 1.105,
+      "step": 9523
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.659818135370763e-07,
+      "loss": 1.0156,
+      "step": 9524
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6514707354698644e-07,
+      "loss": 1.0837,
+      "step": 9525
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6431326888782924e-07,
+      "loss": 1.0409,
+      "step": 9526
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.634803996405467e-07,
+      "loss": 1.0183,
+      "step": 9527
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.62648465885993e-07,
+      "loss": 1.103,
+      "step": 9528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6181746770493355e-07,
+      "loss": 0.8951,
+      "step": 9529
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6098740517803596e-07,
+      "loss": 1.1104,
+      "step": 9530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6015827838588366e-07,
+      "loss": 0.9967,
+      "step": 9531
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.593300874089656e-07,
+      "loss": 1.1028,
+      "step": 9532
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5850283232767844e-07,
+      "loss": 1.0732,
+      "step": 9533
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5767651322233367e-07,
+      "loss": 1.1677,
+      "step": 9534
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5685113017314697e-07,
+      "loss": 0.9052,
+      "step": 9535
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.560266832602444e-07,
+      "loss": 0.9719,
+      "step": 9536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.55203172563664e-07,
+      "loss": 1.0574,
+      "step": 9537
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5438059816334636e-07,
+      "loss": 1.033,
+      "step": 9538
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.535589601391498e-07,
+      "loss": 1.012,
+      "step": 9539
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.527382585708339e-07,
+      "loss": 1.0376,
+      "step": 9540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.519184935380704e-07,
+      "loss": 1.1045,
+      "step": 9541
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5109966512044344e-07,
+      "loss": 1.1392,
+      "step": 9542
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.502817733974395e-07,
+      "loss": 0.9767,
+      "step": 9543
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.494648184484606e-07,
+      "loss": 1.1191,
+      "step": 9544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4864880035281323e-07,
+      "loss": 0.9739,
+      "step": 9545
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.478337191897152e-07,
+      "loss": 1.1245,
+      "step": 9546
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470195750382943e-07,
+      "loss": 1.0604,
+      "step": 9547
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46206367977584e-07,
+      "loss": 1.0688,
+      "step": 9548
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.453940980865289e-07,
+      "loss": 1.041,
+      "step": 9549
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.445827654439815e-07,
+      "loss": 1.1802,
+      "step": 9550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.437723701287077e-07,
+      "loss": 1.1047,
+      "step": 9551
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4296291221937563e-07,
+      "loss": 1.0416,
+      "step": 9552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.42154391794568e-07,
+      "loss": 1.1187,
+      "step": 9553
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4134680893277315e-07,
+      "loss": 1.0295,
+      "step": 9554
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.405401637123873e-07,
+      "loss": 1.0845,
+      "step": 9555
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.397344562117222e-07,
+      "loss": 1.1958,
+      "step": 9556
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3892968650899084e-07,
+      "loss": 0.9603,
+      "step": 9557
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3812585468231963e-07,
+      "loss": 1.0709,
+      "step": 9558
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.373229608097439e-07,
+      "loss": 1.0305,
+      "step": 9559
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.365210049692047e-07,
+      "loss": 1.1675,
+      "step": 9560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3571998723855634e-07,
+      "loss": 1.0637,
+      "step": 9561
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3491990769556e-07,
+      "loss": 1.0408,
+      "step": 9562
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.341207664178825e-07,
+      "loss": 0.8917,
+      "step": 9563
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.333225634831072e-07,
+      "loss": 1.1685,
+      "step": 9564
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3252529896871777e-07,
+      "loss": 1.1133,
+      "step": 9565
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3172897295211225e-07,
+      "loss": 1.0867,
+      "step": 9566
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3093358551059887e-07,
+      "loss": 0.8851,
+      "step": 9567
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3013913672138906e-07,
+      "loss": 1.1404,
+      "step": 9568
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2934562666160794e-07,
+      "loss": 1.0371,
+      "step": 9569
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2855305540828717e-07,
+      "loss": 0.9965,
+      "step": 9570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2776142303836746e-07,
+      "loss": 1.1146,
+      "step": 9571
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.269707296286995e-07,
+      "loss": 1.0863,
+      "step": 9572
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2618097525604076e-07,
+      "loss": 0.9301,
+      "step": 9573
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2539215999706e-07,
+      "loss": 0.9958,
+      "step": 9574
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.246042839283348e-07,
+      "loss": 1.0787,
+      "step": 9575
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.238173471263495e-07,
+      "loss": 1.0364,
+      "step": 9576
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.230313496674975e-07,
+      "loss": 1.0156,
+      "step": 9577
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.222462916280822e-07,
+      "loss": 1.0278,
+      "step": 9578
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2146217308431484e-07,
+      "loss": 1.0969,
+      "step": 9579
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2067899411231783e-07,
+      "loss": 0.9082,
+      "step": 9580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1989675478811824e-07,
+      "loss": 0.9477,
+      "step": 9581
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1911545518765517e-07,
+      "loss": 1.1428,
+      "step": 9582
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.183350953867781e-07,
+      "loss": 1.0308,
+      "step": 9583
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.175556754612374e-07,
+      "loss": 0.9641,
+      "step": 9584
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1677719548670047e-07,
+      "loss": 0.9926,
+      "step": 9585
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.159996555387412e-07,
+      "loss": 0.9167,
+      "step": 9586
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.152230556928393e-07,
+      "loss": 1.0642,
+      "step": 9587
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1444739602438877e-07,
+      "loss": 0.9955,
+      "step": 9588
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1367267660868507e-07,
+      "loss": 1.0742,
+      "step": 9589
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1289889752093905e-07,
+      "loss": 1.1003,
+      "step": 9590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1212605883626734e-07,
+      "loss": 1.1636,
+      "step": 9591
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1135416062969435e-07,
+      "loss": 0.9532,
+      "step": 9592
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.105832029761535e-07,
+      "loss": 1.0121,
+      "step": 9593
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0981318595049046e-07,
+      "loss": 0.9829,
+      "step": 9594
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0904410962745437e-07,
+      "loss": 0.9829,
+      "step": 9595
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.082759740817065e-07,
+      "loss": 0.9993,
+      "step": 9596
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.075087793878162e-07,
+      "loss": 0.9692,
+      "step": 9597
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0674252562026053e-07,
+      "loss": 1.1272,
+      "step": 9598
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.059772128534266e-07,
+      "loss": 1.0188,
+      "step": 9599
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.052128411616073e-07,
+      "loss": 1.0088,
+      "step": 9600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.044494106190077e-07,
+      "loss": 1.0374,
+      "step": 9601
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0368692129974176e-07,
+      "loss": 1.063,
+      "step": 9602
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.029253732778259e-07,
+      "loss": 1.0126,
+      "step": 9603
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.021647666271943e-07,
+      "loss": 1.0396,
+      "step": 9604
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0140510142168234e-07,
+      "loss": 0.9966,
+      "step": 9605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0064637773503547e-07,
+      "loss": 0.989,
+      "step": 9606
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9988859564091257e-07,
+      "loss": 0.9646,
+      "step": 9607
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9913175521287475e-07,
+      "loss": 0.9978,
+      "step": 9608
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.983758565243944e-07,
+      "loss": 1.1179,
+      "step": 9609
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.97620899648855e-07,
+      "loss": 1.046,
+      "step": 9610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.968668846595435e-07,
+      "loss": 1.0947,
+      "step": 9611
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9611381162966024e-07,
+      "loss": 1.0831,
+      "step": 9612
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.953616806323112e-07,
+      "loss": 1.0649,
+      "step": 9613
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.94610491740509e-07,
+      "loss": 1.0789,
+      "step": 9614
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9386024502718214e-07,
+      "loss": 1.0278,
+      "step": 9615
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.931109405651589e-07,
+      "loss": 1.0559,
+      "step": 9616
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.923625784271822e-07,
+      "loss": 1.0205,
+      "step": 9617
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9161515868590285e-07,
+      "loss": 1.0938,
+      "step": 9618
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.90868681413875e-07,
+      "loss": 0.9647,
+      "step": 9619
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9012314668356854e-07,
+      "loss": 0.9858,
+      "step": 9620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8937855456735665e-07,
+      "loss": 1.0889,
+      "step": 9621
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.886349051375215e-07,
+      "loss": 1.0679,
+      "step": 9622
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.878921984662575e-07,
+      "loss": 1.0171,
+      "step": 9623
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.871504346256637e-07,
+      "loss": 1.0964,
+      "step": 9624
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.864096136877492e-07,
+      "loss": 0.9843,
+      "step": 9625
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.856697357244331e-07,
+      "loss": 1.1125,
+      "step": 9626
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.849308008075369e-07,
+      "loss": 1.1189,
+      "step": 9627
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8419280900879975e-07,
+      "loss": 1.1304,
+      "step": 9628
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.834557603998611e-07,
+      "loss": 1.0415,
+      "step": 9629
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.827196550522726e-07,
+      "loss": 1.1057,
+      "step": 9630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8198449303749374e-07,
+      "loss": 1.0916,
+      "step": 9631
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.81250274426893e-07,
+      "loss": 0.9816,
+      "step": 9632
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.805169992917467e-07,
+      "loss": 0.991,
+      "step": 9633
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.797846677032401e-07,
+      "loss": 0.9467,
+      "step": 9634
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.790532797324663e-07,
+      "loss": 1.043,
+      "step": 9635
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7832283545042415e-07,
+      "loss": 1.1257,
+      "step": 9636
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.77593334928028e-07,
+      "loss": 1.0647,
+      "step": 9637
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7686477823609225e-07,
+      "loss": 0.8574,
+      "step": 9638
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7613716544534596e-07,
+      "loss": 0.9741,
+      "step": 9639
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.754104966264248e-07,
+      "loss": 1.0708,
+      "step": 9640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.746847718498691e-07,
+      "loss": 1.0813,
+      "step": 9641
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.739599911861346e-07,
+      "loss": 1.0581,
+      "step": 9642
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.732361547055795e-07,
+      "loss": 1.1155,
+      "step": 9643
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.72513262478471e-07,
+      "loss": 0.9841,
+      "step": 9644
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7179131457498844e-07,
+      "loss": 1.0386,
+      "step": 9645
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7107031106521573e-07,
+      "loss": 1.0178,
+      "step": 9646
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.703502520191448e-07,
+      "loss": 1.05,
+      "step": 9647
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6963113750668066e-07,
+      "loss": 1.1028,
+      "step": 9648
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.68912967597631e-07,
+      "loss": 1.0552,
+      "step": 9649
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.681957423617165e-07,
+      "loss": 1.1921,
+      "step": 9650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.674794618685617e-07,
+      "loss": 0.9385,
+      "step": 9651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.667641261877019e-07,
+      "loss": 1.1309,
+      "step": 9652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6604973538858046e-07,
+      "loss": 1.042,
+      "step": 9653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.653362895405498e-07,
+      "loss": 0.9775,
+      "step": 9654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6462378871286777e-07,
+      "loss": 1.0725,
+      "step": 9655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.639122329747057e-07,
+      "loss": 1.005,
+      "step": 9656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6320162239513725e-07,
+      "loss": 1.0698,
+      "step": 9657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6249195704314724e-07,
+      "loss": 0.9707,
+      "step": 9658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.617832369876294e-07,
+      "loss": 1.147,
+      "step": 9659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6107546229738433e-07,
+      "loss": 1.0093,
+      "step": 9660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.603686330411215e-07,
+      "loss": 0.996,
+      "step": 9661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5966274928745705e-07,
+      "loss": 1.0148,
+      "step": 9662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5895781110491735e-07,
+      "loss": 1.0676,
+      "step": 9663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5825381856193876e-07,
+      "loss": 1.1472,
+      "step": 9664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5755077172685995e-07,
+      "loss": 1.1772,
+      "step": 9665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.568486706679318e-07,
+      "loss": 1.1284,
+      "step": 9666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5614751545331553e-07,
+      "loss": 1.043,
+      "step": 9667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5544730615107425e-07,
+      "loss": 1.101,
+      "step": 9668
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5474804282918487e-07,
+      "loss": 0.9238,
+      "step": 9669
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5404972555552967e-07,
+      "loss": 1.0337,
+      "step": 9670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5335235439789885e-07,
+      "loss": 1.071,
+      "step": 9671
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.526559294239961e-07,
+      "loss": 0.8737,
+      "step": 9672
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5196045070142283e-07,
+      "loss": 0.9723,
+      "step": 9673
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.512659182976973e-07,
+      "loss": 1.0759,
+      "step": 9674
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.505723322802445e-07,
+      "loss": 1.1316,
+      "step": 9675
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.498796927163938e-07,
+      "loss": 0.9474,
+      "step": 9676
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4918799967338703e-07,
+      "loss": 1.0127,
+      "step": 9677
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.484972532183716e-07,
+      "loss": 1.0094,
+      "step": 9678
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4780745341840497e-07,
+      "loss": 1.0179,
+      "step": 9679
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4711860034044907e-07,
+      "loss": 1.0176,
+      "step": 9680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.464306940513772e-07,
+      "loss": 1.0577,
+      "step": 9681
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4574373461797027e-07,
+      "loss": 0.9089,
+      "step": 9682
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4505772210691837e-07,
+      "loss": 1.0708,
+      "step": 9683
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.443726565848137e-07,
+      "loss": 1.0332,
+      "step": 9684
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.436885381181664e-07,
+      "loss": 1.1787,
+      "step": 9685
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.430053667733856e-07,
+      "loss": 1.1016,
+      "step": 9686
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.423231426167916e-07,
+      "loss": 1.1038,
+      "step": 9687
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.416418657146158e-07,
+      "loss": 1.0214,
+      "step": 9688
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.409615361329942e-07,
+      "loss": 1.061,
+      "step": 9689
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4028215393797073e-07,
+      "loss": 1.0724,
+      "step": 9690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.396037191955003e-07,
+      "loss": 1.0835,
+      "step": 9691
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3892623197144136e-07,
+      "loss": 1.0546,
+      "step": 9692
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.382496923315647e-07,
+      "loss": 1.1555,
+      "step": 9693
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3757410034154883e-07,
+      "loss": 0.9436,
+      "step": 9694
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3689945606697463e-07,
+      "loss": 1.0964,
+      "step": 9695
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3622575957333748e-07,
+      "loss": 1.1362,
+      "step": 9696
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3555301092604066e-07,
+      "loss": 0.9897,
+      "step": 9697
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3488121019038857e-07,
+      "loss": 1.0452,
+      "step": 9698
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3421035743160125e-07,
+      "loss": 1.0649,
+      "step": 9699
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.335404527148022e-07,
+      "loss": 1.0975,
+      "step": 9700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3287149610502602e-07,
+      "loss": 1.1099,
+      "step": 9701
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.32203487667213e-07,
+      "loss": 1.0662,
+      "step": 9702
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.315364274662102e-07,
+      "loss": 0.9827,
+      "step": 9703
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3087031556677574e-07,
+      "loss": 0.9989,
+      "step": 9704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3020515203357574e-07,
+      "loss": 1.0227,
+      "step": 9705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2954093693117962e-07,
+      "loss": 1.0327,
+      "step": 9706
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2887767032407138e-07,
+      "loss": 1.0779,
+      "step": 9707
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2821535227663838e-07,
+      "loss": 1.0415,
+      "step": 9708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2755398285317477e-07,
+      "loss": 1.0957,
+      "step": 9709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2689356211788805e-07,
+      "loss": 1.0708,
+      "step": 9710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2623409013488806e-07,
+      "loss": 1.0066,
+      "step": 9711
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2557556696819582e-07,
+      "loss": 1.0618,
+      "step": 9712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.249179926817413e-07,
+      "loss": 1.0728,
+      "step": 9713
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2426136733935677e-07,
+      "loss": 1.1411,
+      "step": 9714
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.23605691004789e-07,
+      "loss": 1.0962,
+      "step": 9715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2295096374168822e-07,
+      "loss": 1.1255,
+      "step": 9716
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2229718561361468e-07,
+      "loss": 1.0303,
+      "step": 9717
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.216443566840354e-07,
+      "loss": 1.1602,
+      "step": 9718
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2099247701632408e-07,
+      "loss": 1.0464,
+      "step": 9719
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2034154667376574e-07,
+      "loss": 1.0988,
+      "step": 9720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1969156571955196e-07,
+      "loss": 1.1411,
+      "step": 9721
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.19042534216779e-07,
+      "loss": 1.0427,
+      "step": 9722
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1839445222845645e-07,
+      "loss": 1.0654,
+      "step": 9723
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1774731981749618e-07,
+      "loss": 1.0723,
+      "step": 9724
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.171011370467213e-07,
+      "loss": 1.0725,
+      "step": 9725
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1645590397886273e-07,
+      "loss": 1.1575,
+      "step": 9726
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1581162067655592e-07,
+      "loss": 1.1687,
+      "step": 9727
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1516828720234862e-07,
+      "loss": 1.1211,
+      "step": 9728
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.145259036186953e-07,
+      "loss": 1.1572,
+      "step": 9729
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1388446998795498e-07,
+      "loss": 1.0237,
+      "step": 9730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1324398637239563e-07,
+      "loss": 1.0356,
+      "step": 9731
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.126044528341975e-07,
+      "loss": 1.0215,
+      "step": 9732
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1196586943544207e-07,
+      "loss": 0.9849,
+      "step": 9733
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1132823623812416e-07,
+      "loss": 1.0463,
+      "step": 9734
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1069155330414204e-07,
+      "loss": 1.0608,
+      "step": 9735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1005582069530295e-07,
+      "loss": 0.9529,
+      "step": 9736
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0942103847332528e-07,
+      "loss": 1.0969,
+      "step": 9737
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0878720669983088e-07,
+      "loss": 1.1873,
+      "step": 9738
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0815432543634938e-07,
+      "loss": 0.9771,
+      "step": 9739
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0752239474432168e-07,
+      "loss": 1.0587,
+      "step": 9740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.06891414685092e-07,
+      "loss": 0.9763,
+      "step": 9741
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.06261385319918e-07,
+      "loss": 0.8838,
+      "step": 9742
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.056323067099575e-07,
+      "loss": 1.0378,
+      "step": 9743
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0500417891628155e-07,
+      "loss": 1.1545,
+      "step": 9744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.043770019998703e-07,
+      "loss": 1.0442,
+      "step": 9745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.037507760216051e-07,
+      "loss": 1.0938,
+      "step": 9746
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0312550104227947e-07,
+      "loss": 0.995,
+      "step": 9747
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0250117712259487e-07,
+      "loss": 1.0576,
+      "step": 9748
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.018778043231584e-07,
+      "loss": 1.0402,
+      "step": 9749
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0125538270448608e-07,
+      "loss": 1.0232,
+      "step": 9750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.006339123270007e-07,
+      "loss": 1.1141,
+      "step": 9751
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0001339325103508e-07,
+      "loss": 1.0469,
+      "step": 9752
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.993938255368244e-07,
+      "loss": 0.9991,
+      "step": 9753
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9877520924451943e-07,
+      "loss": 1.0093,
+      "step": 9754
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9815754443417102e-07,
+      "loss": 1.146,
+      "step": 9755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.975408311657412e-07,
+      "loss": 1.1716,
+      "step": 9756
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9692506949909985e-07,
+      "loss": 1.0227,
+      "step": 9757
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.963102594940236e-07,
+      "loss": 1.0669,
+      "step": 9758
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9569640121019695e-07,
+      "loss": 1.0789,
+      "step": 9759
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.950834947072122e-07,
+      "loss": 1.0708,
+      "step": 9760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9447154004456737e-07,
+      "loss": 1.0269,
+      "step": 9761
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9386053728167152e-07,
+      "loss": 1.0872,
+      "step": 9762
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.932504864778395e-07,
+      "loss": 1.0791,
+      "step": 9763
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.926413876922928e-07,
+      "loss": 0.9882,
+      "step": 9764
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9203324098416076e-07,
+      "loss": 1.1865,
+      "step": 9765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9142604641248285e-07,
+      "loss": 0.9924,
+      "step": 9766
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9081980403620416e-07,
+      "loss": 1.1992,
+      "step": 9767
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9021451391417534e-07,
+      "loss": 1.2634,
+      "step": 9768
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.896101761051583e-07,
+      "loss": 1.0137,
+      "step": 9769
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8900679066782169e-07,
+      "loss": 0.8772,
+      "step": 9770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.884043576607375e-07,
+      "loss": 0.978,
+      "step": 9771
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.878028771423923e-07,
+      "loss": 1.0889,
+      "step": 9772
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8720234917117496e-07,
+      "loss": 1.0439,
+      "step": 9773
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8660277380538327e-07,
+      "loss": 1.1165,
+      "step": 9774
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.86004151103224e-07,
+      "loss": 1.0266,
+      "step": 9775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8540648112280735e-07,
+      "loss": 1.1233,
+      "step": 9776
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8480976392215578e-07,
+      "loss": 1.0972,
+      "step": 9777
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.842139995591985e-07,
+      "loss": 1.0322,
+      "step": 9778
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8361918809176815e-07,
+      "loss": 1.1064,
+      "step": 9779
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8302532957761076e-07,
+      "loss": 0.8932,
+      "step": 9780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8243242407437466e-07,
+      "loss": 1.001,
+      "step": 9781
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.818404716396194e-07,
+      "loss": 1.092,
+      "step": 9782
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8124947233081002e-07,
+      "loss": 1.0671,
+      "step": 9783
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.806594262053174e-07,
+      "loss": 1.0979,
+      "step": 9784
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8007033332042456e-07,
+      "loss": 1.0237,
+      "step": 9785
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.794821937333191e-07,
+      "loss": 1.2031,
+      "step": 9786
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7889500750109422e-07,
+      "loss": 1.0499,
+      "step": 9787
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7830877468075658e-07,
+      "loss": 0.9781,
+      "step": 9788
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7772349532921286e-07,
+      "loss": 1.0465,
+      "step": 9789
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7713916950328092e-07,
+      "loss": 1.0171,
+      "step": 9790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7655579725968875e-07,
+      "loss": 1.1091,
+      "step": 9791
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7597337865506548e-07,
+      "loss": 0.9814,
+      "step": 9792
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7539191374595255e-07,
+      "loss": 1.1049,
+      "step": 9793
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7481140258879814e-07,
+      "loss": 1.0474,
+      "step": 9794
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7423184523995606e-07,
+      "loss": 1.0336,
+      "step": 9795
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.736532417556891e-07,
+      "loss": 0.9631,
+      "step": 9796
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7307559219216675e-07,
+      "loss": 0.9589,
+      "step": 9797
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7249889660546416e-07,
+      "loss": 1.0825,
+      "step": 9798
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7192315505156875e-07,
+      "loss": 0.959,
+      "step": 9799
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7134836758636919e-07,
+      "loss": 1.0291,
+      "step": 9800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7077453426566749e-07,
+      "loss": 1.2681,
+      "step": 9801
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7020165514516794e-07,
+      "loss": 1.01,
+      "step": 9802
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6962973028048614e-07,
+      "loss": 1.031,
+      "step": 9803
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6905875972714314e-07,
+      "loss": 1.0249,
+      "step": 9804
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6848874354056577e-07,
+      "loss": 1.0681,
+      "step": 9805
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6791968177609197e-07,
+      "loss": 1.0468,
+      "step": 9806
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.673515744889642e-07,
+      "loss": 1.1187,
+      "step": 9807
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6678442173433173e-07,
+      "loss": 0.9784,
+      "step": 9808
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6621822356725492e-07,
+      "loss": 1.0972,
+      "step": 9809
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.656529800426987e-07,
+      "loss": 1.1008,
+      "step": 9810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6508869121553473e-07,
+      "loss": 1.0745,
+      "step": 9811
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.645253571405425e-07,
+      "loss": 0.9961,
+      "step": 9812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6396297787241166e-07,
+      "loss": 1.083,
+      "step": 9813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6340155346573404e-07,
+      "loss": 1.0215,
+      "step": 9814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6284108397501386e-07,
+      "loss": 0.9385,
+      "step": 9815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6228156945465756e-07,
+      "loss": 1.0562,
+      "step": 9816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6172300995898395e-07,
+      "loss": 1.1284,
+      "step": 9817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6116540554221627e-07,
+      "loss": 1.0737,
+      "step": 9818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606087562584857e-07,
+      "loss": 0.9552,
+      "step": 9819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6005306216182903e-07,
+      "loss": 1.072,
+      "step": 9820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.594983233061942e-07,
+      "loss": 1.0054,
+      "step": 9821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5894453974543256e-07,
+      "loss": 1.0264,
+      "step": 9822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5839171153330445e-07,
+      "loss": 1.0564,
+      "step": 9823
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.578398387234781e-07,
+      "loss": 0.9681,
+      "step": 9824
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5728892136952612e-07,
+      "loss": 0.9785,
+      "step": 9825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5673895952493246e-07,
+      "loss": 1.0342,
+      "step": 9826
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.561899532430866e-07,
+      "loss": 1.0043,
+      "step": 9827
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5564190257728151e-07,
+      "loss": 1.0994,
+      "step": 9828
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5509480758072459e-07,
+      "loss": 0.9994,
+      "step": 9829
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.545486683065245e-07,
+      "loss": 1.1572,
+      "step": 9830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5400348480769988e-07,
+      "loss": 1.1055,
+      "step": 9831
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5345925713717625e-07,
+      "loss": 1.1001,
+      "step": 9832
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5291598534778685e-07,
+      "loss": 1.0139,
+      "step": 9833
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.523736694922684e-07,
+      "loss": 1.1453,
+      "step": 9834
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.51832309623271e-07,
+      "loss": 1.1206,
+      "step": 9835
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5129190579334707e-07,
+      "loss": 1.0762,
+      "step": 9836
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5075245805495798e-07,
+      "loss": 1.0803,
+      "step": 9837
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.502139664604718e-07,
+      "loss": 1.038,
+      "step": 9838
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4967643106216457e-07,
+      "loss": 1.0754,
+      "step": 9839
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4913985191222115e-07,
+      "loss": 0.9949,
+      "step": 9840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4860422906272764e-07,
+      "loss": 1.082,
+      "step": 9841
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4806956256568362e-07,
+      "loss": 1.0332,
+      "step": 9842
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4753585247299307e-07,
+      "loss": 1.0405,
+      "step": 9843
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.470030988364668e-07,
+      "loss": 0.9915,
+      "step": 9844
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4647130170782453e-07,
+      "loss": 0.9979,
+      "step": 9845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4594046113869053e-07,
+      "loss": 1.1322,
+      "step": 9846
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4541057718059804e-07,
+      "loss": 1.0906,
+      "step": 9847
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4488164988498922e-07,
+      "loss": 1.0674,
+      "step": 9848
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4435367930320743e-07,
+      "loss": 1.0669,
+      "step": 9849
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4382666548650948e-07,
+      "loss": 1.0959,
+      "step": 9850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4330060848605554e-07,
+      "loss": 1.0923,
+      "step": 9851
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4277550835291476e-07,
+      "loss": 1.1636,
+      "step": 9852
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4225136513806414e-07,
+      "loss": 1.0839,
+      "step": 9853
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.41728178892383e-07,
+      "loss": 1.0488,
+      "step": 9854
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4120594966666512e-07,
+      "loss": 1.0745,
+      "step": 9855
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4068467751160442e-07,
+      "loss": 0.8651,
+      "step": 9856
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4016436247780485e-07,
+      "loss": 1.0266,
+      "step": 9857
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3964500461577934e-07,
+      "loss": 1.0884,
+      "step": 9858
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3912660397594534e-07,
+      "loss": 1.0935,
+      "step": 9859
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3860916060862705e-07,
+      "loss": 1.0104,
+      "step": 9860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.380926745640576e-07,
+      "loss": 1.1726,
+      "step": 9861
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.375771458923769e-07,
+      "loss": 1.0581,
+      "step": 9862
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3706257464363048e-07,
+      "loss": 0.861,
+      "step": 9863
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3654896086777281e-07,
+      "loss": 1.063,
+      "step": 9864
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3603630461466288e-07,
+      "loss": 1.0684,
+      "step": 9865
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.355246059340698e-07,
+      "loss": 1.0607,
+      "step": 9866
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3501386487566714e-07,
+      "loss": 0.9443,
+      "step": 9867
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3450408148903638e-07,
+      "loss": 1.1301,
+      "step": 9868
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3399525582366792e-07,
+      "loss": 1.0483,
+      "step": 9869
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3348738792895666e-07,
+      "loss": 1.095,
+      "step": 9870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3298047785420321e-07,
+      "loss": 1.0109,
+      "step": 9871
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.324745256486204e-07,
+      "loss": 1.1472,
+      "step": 9872
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.319695313613234e-07,
+      "loss": 1.0042,
+      "step": 9873
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3146549504133522e-07,
+      "loss": 1.1409,
+      "step": 9874
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.309624167375889e-07,
+      "loss": 1.1362,
+      "step": 9875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3046029649891988e-07,
+      "loss": 1.0549,
+      "step": 9876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.299591343740758e-07,
+      "loss": 1.0632,
+      "step": 9877
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2945893041170553e-07,
+      "loss": 1.0125,
+      "step": 9878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2895968466036912e-07,
+      "loss": 0.9399,
+      "step": 9879
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2846139716853111e-07,
+      "loss": 1.062,
+      "step": 9880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.279640679845662e-07,
+      "loss": 0.9033,
+      "step": 9881
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2746769715675234e-07,
+      "loss": 1.0579,
+      "step": 9882
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.269722847332766e-07,
+      "loss": 0.9421,
+      "step": 9883
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2647783076223274e-07,
+      "loss": 1.0884,
+      "step": 9884
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2598433529162235e-07,
+      "loss": 0.9681,
+      "step": 9885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2549179836935155e-07,
+      "loss": 0.9893,
+      "step": 9886
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.250002200432343e-07,
+      "loss": 1.0127,
+      "step": 9887
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2450960036099247e-07,
+      "loss": 1.0909,
+      "step": 9888
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.240199393702557e-07,
+      "loss": 1.0286,
+      "step": 9889
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2353123711855707e-07,
+      "loss": 1.074,
+      "step": 9890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.230434936533409e-07,
+      "loss": 1.1459,
+      "step": 9891
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2255670902195592e-07,
+      "loss": 1.0405,
+      "step": 9892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2207088327165552e-07,
+      "loss": 1.0376,
+      "step": 9893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2158601644960521e-07,
+      "loss": 1.0618,
+      "step": 9894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2110210860287408e-07,
+      "loss": 0.9603,
+      "step": 9895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2061915977843897e-07,
+      "loss": 1.0222,
+      "step": 9896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2013717002318458e-07,
+      "loss": 1.0403,
+      "step": 9897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1965613938389907e-07,
+      "loss": 1.1086,
+      "step": 9898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1917606790728286e-07,
+      "loss": 1.0483,
+      "step": 9899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1869695563993755e-07,
+      "loss": 1.0192,
+      "step": 9900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1821880262837482e-07,
+      "loss": 1.0939,
+      "step": 9901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1774160891901421e-07,
+      "loss": 0.9836,
+      "step": 9902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1726537455817866e-07,
+      "loss": 1.0107,
+      "step": 9903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.167900995921012e-07,
+      "loss": 0.9658,
+      "step": 9904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1631578406692156e-07,
+      "loss": 0.9091,
+      "step": 9905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1584242802868295e-07,
+      "loss": 0.8217,
+      "step": 9906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.153700315233397e-07,
+      "loss": 1.067,
+      "step": 9907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1489859459675068e-07,
+      "loss": 1.0967,
+      "step": 9908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1442811729468039e-07,
+      "loss": 1.0515,
+      "step": 9909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1395859966280453e-07,
+      "loss": 0.9093,
+      "step": 9910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1349004174669886e-07,
+      "loss": 1.0061,
+      "step": 9911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1302244359185365e-07,
+      "loss": 1.0,
+      "step": 9912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.125558052436615e-07,
+      "loss": 1.0269,
+      "step": 9913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1209012674742281e-07,
+      "loss": 1.0188,
+      "step": 9914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1162540814834255e-07,
+      "loss": 1.0994,
+      "step": 9915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1116164949153686e-07,
+      "loss": 1.1111,
+      "step": 9916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1069885082202526e-07,
+      "loss": 0.9335,
+      "step": 9917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1023701218473626e-07,
+      "loss": 1.084,
+      "step": 9918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.097761336245029e-07,
+      "loss": 1.1475,
+      "step": 9919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0931621518606717e-07,
+      "loss": 1.026,
+      "step": 9920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0885725691407778e-07,
+      "loss": 1.0704,
+      "step": 9921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.08399258853088e-07,
+      "loss": 1.0989,
+      "step": 9922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0794222104755891e-07,
+      "loss": 1.0778,
+      "step": 9923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.074861435418606e-07,
+      "loss": 1.0559,
+      "step": 9924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0703102638026652e-07,
+      "loss": 0.9694,
+      "step": 9925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0657686960696023e-07,
+      "loss": 1.0566,
+      "step": 9926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0612367326602757e-07,
+      "loss": 1.1493,
+      "step": 9927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0567143740146558e-07,
+      "loss": 1.0996,
+      "step": 9928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.05220162057178e-07,
+      "loss": 1.0471,
+      "step": 9929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0476984727696981e-07,
+      "loss": 0.9777,
+      "step": 9930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0432049310455939e-07,
+      "loss": 1.1377,
+      "step": 9931
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0387209958356848e-07,
+      "loss": 1.0212,
+      "step": 9932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0342466675752561e-07,
+      "loss": 1.1572,
+      "step": 9933
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0297819466986714e-07,
+      "loss": 1.1028,
+      "step": 9934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0253268336393507e-07,
+      "loss": 0.9384,
+      "step": 9935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0208813288298036e-07,
+      "loss": 0.8523,
+      "step": 9936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0164454327015738e-07,
+      "loss": 1.1145,
+      "step": 9937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0120191456852945e-07,
+      "loss": 1.0669,
+      "step": 9938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0076024682106444e-07,
+      "loss": 1.0728,
+      "step": 9939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0031954007064137e-07,
+      "loss": 1.0693,
+      "step": 9940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.987979436004158e-08,
+      "loss": 1.0859,
+      "step": 9941
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.944100973195425e-08,
+      "loss": 1.0769,
+      "step": 9942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.900318622897642e-08,
+      "loss": 0.9939,
+      "step": 9943
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.856632389361076e-08,
+      "loss": 1.0222,
+      "step": 9944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.813042276826779e-08,
+      "loss": 1.0217,
+      "step": 9945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.769548289526143e-08,
+      "loss": 1.0012,
+      "step": 9946
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.726150431681681e-08,
+      "loss": 1.0334,
+      "step": 9947
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.682848707506354e-08,
+      "loss": 0.9639,
+      "step": 9948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.63964312120369e-08,
+      "loss": 1.0466,
+      "step": 9949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.596533676968112e-08,
+      "loss": 1.1145,
+      "step": 9950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.553520378984604e-08,
+      "loss": 0.9601,
+      "step": 9951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.510603231428606e-08,
+      "loss": 1.1061,
+      "step": 9952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.467782238466672e-08,
+      "loss": 1.1165,
+      "step": 9953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.425057404255589e-08,
+      "loss": 0.9783,
+      "step": 9954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.38242873294315e-08,
+      "loss": 1.1624,
+      "step": 9955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.339896228667488e-08,
+      "loss": 1.1208,
+      "step": 9956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.297459895557747e-08,
+      "loss": 1.0061,
+      "step": 9957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.255119737733409e-08,
+      "loss": 1.0442,
+      "step": 9958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.212875759304851e-08,
+      "loss": 1.0847,
+      "step": 9959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.170727964372905e-08,
+      "loss": 1.1464,
+      "step": 9960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.128676357029298e-08,
+      "loss": 1.1052,
+      "step": 9961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.08672094135643e-08,
+      "loss": 1.024,
+      "step": 9962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.044861721426934e-08,
+      "loss": 0.9305,
+      "step": 9963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.003098701304669e-08,
+      "loss": 1.0974,
+      "step": 9964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.961431885043836e-08,
+      "loss": 1.0996,
+      "step": 9965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.919861276689312e-08,
+      "loss": 0.9847,
+      "step": 9966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.878386880276757e-08,
+      "loss": 1.0527,
+      "step": 9967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.837008699832284e-08,
+      "loss": 1.0486,
+      "step": 9968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.79572673937279e-08,
+      "loss": 0.9863,
+      "step": 9969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.75454100290618e-08,
+      "loss": 0.9924,
+      "step": 9970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.713451494430259e-08,
+      "loss": 1.0894,
+      "step": 9971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.672458217934166e-08,
+      "loss": 1.1018,
+      "step": 9972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.631561177397274e-08,
+      "loss": 1.1399,
+      "step": 9973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.590760376789964e-08,
+      "loss": 0.9783,
+      "step": 9974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.550055820072955e-08,
+      "loss": 0.9581,
+      "step": 9975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.509447511197644e-08,
+      "loss": 1.0603,
+      "step": 9976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.46893545410643e-08,
+      "loss": 1.1912,
+      "step": 9977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.42851965273217e-08,
+      "loss": 1.1562,
+      "step": 9978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.388200110998168e-08,
+      "loss": 1.1274,
+      "step": 9979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.347976832818627e-08,
+      "loss": 1.1257,
+      "step": 9980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.307849822098313e-08,
+      "loss": 1.0745,
+      "step": 9981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.267819082732664e-08,
+      "loss": 1.0168,
+      "step": 9982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.227884618607796e-08,
+      "loss": 1.0508,
+      "step": 9983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.188046433600383e-08,
+      "loss": 1.0354,
+      "step": 9984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.148304531577999e-08,
+      "loss": 1.0457,
+      "step": 9985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.108658916398671e-08,
+      "loss": 1.0854,
+      "step": 9986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.069109591910984e-08,
+      "loss": 1.1304,
+      "step": 9987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.029656561954424e-08,
+      "loss": 0.9611,
+      "step": 9988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.990299830358928e-08,
+      "loss": 1.0088,
+      "step": 9989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.951039400945215e-08,
+      "loss": 1.0369,
+      "step": 9990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.911875277524572e-08,
+      "loss": 1.074,
+      "step": 9991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.872807463898958e-08,
+      "loss": 1.1155,
+      "step": 9992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.833835963861002e-08,
+      "loss": 1.0093,
+      "step": 9993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.794960781194017e-08,
+      "loss": 1.0806,
+      "step": 9994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.756181919671869e-08,
+      "loss": 1.0894,
+      "step": 9995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.717499383059213e-08,
+      "loss": 1.0032,
+      "step": 9996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.678913175111157e-08,
+      "loss": 1.074,
+      "step": 9997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.640423299573596e-08,
+      "loss": 1.0054,
+      "step": 9998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.602029760183095e-08,
+      "loss": 1.0803,
+      "step": 9999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.56373256066667e-08,
+      "loss": 1.0225,
+      "step": 10000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.52553170474235e-08,
+      "loss": 1.0398,
+      "step": 10001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.487427196118501e-08,
+      "loss": 0.9816,
+      "step": 10002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.449419038494166e-08,
+      "loss": 1.2356,
+      "step": 10003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.411507235559168e-08,
+      "loss": 1.0117,
+      "step": 10004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.373691790993897e-08,
+      "loss": 0.9761,
+      "step": 10005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.335972708469308e-08,
+      "loss": 1.2297,
+      "step": 10006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.298349991647135e-08,
+      "loss": 1.0881,
+      "step": 10007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.260823644179793e-08,
+      "loss": 1.0331,
+      "step": 10008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.223393669710255e-08,
+      "loss": 1.0203,
+      "step": 10009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.186060071872059e-08,
+      "loss": 1.0477,
+      "step": 10010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.148822854289417e-08,
+      "loss": 1.0266,
+      "step": 10011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.111682020577437e-08,
+      "loss": 1.1316,
+      "step": 10012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.074637574341459e-08,
+      "loss": 1.1318,
+      "step": 10013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.037689519177826e-08,
+      "loss": 1.1392,
+      "step": 10014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.000837858673338e-08,
+      "loss": 1.0598,
+      "step": 10015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.964082596405463e-08,
+      "loss": 1.02,
+      "step": 10016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.927423735942352e-08,
+      "loss": 1.1465,
+      "step": 10017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.890861280842709e-08,
+      "loss": 1.1416,
+      "step": 10018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.85439523465592e-08,
+      "loss": 0.9949,
+      "step": 10019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.818025600922151e-08,
+      "loss": 1.1211,
+      "step": 10020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.781752383172024e-08,
+      "loss": 1.1233,
+      "step": 10021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.745575584926722e-08,
+      "loss": 0.9817,
+      "step": 10022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.709495209698546e-08,
+      "loss": 0.9961,
+      "step": 10023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.673511260989807e-08,
+      "loss": 1.0496,
+      "step": 10024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.63762374229382e-08,
+      "loss": 1.2656,
+      "step": 10025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.601832657094575e-08,
+      "loss": 1.0631,
+      "step": 10026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.566138008866518e-08,
+      "loss": 1.1182,
+      "step": 10027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.530539801074765e-08,
+      "loss": 1.1035,
+      "step": 10028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.495038037175106e-08,
+      "loss": 1.0759,
+      "step": 10029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.459632720614118e-08,
+      "loss": 0.9857,
+      "step": 10030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.424323854828718e-08,
+      "loss": 1.0411,
+      "step": 10031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.38911144324672e-08,
+      "loss": 1.1316,
+      "step": 10032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.35399548928639e-08,
+      "loss": 1.1743,
+      "step": 10033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.31897599635678e-08,
+      "loss": 0.8921,
+      "step": 10034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.284052967857501e-08,
+      "loss": 1.0046,
+      "step": 10035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.249226407178732e-08,
+      "loss": 0.9944,
+      "step": 10036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.214496317701435e-08,
+      "loss": 0.9822,
+      "step": 10037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.179862702797135e-08,
+      "loss": 0.9771,
+      "step": 10038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.14532556582792e-08,
+      "loss": 0.9364,
+      "step": 10039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.110884910146664e-08,
+      "loss": 1.1619,
+      "step": 10040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.076540739096804e-08,
+      "loss": 0.9934,
+      "step": 10041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.042293056012228e-08,
+      "loss": 1.1272,
+      "step": 10042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.008141864217831e-08,
+      "loss": 0.9048,
+      "step": 10043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.974087167028741e-08,
+      "loss": 1.02,
+      "step": 10044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.940128967751091e-08,
+      "loss": 1.0217,
+      "step": 10045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.906267269681465e-08,
+      "loss": 0.9803,
+      "step": 10046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.872502076107012e-08,
+      "loss": 1.0747,
+      "step": 10047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.838833390305554e-08,
+      "loss": 0.8773,
+      "step": 10048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8052612155455877e-08,
+      "loss": 0.9983,
+      "step": 10049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7717855550862846e-08,
+      "loss": 0.9988,
+      "step": 10050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.738406412177488e-08,
+      "loss": 0.9574,
+      "step": 10051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.705123790059275e-08,
+      "loss": 0.9972,
+      "step": 10052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6719376919629475e-08,
+      "loss": 0.9558,
+      "step": 10053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.638848121110041e-08,
+      "loss": 0.928,
+      "step": 10054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.605855080712763e-08,
+      "loss": 0.9698,
+      "step": 10055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5729585739739966e-08,
+      "loss": 1.073,
+      "step": 10056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.540158604087298e-08,
+      "loss": 1.0771,
+      "step": 10057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.507455174236898e-08,
+      "loss": 1.0715,
+      "step": 10058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.474848287597478e-08,
+      "loss": 1.1084,
+      "step": 10059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.442337947334508e-08,
+      "loss": 1.051,
+      "step": 10060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4099241566040184e-08,
+      "loss": 1.0264,
+      "step": 10061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.377606918552603e-08,
+      "loss": 1.0642,
+      "step": 10062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.34538623631764e-08,
+      "loss": 1.0448,
+      "step": 10063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.313262113026962e-08,
+      "loss": 1.0367,
+      "step": 10064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.281234551799075e-08,
+      "loss": 1.0388,
+      "step": 10065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.249303555743268e-08,
+      "loss": 1.0735,
+      "step": 10066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2174691279593955e-08,
+      "loss": 0.9784,
+      "step": 10067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.185731271537653e-08,
+      "loss": 0.946,
+      "step": 10068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1540899895591304e-08,
+      "loss": 0.9109,
+      "step": 10069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1225452850957035e-08,
+      "loss": 1.092,
+      "step": 10070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.091097161209369e-08,
+      "loss": 1.1836,
+      "step": 10071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.059745620953349e-08,
+      "loss": 1.1338,
+      "step": 10072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.028490667370989e-08,
+      "loss": 1.1123,
+      "step": 10073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.997332303496416e-08,
+      "loss": 0.9961,
+      "step": 10074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9662705323545444e-08,
+      "loss": 1.0903,
+      "step": 10075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.93530535696074e-08,
+      "loss": 0.9753,
+      "step": 10076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.904436780321043e-08,
+      "loss": 0.9767,
+      "step": 10077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.873664805432054e-08,
+      "loss": 1.1104,
+      "step": 10078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8429894352810534e-08,
+      "loss": 1.1041,
+      "step": 10079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8124106728461016e-08,
+      "loss": 1.0347,
+      "step": 10080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.781928521095491e-08,
+      "loss": 1.0151,
+      "step": 10081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7515429829885215e-08,
+      "loss": 1.0334,
+      "step": 10082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.721254061474945e-08,
+      "loss": 1.0259,
+      "step": 10083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.691061759495075e-08,
+      "loss": 1.0984,
+      "step": 10084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6609660799799005e-08,
+      "loss": 1.0779,
+      "step": 10085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6309670258511964e-08,
+      "loss": 0.9984,
+      "step": 10086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.601064600020966e-08,
+      "loss": 1.0625,
+      "step": 10087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.571258805392331e-08,
+      "loss": 1.0784,
+      "step": 10088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.541549644858645e-08,
+      "loss": 1.1836,
+      "step": 10089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.511937121303933e-08,
+      "loss": 0.8899,
+      "step": 10090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.482421237603007e-08,
+      "loss": 1.0842,
+      "step": 10091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.453001996621242e-08,
+      "loss": 0.9467,
+      "step": 10092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.423679401214576e-08,
+      "loss": 0.8955,
+      "step": 10093
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.394453454229508e-08,
+      "loss": 1.1367,
+      "step": 10094
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.365324158503326e-08,
+      "loss": 1.0625,
+      "step": 10095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.336291516863767e-08,
+      "loss": 1.0062,
+      "step": 10096
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.307355532129243e-08,
+      "loss": 1.0171,
+      "step": 10097
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.278516207108952e-08,
+      "loss": 1.1133,
+      "step": 10098
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2497735446023204e-08,
+      "loss": 0.9896,
+      "step": 10099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.221127547399895e-08,
+      "loss": 1.0542,
+      "step": 10100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.192578218282339e-08,
+      "loss": 1.0442,
+      "step": 10101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1641255600212146e-08,
+      "loss": 1.0057,
+      "step": 10102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1357695753786456e-08,
+      "loss": 1.0393,
+      "step": 10103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.107510267107429e-08,
+      "loss": 1.0916,
+      "step": 10104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0793476379510365e-08,
+      "loss": 0.98,
+      "step": 10105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0512816906430604e-08,
+      "loss": 0.9265,
+      "step": 10106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0233124279084327e-08,
+      "loss": 1.0862,
+      "step": 10107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9954398524622015e-08,
+      "loss": 1.126,
+      "step": 10108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9676639670102044e-08,
+      "loss": 1.1929,
+      "step": 10109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.93998477424884e-08,
+      "loss": 1.0354,
+      "step": 10110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9124022768651794e-08,
+      "loss": 1.0891,
+      "step": 10111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.88491647753686e-08,
+      "loss": 0.9922,
+      "step": 10112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.857527378932191e-08,
+      "loss": 0.9474,
+      "step": 10113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.830234983709935e-08,
+      "loss": 1.0076,
+      "step": 10114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.803039294519639e-08,
+      "loss": 0.8843,
+      "step": 10115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7759403140014136e-08,
+      "loss": 1.0251,
+      "step": 10116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.748938044785933e-08,
+      "loss": 1.027,
+      "step": 10117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.722032489494653e-08,
+      "loss": 1.0449,
+      "step": 10118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6952236507392655e-08,
+      "loss": 0.9084,
+      "step": 10119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.668511531122576e-08,
+      "loss": 1.0332,
+      "step": 10120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.641896133237621e-08,
+      "loss": 0.9634,
+      "step": 10121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6153774596681125e-08,
+      "loss": 1.0645,
+      "step": 10122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.588955512988434e-08,
+      "loss": 0.9592,
+      "step": 10123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5626302957637584e-08,
+      "loss": 0.9982,
+      "step": 10124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5364018105494834e-08,
+      "loss": 1.0098,
+      "step": 10125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.510270059891796e-08,
+      "loss": 1.0701,
+      "step": 10126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4842350463276664e-08,
+      "loss": 0.9561,
+      "step": 10127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.458296772384406e-08,
+      "loss": 1.1854,
+      "step": 10128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4324552405801124e-08,
+      "loss": 0.9031,
+      "step": 10129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4067104534233345e-08,
+      "loss": 1.0619,
+      "step": 10130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.381062413413405e-08,
+      "loss": 0.9998,
+      "step": 10131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.355511123040112e-08,
+      "loss": 1.0496,
+      "step": 10132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.330056584784025e-08,
+      "loss": 0.9626,
+      "step": 10133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3046988011161685e-08,
+      "loss": 1.0955,
+      "step": 10134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2794377744981286e-08,
+      "loss": 1.0242,
+      "step": 10135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.254273507382277e-08,
+      "loss": 1.095,
+      "step": 10136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.229206002211549e-08,
+      "loss": 0.8765,
+      "step": 10137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.204235261419442e-08,
+      "loss": 1.0812,
+      "step": 10138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.179361287429905e-08,
+      "loss": 1.1414,
+      "step": 10139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1545840826577855e-08,
+      "loss": 1.0199,
+      "step": 10140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.129903649508492e-08,
+      "loss": 1.1489,
+      "step": 10141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1053199903776645e-08,
+      "loss": 1.0334,
+      "step": 10142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.08083310765217e-08,
+      "loss": 0.9927,
+      "step": 10143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0564430037088866e-08,
+      "loss": 1.0071,
+      "step": 10144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.032149680915586e-08,
+      "loss": 1.063,
+      "step": 10145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.007953141630715e-08,
+      "loss": 1.0841,
+      "step": 10146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9838533882031727e-08,
+      "loss": 1.0203,
+      "step": 10147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.959850422972532e-08,
+      "loss": 1.1035,
+      "step": 10148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9359442482689292e-08,
+      "loss": 1.1187,
+      "step": 10149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9121348664131745e-08,
+      "loss": 1.1086,
+      "step": 10150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8884222797165296e-08,
+      "loss": 1.2122,
+      "step": 10151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8648064904809313e-08,
+      "loss": 1.0645,
+      "step": 10152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8412875009992124e-08,
+      "loss": 0.9956,
+      "step": 10153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8178653135542132e-08,
+      "loss": 1.0708,
+      "step": 10154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7945399304198927e-08,
+      "loss": 1.0168,
+      "step": 10155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7713113538606616e-08,
+      "loss": 1.0593,
+      "step": 10156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7481795861313832e-08,
+      "loss": 1.1663,
+      "step": 10157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.725144629477705e-08,
+      "loss": 0.9919,
+      "step": 10158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7022064861358388e-08,
+      "loss": 1.1022,
+      "step": 10159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6793651583325587e-08,
+      "loss": 1.012,
+      "step": 10160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.656620648285091e-08,
+      "loss": 0.981,
+      "step": 10161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6339729582016693e-08,
+      "loss": 1.0554,
+      "step": 10162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6114220902807574e-08,
+      "loss": 0.9417,
+      "step": 10163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.588968046711604e-08,
+      "loss": 0.9125,
+      "step": 10164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.56661082967391e-08,
+      "loss": 0.938,
+      "step": 10165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5443504413380504e-08,
+      "loss": 1.0811,
+      "step": 10166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5221868838651854e-08,
+      "loss": 0.9362,
+      "step": 10167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5001201594068157e-08,
+      "loss": 1.082,
+      "step": 10168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4781502701050065e-08,
+      "loss": 0.9648,
+      "step": 10169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4562772180927174e-08,
+      "loss": 1.0137,
+      "step": 10170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4345010054933614e-08,
+      "loss": 1.1223,
+      "step": 10171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.412821634420692e-08,
+      "loss": 0.9247,
+      "step": 10172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3912391069794706e-08,
+      "loss": 1.072,
+      "step": 10173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3697534252649092e-08,
+      "loss": 1.1504,
+      "step": 10174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.348364591362784e-08,
+      "loss": 1.0828,
+      "step": 10175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3270726073493233e-08,
+      "loss": 1.1311,
+      "step": 10176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3058774752917623e-08,
+      "loss": 1.0225,
+      "step": 10177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2847791972474553e-08,
+      "loss": 1.1802,
+      "step": 10178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.263777775264653e-08,
+      "loss": 1.1011,
+      "step": 10179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2428732113821683e-08,
+      "loss": 1.1509,
+      "step": 10180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2220655076292675e-08,
+      "loss": 1.0439,
+      "step": 10181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.201354666026001e-08,
+      "loss": 1.0605,
+      "step": 10182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.180740688582983e-08,
+      "loss": 0.978,
+      "step": 10183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1602235773012792e-08,
+      "loss": 0.9568,
+      "step": 10184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.13980333417263e-08,
+      "loss": 1.0288,
+      "step": 10185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1194799611795603e-08,
+      "loss": 1.1349,
+      "step": 10186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0992534602948256e-08,
+      "loss": 1.1018,
+      "step": 10187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0791238334820772e-08,
+      "loss": 0.9757,
+      "step": 10188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.059091082695419e-08,
+      "loss": 0.8984,
+      "step": 10189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0391552098795174e-08,
+      "loss": 1.04,
+      "step": 10190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0193162169698243e-08,
+      "loss": 1.0955,
+      "step": 10191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995741058922437e-08,
+      "loss": 1.0376,
+      "step": 10192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.979928878563242e-08,
+      "loss": 1.0665,
+      "step": 10193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.960380536889961e-08,
+      "loss": 1.1147,
+      "step": 10194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9409290827701043e-08,
+      "loss": 1.0029,
+      "step": 10195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.921574518092051e-08,
+      "loss": 0.9753,
+      "step": 10196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9023168447346308e-08,
+      "loss": 1.0219,
+      "step": 10197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8831560645673484e-08,
+      "loss": 1.0334,
+      "step": 10198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.864092179450272e-08,
+      "loss": 0.9452,
+      "step": 10199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.845125191234143e-08,
+      "loss": 1.1509,
+      "step": 10200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.826255101760155e-08,
+      "loss": 0.9839,
+      "step": 10201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.807481912860287e-08,
+      "loss": 1.1062,
+      "step": 10202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7888056263568598e-08,
+      "loss": 0.8947,
+      "step": 10203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.770226244062978e-08,
+      "loss": 1.0249,
+      "step": 10204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7517437677823102e-08,
+      "loss": 0.9658,
+      "step": 10205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7333581993090875e-08,
+      "loss": 0.9624,
+      "step": 10206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.715069540428216e-08,
+      "loss": 1.0737,
+      "step": 10207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6968777929150525e-08,
+      "loss": 1.0349,
+      "step": 10208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6787829585355187e-08,
+      "loss": 0.9086,
+      "step": 10209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.660785039046431e-08,
+      "loss": 1.083,
+      "step": 10210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.642884036194836e-08,
+      "loss": 0.9634,
+      "step": 10211
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6250799517185668e-08,
+      "loss": 1.0433,
+      "step": 10212
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6073727873459066e-08,
+      "loss": 1.033,
+      "step": 10213
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5897625447960363e-08,
+      "loss": 1.0352,
+      "step": 10214
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5722492257783663e-08,
+      "loss": 1.0171,
+      "step": 10215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5548328319930915e-08,
+      "loss": 1.098,
+      "step": 10216
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5375133651309716e-08,
+      "loss": 1.135,
+      "step": 10217
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5202908268733274e-08,
+      "loss": 1.0509,
+      "step": 10218
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5031652188920442e-08,
+      "loss": 1.1033,
+      "step": 10219
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4861365428497921e-08,
+      "loss": 1.0161,
+      "step": 10220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4692048003994707e-08,
+      "loss": 1.0969,
+      "step": 10221
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4523699931849877e-08,
+      "loss": 1.0466,
+      "step": 10222
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4356321228404801e-08,
+      "loss": 0.9868,
+      "step": 10223
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4189911909907594e-08,
+      "loss": 1.0117,
+      "step": 10224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4024471992515331e-08,
+      "loss": 0.9885,
+      "step": 10225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.386000149228628e-08,
+      "loss": 1.0011,
+      "step": 10226
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3696500425188774e-08,
+      "loss": 1.0911,
+      "step": 10227
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3533968807093457e-08,
+      "loss": 1.229,
+      "step": 10228
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3372406653779923e-08,
+      "loss": 0.9893,
+      "step": 10229
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3211813980931187e-08,
+      "loss": 0.9293,
+      "step": 10230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.305219080413811e-08,
+      "loss": 0.9917,
+      "step": 10231
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2893537138896073e-08,
+      "loss": 1.072,
+      "step": 10232
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.273585300060609e-08,
+      "loss": 1.0881,
+      "step": 10233
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.257913840457703e-08,
+      "loss": 1.0667,
+      "step": 10234
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2423393366022274e-08,
+      "loss": 1.1277,
+      "step": 10235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2268617900060842e-08,
+      "loss": 1.0493,
+      "step": 10236
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2114812021718491e-08,
+      "loss": 1.1101,
+      "step": 10237
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1961975745926613e-08,
+      "loss": 0.9707,
+      "step": 10238
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1810109087520005e-08,
+      "loss": 1.05,
+      "step": 10239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1659212061244651e-08,
+      "loss": 0.9484,
+      "step": 10240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1509284681746612e-08,
+      "loss": 1.09,
+      "step": 10241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1360326963582024e-08,
+      "loss": 1.0908,
+      "step": 10242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1212338921211542e-08,
+      "loss": 1.0222,
+      "step": 10243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1065320569001448e-08,
+      "loss": 1.1216,
+      "step": 10244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0919271921222552e-08,
+      "loss": 1.1948,
+      "step": 10245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0774192992054622e-08,
+      "loss": 1.1824,
+      "step": 10246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0630083795580837e-08,
+      "loss": 1.011,
+      "step": 10247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0486944345791118e-08,
+      "loss": 0.9891,
+      "step": 10248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0344774656581014e-08,
+      "loss": 1.1169,
+      "step": 10249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.020357474175282e-08,
+      "loss": 1.1189,
+      "step": 10250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0063344615013349e-08,
+      "loss": 1.0166,
+      "step": 10251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.924084289975045e-09,
+      "loss": 0.9731,
+      "step": 10252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.785793780158203e-09,
+      "loss": 0.995,
+      "step": 10253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.648473098987643e-09,
+      "loss": 1.0181,
+      "step": 10254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.51212225979381e-09,
+      "loss": 1.0469,
+      "step": 10255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.376741275812784e-09,
+      "loss": 1.0449,
+      "step": 10256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.242330160187386e-09,
+      "loss": 1.2314,
+      "step": 10257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.108888925967175e-09,
+      "loss": 0.9559,
+      "step": 10258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.976417586105124e-09,
+      "loss": 1.0366,
+      "step": 10259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.844916153462058e-09,
+      "loss": 0.993,
+      "step": 10260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.714384640802209e-09,
+      "loss": 1.0713,
+      "step": 10261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.584823060799885e-09,
+      "loss": 1.0596,
+      "step": 10262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.45623142602947e-09,
+      "loss": 0.9822,
+      "step": 10263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.328609748977645e-09,
+      "loss": 1.1335,
+      "step": 10264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.201958042031166e-09,
+      "loss": 0.9529,
+      "step": 10265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.076276317486864e-09,
+      "loss": 0.9708,
+      "step": 10266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.95156458754498e-09,
+      "loss": 1.0614,
+      "step": 10267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.827822864311386e-09,
+      "loss": 1.2424,
+      "step": 10268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.705051159798693e-09,
+      "loss": 1.1162,
+      "step": 10269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.583249485927369e-09,
+      "loss": 1.084,
+      "step": 10270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.462417854520176e-09,
+      "loss": 0.9574,
+      "step": 10271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.3425562773066225e-09,
+      "loss": 1.0752,
+      "step": 10272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.223664765924065e-09,
+      "loss": 1.1206,
+      "step": 10273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.1057433319132725e-09,
+      "loss": 1.1152,
+      "step": 10274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.988791986721755e-09,
+      "loss": 1.0092,
+      "step": 10275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.872810741702651e-09,
+      "loss": 1.0126,
+      "step": 10276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.757799608116955e-09,
+      "loss": 1.063,
+      "step": 10277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.643758597127958e-09,
+      "loss": 0.9778,
+      "step": 10278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.5306877198068054e-09,
+      "loss": 1.1208,
+      "step": 10279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.418586987130271e-09,
+      "loss": 1.0466,
+      "step": 10280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.307456409980761e-09,
+      "loss": 1.0664,
+      "step": 10281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.197295999147423e-09,
+      "loss": 0.8767,
+      "step": 10282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.0881057653239264e-09,
+      "loss": 1.1069,
+      "step": 10283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.979885719109568e-09,
+      "loss": 0.9841,
+      "step": 10284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8726358710103905e-09,
+      "loss": 1.0371,
+      "step": 10285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.766356231439174e-09,
+      "loss": 1.0623,
+      "step": 10286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6610468107110015e-09,
+      "loss": 0.9585,
+      "step": 10287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5567076190510275e-09,
+      "loss": 0.991,
+      "step": 10288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.453338666587815e-09,
+      "loss": 0.9733,
+      "step": 10289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3509399633566716e-09,
+      "loss": 1.0444,
+      "step": 10290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2495115192963135e-09,
+      "loss": 0.9819,
+      "step": 10291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1490533442555276e-09,
+      "loss": 1.0842,
+      "step": 10292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.049565447985405e-09,
+      "loss": 1.0461,
+      "step": 10293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.951047840143774e-09,
+      "loss": 1.0479,
+      "step": 10294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8535005302952074e-09,
+      "loss": 1.0018,
+      "step": 10295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.756923527908796e-09,
+      "loss": 1.1604,
+      "step": 10296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6613168423614854e-09,
+      "loss": 1.0303,
+      "step": 10297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.566680482932517e-09,
+      "loss": 0.9366,
+      "step": 10298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.473014458808988e-09,
+      "loss": 1.0232,
+      "step": 10299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.380318779085846e-09,
+      "loss": 1.0203,
+      "step": 10300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.288593452760337e-09,
+      "loss": 1.0068,
+      "step": 10301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.19783848873756e-09,
+      "loss": 1.1609,
+      "step": 10302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.108053895827135e-09,
+      "loss": 1.1074,
+      "step": 10303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.019239682746534e-09,
+      "loss": 0.9275,
+      "step": 10304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.931395858115528e-09,
+      "loss": 1.0354,
+      "step": 10305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.844522430462849e-09,
+      "loss": 1.1008,
+      "step": 10306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.758619408222864e-09,
+      "loss": 1.0841,
+      "step": 10307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.673686799734455e-09,
+      "loss": 0.9431,
+      "step": 10308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.58972461324103e-09,
+      "loss": 0.9127,
+      "step": 10309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5067328568949566e-09,
+      "loss": 1.0615,
+      "step": 10310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4247115387531225e-09,
+      "loss": 1.1406,
+      "step": 10311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.343660666778048e-09,
+      "loss": 1.0507,
+      "step": 10312
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.263580248836773e-09,
+      "loss": 0.8798,
+      "step": 10313
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.184470292704189e-09,
+      "loss": 0.9441,
+      "step": 10314
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.10633080605971e-09,
+      "loss": 1.0754,
+      "step": 10315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.029161796489488e-09,
+      "loss": 1.2192,
+      "step": 10316
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9529632714841994e-09,
+      "loss": 1.08,
+      "step": 10317
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8777352384423696e-09,
+      "loss": 1.0592,
+      "step": 10318
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8034777046648255e-09,
+      "loss": 0.9283,
+      "step": 10319
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7301906773624653e-09,
+      "loss": 1.0156,
+      "step": 10320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6578741636484885e-09,
+      "loss": 1.1135,
+      "step": 10321
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.586528170543945e-09,
+      "loss": 1.0634,
+      "step": 10322
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5161527049755165e-09,
+      "loss": 1.0806,
+      "step": 10323
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4467477737732946e-09,
+      "loss": 1.0266,
+      "step": 10324
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.378313383676334e-09,
+      "loss": 0.9464,
+      "step": 10325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.310849541327098e-09,
+      "loss": 1.1997,
+      "step": 10326
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.244356253275903e-09,
+      "loss": 0.9463,
+      "step": 10327
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.178833525977586e-09,
+      "loss": 1.0579,
+      "step": 10328
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.114281365792614e-09,
+      "loss": 1.0196,
+      "step": 10329
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0506997789870865e-09,
+      "loss": 1.0424,
+      "step": 10330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9880887717338425e-09,
+      "loss": 0.9975,
+      "step": 10331
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9264483501124643e-09,
+      "loss": 1.0764,
+      "step": 10332
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.865778520104833e-09,
+      "loss": 0.9031,
+      "step": 10333
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8060792876006815e-09,
+      "loss": 1.0737,
+      "step": 10334
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.747350658396485e-09,
+      "loss": 1.0745,
+      "step": 10335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6895926381932381e-09,
+      "loss": 0.9646,
+      "step": 10336
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6328052325975675e-09,
+      "loss": 1.0264,
+      "step": 10337
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5769884471239506e-09,
+      "loss": 1.0554,
+      "step": 10338
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5221422871880553e-09,
+      "loss": 1.0356,
+      "step": 10339
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.468266758116732e-09,
+      "loss": 1.0746,
+      "step": 10340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4153618651380208e-09,
+      "loss": 1.092,
+      "step": 10341
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3634276133900337e-09,
+      "loss": 0.9863,
+      "step": 10342
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3124640079131835e-09,
+      "loss": 1.0833,
+      "step": 10343
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2624710536546235e-09,
+      "loss": 1.1443,
+      "step": 10344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2134487554682495e-09,
+      "loss": 1.1394,
+      "step": 10345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1653971181124767e-09,
+      "loss": 1.0923,
+      "step": 10346
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1183161462524627e-09,
+      "loss": 1.0598,
+      "step": 10347
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0722058444589955e-09,
+      "loss": 1.092,
+      "step": 10348
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0270662172073842e-09,
+      "loss": 1.0204,
+      "step": 10349
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.82897268880789e-10,
+      "loss": 0.926,
+      "step": 10350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.396990037668919e-10,
+      "loss": 1.073,
+      "step": 10351
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.974714260578943e-10,
+      "loss": 1.1213,
+      "step": 10352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.562145398549604e-10,
+      "loss": 1.1545,
+      "step": 10353
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.159283491626646e-10,
+      "loss": 0.9441,
+      "step": 10354
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.766128578912124e-10,
+      "loss": 1.0408,
+      "step": 10355
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.382680698586608e-10,
+      "loss": 1.0204,
+      "step": 10356
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.008939887853672e-10,
+      "loss": 1.0415,
+      "step": 10357
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.644906183017608e-10,
+      "loss": 1.04,
+      "step": 10358
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.29057961939461e-10,
+      "loss": 1.0884,
+      "step": 10359
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.945960231401593e-10,
+      "loss": 1.0667,
+      "step": 10360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.611048052489576e-10,
+      "loss": 1.1392,
+      "step": 10361
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.285843115165889e-10,
+      "loss": 1.054,
+      "step": 10362
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.970345451005276e-10,
+      "loss": 0.9548,
+      "step": 10363
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.664555090627687e-10,
+      "loss": 1.0743,
+      "step": 10364
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.368472063731588e-10,
+      "loss": 0.8787,
+      "step": 10365
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.082096399049551e-10,
+      "loss": 1.0098,
+      "step": 10366
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.80542812438156e-10,
+      "loss": 0.848,
+      "step": 10367
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.538467266595014e-10,
+      "loss": 1.1274,
+      "step": 10368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.281213851602516e-10,
+      "loss": 1.0231,
+      "step": 10369
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.033667904372983e-10,
+      "loss": 1.1006,
+      "step": 10370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.795829448942744e-10,
+      "loss": 1.1641,
+      "step": 10371
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.567698508393335e-10,
+      "loss": 1.0289,
+      "step": 10372
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.349275104884807e-10,
+      "loss": 1.0409,
+      "step": 10373
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.140559259611319e-10,
+      "loss": 1.0183,
+      "step": 10374
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.94155099283444e-10,
+      "loss": 1.0459,
+      "step": 10375
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.752250323872051e-10,
+      "loss": 1.0195,
+      "step": 10376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5726572710983435e-10,
+      "loss": 1.063,
+      "step": 10377
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4027718519660227e-10,
+      "loss": 1.0548,
+      "step": 10378
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2425940829507987e-10,
+      "loss": 1.0754,
+      "step": 10379
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.092123979606896e-10,
+      "loss": 0.9457,
+      "step": 10380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.513615565448498e-11,
+      "loss": 1.114,
+      "step": 10381
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.203068274204029e-11,
+      "loss": 1.0225,
+      "step": 10382
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.989598049567115e-11,
+      "loss": 1.092,
+      "step": 10383
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.873205009443438e-11,
+      "loss": 1.022,
+      "step": 10384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.85388926219077e-11,
+      "loss": 1.0217,
+      "step": 10385
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.931650906729978e-11,
+      "loss": 1.0444,
+      "step": 10386
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.10649003254504e-11,
+      "loss": 0.933,
+      "step": 10387
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3784067197940575e-11,
+      "loss": 1.0835,
+      "step": 10388
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7474010390872153e-11,
+      "loss": 1.1177,
+      "step": 10389
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2134730518198468e-11,
+      "loss": 1.1799,
+      "step": 10390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.766228097283447e-12,
+      "loss": 0.9944,
+      "step": 10391
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.368503552232284e-12,
+      "loss": 0.9736,
+      "step": 10392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.94155721278122e-12,
+      "loss": 1.0859,
+      "step": 10393
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.853893154077583e-13,
+      "loss": 1.1167,
+      "step": 10394
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.0468,
+      "step": 10395
+    },
+    {
+      "epoch": 1.0,
+      "step": 10395,
+      "total_flos": 6.342282135570743e+18,
+      "train_loss": 1.0989449050117996,
+      "train_runtime": 125288.331,
+      "train_samples_per_second": 5.31,
+      "train_steps_per_second": 0.083
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 10395,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 5000,
+  "total_flos": 6.342282135570743e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}