diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31194 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998796833265153,
+  "eval_steps": 500,
+  "global_step": 5194,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-07,
+      "loss": 1.4568,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-07,
+      "loss": 1.3519,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 1.5023,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 1.3879,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.41025641025641e-07,
+      "loss": 1.3476,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 1.3944,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.974358974358975e-07,
+      "loss": 1.3298,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 1.3694,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 1.3934,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 1.3392,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4102564102564104e-06,
+      "loss": 1.3051,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.327,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.2895,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.794871794871795e-06,
+      "loss": 1.3203,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 1.1865,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 1.232,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1794871794871797e-06,
+      "loss": 1.2591,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.1657,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.435897435897436e-06,
+      "loss": 1.1379,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 1.1048,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 1.1536,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8205128205128207e-06,
+      "loss": 1.2042,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.948717948717949e-06,
+      "loss": 1.0689,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.1505,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.205128205128206e-06,
+      "loss": 1.1577,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.9724,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 1.0357,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 1.0858,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7179487179487184e-06,
+      "loss": 1.1227,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.1363,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.974358974358974e-06,
+      "loss": 1.0939,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 1.0564,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 1.0516,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.358974358974359e-06,
+      "loss": 1.0718,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.487179487179488e-06,
+      "loss": 1.0455,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 0.3501,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.743589743589744e-06,
+      "loss": 1.0192,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 1.0057,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 1.0616,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 1.0402,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.256410256410257e-06,
+      "loss": 1.0165,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 1.0511,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.512820512820514e-06,
+      "loss": 1.0954,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 1.0044,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 1.01,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.897435897435898e-06,
+      "loss": 1.0478,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.025641025641026e-06,
+      "loss": 1.0311,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.0003,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.282051282051282e-06,
+      "loss": 0.9851,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.410256410256412e-06,
+      "loss": 0.3461,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 1.0061,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.9571,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.794871794871796e-06,
+      "loss": 1.0099,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 1.0519,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.051282051282053e-06,
+      "loss": 0.9441,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.17948717948718e-06,
+      "loss": 0.9394,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 1.0082,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.435897435897437e-06,
+      "loss": 0.9644,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.564102564102564e-06,
+      "loss": 0.9238,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.0289,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.820512820512822e-06,
+      "loss": 0.9353,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.948717948717949e-06,
+      "loss": 1.017,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 0.925,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 0.931,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.9962,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 0.9417,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.58974358974359e-06,
+      "loss": 0.9704,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 0.9278,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 0.9516,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 0.9396,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.102564102564104e-06,
+      "loss": 0.9596,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.0327,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.358974358974359e-06,
+      "loss": 0.9283,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.487179487179487e-06,
+      "loss": 0.9486,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 0.931,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 0.9218,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.871794871794872e-06,
+      "loss": 0.9939,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.9909,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.012820512820513e-05,
+      "loss": 0.976,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 0.9342,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 0.8766,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0512820512820514e-05,
+      "loss": 0.9775,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0641025641025643e-05,
+      "loss": 1.0144,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.3455,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0897435897435898e-05,
+      "loss": 0.9266,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1025641025641028e-05,
+      "loss": 0.9761,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 0.9658,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 0.9105,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1410256410256411e-05,
+      "loss": 0.9438,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 0.9046,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.9083,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1794871794871796e-05,
+      "loss": 1.0173,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 1.0603,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2051282051282051e-05,
+      "loss": 0.9464,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.217948717948718e-05,
+      "loss": 0.965,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.9758,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2435897435897436e-05,
+      "loss": 0.9181,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2564102564102565e-05,
+      "loss": 0.3244,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 0.9932,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2820512820512823e-05,
+      "loss": 0.9672,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.294871794871795e-05,
+      "loss": 1.0127,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 0.9119,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3205128205128207e-05,
+      "loss": 0.3234,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.9469,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3461538461538463e-05,
+      "loss": 0.9539,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3589743589743592e-05,
+      "loss": 0.9101,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3717948717948718e-05,
+      "loss": 0.9747,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.2806,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3974358974358975e-05,
+      "loss": 0.9521,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4102564102564105e-05,
+      "loss": 0.9844,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4230769230769232e-05,
+      "loss": 0.914,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.435897435897436e-05,
+      "loss": 0.941,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4487179487179489e-05,
+      "loss": 0.9343,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 0.9326,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4743589743589745e-05,
+      "loss": 0.9253,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4871794871794874e-05,
+      "loss": 0.2921,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.9224,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5128205128205129e-05,
+      "loss": 0.8941,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5256410256410257e-05,
+      "loss": 0.972,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.9595,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5512820512820516e-05,
+      "loss": 0.9443,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5641025641025644e-05,
+      "loss": 0.976,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 0.9791,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 0.961,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.602564102564103e-05,
+      "loss": 0.9065,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 0.9956,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6282051282051282e-05,
+      "loss": 0.9804,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 0.9493,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 0.8838,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.9539,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6794871794871796e-05,
+      "loss": 0.9201,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.9596,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7051282051282053e-05,
+      "loss": 0.9175,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.717948717948718e-05,
+      "loss": 0.9245,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 0.9567,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7435897435897438e-05,
+      "loss": 0.9419,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7564102564102566e-05,
+      "loss": 0.9733,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 0.8945,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7820512820512823e-05,
+      "loss": 1.0129,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 0.9959,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.807692307692308e-05,
+      "loss": 0.991,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8205128205128208e-05,
+      "loss": 0.9743,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.9711,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.9616,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8589743589743593e-05,
+      "loss": 0.9673,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8717948717948718e-05,
+      "loss": 0.9397,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 0.9214,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 0.9446,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9102564102564106e-05,
+      "loss": 0.8336,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 0.3072,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.935897435897436e-05,
+      "loss": 0.8983,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 0.9582,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 0.9264,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9743589743589745e-05,
+      "loss": 0.2887,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9871794871794873e-05,
+      "loss": 0.9143,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.9342,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998055744173e-05,
+      "loss": 0.9571,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999992222977445e-05,
+      "loss": 0.943,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998250170209e-05,
+      "loss": 0.961,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999968891921883e-05,
+      "loss": 1.0011,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999951393642117e-05,
+      "loss": 0.9542,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999930006869593e-05,
+      "loss": 0.3099,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999904731612635e-05,
+      "loss": 0.9921,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999987556788107e-05,
+      "loss": 0.9413,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999984251568623e-05,
+      "loss": 0.9349,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999805575040976e-05,
+      "loss": 0.3087,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999976474595967e-05,
+      "loss": 1.0105,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999720028458185e-05,
+      "loss": 0.9711,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999671422553914e-05,
+      "loss": 0.9384,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999618928265758e-05,
+      "loss": 1.0137,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999562545614125e-05,
+      "loss": 0.8926,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999502274620942e-05,
+      "loss": 0.9739,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999438115309645e-05,
+      "loss": 1.0254,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999937006770518e-05,
+      "loss": 0.9197,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999298131834015e-05,
+      "loss": 0.9614,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999222307724112e-05,
+      "loss": 0.9707,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999142595404966e-05,
+      "loss": 1.0139,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999058994907564e-05,
+      "loss": 0.8866,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999897150626442e-05,
+      "loss": 0.9925,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999888012950955e-05,
+      "loss": 0.9797,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998784864678492e-05,
+      "loss": 0.9003,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998685711808282e-05,
+      "loss": 0.9317,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999858267093748e-05,
+      "loss": 0.984,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998475742106155e-05,
+      "loss": 0.9722,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999836492535588e-05,
+      "loss": 0.9725,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998250220729757e-05,
+      "loss": 0.9111,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998131628272383e-05,
+      "loss": 0.947,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999800914802987e-05,
+      "loss": 0.93,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997882780049847e-05,
+      "loss": 0.9675,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999775252438145e-05,
+      "loss": 0.9688,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997618381075335e-05,
+      "loss": 0.9215,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999748035018366e-05,
+      "loss": 0.8832,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997338431760096e-05,
+      "loss": 0.9339,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997192625859835e-05,
+      "loss": 0.8675,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997042932539568e-05,
+      "loss": 0.9353,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996889351857505e-05,
+      "loss": 0.9848,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996731883873367e-05,
+      "loss": 0.9291,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996570528648386e-05,
+      "loss": 0.9514,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99964052862453e-05,
+      "loss": 0.9889,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999623615672837e-05,
+      "loss": 0.9607,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999606314016336e-05,
+      "loss": 0.3042,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995886236617547e-05,
+      "loss": 0.8853,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995705446159724e-05,
+      "loss": 0.9036,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995520768860185e-05,
+      "loss": 0.8576,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995332204790746e-05,
+      "loss": 0.9105,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999513975402473e-05,
+      "loss": 0.9676,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999494341663697e-05,
+      "loss": 0.9226,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994743192703815e-05,
+      "loss": 0.9778,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994539082303118e-05,
+      "loss": 0.9867,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994331085514255e-05,
+      "loss": 0.9145,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99941192024181e-05,
+      "loss": 0.9365,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999390343309704e-05,
+      "loss": 0.9651,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993683777634987e-05,
+      "loss": 0.9727,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993460236117347e-05,
+      "loss": 0.8779,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999323280863105e-05,
+      "loss": 0.9335,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993001495264527e-05,
+      "loss": 0.9109,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992766296107726e-05,
+      "loss": 0.8789,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992527211252103e-05,
+      "loss": 0.8793,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992284240790628e-05,
+      "loss": 0.8699,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999203738481778e-05,
+      "loss": 0.2849,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991786643429553e-05,
+      "loss": 0.9907,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999153201672344e-05,
+      "loss": 0.28,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991273504798456e-05,
+      "loss": 0.949,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991011107755126e-05,
+      "loss": 0.8695,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990744825695485e-05,
+      "loss": 0.3352,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990474658723073e-05,
+      "loss": 0.9634,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990200606942943e-05,
+      "loss": 0.9579,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998992267046167e-05,
+      "loss": 0.9451,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989640849387318e-05,
+      "loss": 0.9514,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989355143829483e-05,
+      "loss": 0.9115,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989065553899255e-05,
+      "loss": 0.8458,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988772079709244e-05,
+      "loss": 0.8999,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998847472137357e-05,
+      "loss": 0.9769,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988173479007857e-05,
+      "loss": 0.919,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987868352729247e-05,
+      "loss": 0.9522,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987559342656387e-05,
+      "loss": 0.9444,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987246448909437e-05,
+      "loss": 0.9304,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998692967161006e-05,
+      "loss": 0.3286,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986609010881447e-05,
+      "loss": 0.9611,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998628446684828e-05,
+      "loss": 0.3254,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985956039636755e-05,
+      "loss": 0.9877,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998562372937459e-05,
+      "loss": 0.9584,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985287536190996e-05,
+      "loss": 0.9814,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984947460216708e-05,
+      "loss": 0.9538,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984603501583963e-05,
+      "loss": 0.8721,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984255660426508e-05,
+      "loss": 0.9236,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983903936879604e-05,
+      "loss": 0.9384,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998354833108002e-05,
+      "loss": 0.908,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983188843166028e-05,
+      "loss": 0.9034,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998282547327742e-05,
+      "loss": 0.9528,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982458221555493e-05,
+      "loss": 0.9685,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998208708814305e-05,
+      "loss": 0.9489,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998171207318441e-05,
+      "loss": 0.8893,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981333176825397e-05,
+      "loss": 0.9553,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980950399213344e-05,
+      "loss": 0.3389,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980563740497096e-05,
+      "loss": 0.9388,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980173200827004e-05,
+      "loss": 0.9072,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979778780354935e-05,
+      "loss": 0.9774,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979380479234253e-05,
+      "loss": 0.9616,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997897829761984e-05,
+      "loss": 1.0027,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978572235668085e-05,
+      "loss": 0.9101,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997816229353689e-05,
+      "loss": 0.9393,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977748471385654e-05,
+      "loss": 0.9229,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977330769375298e-05,
+      "loss": 0.9105,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976909187668247e-05,
+      "loss": 0.8892,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976483726428423e-05,
+      "loss": 0.9886,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997605438582128e-05,
+      "loss": 0.9169,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975621166013758e-05,
+      "loss": 0.9343,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997518406717432e-05,
+      "loss": 0.3304,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974743089472935e-05,
+      "loss": 0.9531,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974298233081072e-05,
+      "loss": 0.9479,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973849498171717e-05,
+      "loss": 0.9579,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997339688491936e-05,
+      "loss": 0.8983,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972940393499997e-05,
+      "loss": 0.9588,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972480024091143e-05,
+      "loss": 0.9858,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972015776871804e-05,
+      "loss": 0.9049,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971547652022512e-05,
+      "loss": 0.9411,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971075649725293e-05,
+      "loss": 0.9551,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970599770163686e-05,
+      "loss": 0.3036,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970120013522737e-05,
+      "loss": 0.9521,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969636379989e-05,
+      "loss": 0.9437,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996914886975054e-05,
+      "loss": 0.3228,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968657482996923e-05,
+      "loss": 0.9468,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968162219919222e-05,
+      "loss": 0.9587,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967663080710026e-05,
+      "loss": 0.9393,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967160065563425e-05,
+      "loss": 0.9231,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966653174675014e-05,
+      "loss": 0.9321,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99661424082419e-05,
+      "loss": 0.8851,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965627766462697e-05,
+      "loss": 0.8957,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965109249537523e-05,
+      "loss": 0.9716,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964586857668002e-05,
+      "loss": 0.858,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996406059105726e-05,
+      "loss": 0.9645,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963530449909954e-05,
+      "loss": 0.8745,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962996434432215e-05,
+      "loss": 0.8933,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962458544831703e-05,
+      "loss": 0.9066,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961916781317576e-05,
+      "loss": 0.9292,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961371144100497e-05,
+      "loss": 0.9256,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960821633392635e-05,
+      "loss": 0.8855,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960268249407674e-05,
+      "loss": 0.9114,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959710992360798e-05,
+      "loss": 0.9223,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995914986246869e-05,
+      "loss": 0.9407,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958584859949558e-05,
+      "loss": 0.9238,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995801598502309e-05,
+      "loss": 0.9465,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957443237910502e-05,
+      "loss": 0.9117,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956866618834508e-05,
+      "loss": 0.9311,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995628612801932e-05,
+      "loss": 1.0003,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955701765690668e-05,
+      "loss": 1.0106,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955113532075784e-05,
+      "loss": 0.9666,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99545214274034e-05,
+      "loss": 0.9482,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953925451903757e-05,
+      "loss": 0.9074,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99533256058086e-05,
+      "loss": 1.0054,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995272188935118e-05,
+      "loss": 0.9164,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952114302766255e-05,
+      "loss": 0.96,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995150284629008e-05,
+      "loss": 0.2877,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950887520160433e-05,
+      "loss": 0.8755,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995026832461657e-05,
+      "loss": 0.895,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949645259899272e-05,
+      "loss": 0.3097,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994901832625082e-05,
+      "loss": 0.8533,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948387523915e-05,
+      "loss": 0.9556,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994775285313709e-05,
+      "loss": 0.9579,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947114314163892e-05,
+      "loss": 0.3286,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99464719072437e-05,
+      "loss": 0.2982,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945825632626313e-05,
+      "loss": 0.9539,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945175490563036e-05,
+      "loss": 0.9009,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994452148130668e-05,
+      "loss": 0.9341,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943863605111558e-05,
+      "loss": 0.9141,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994320186223348e-05,
+      "loss": 0.9452,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994253625292977e-05,
+      "loss": 0.9695,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941866777459253e-05,
+      "loss": 0.9099,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994119343608225e-05,
+      "loss": 0.8982,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940516229060596e-05,
+      "loss": 0.918,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939835156657616e-05,
+      "loss": 0.9044,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939150219138154e-05,
+      "loss": 0.9417,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938461416768544e-05,
+      "loss": 0.307,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937768749816628e-05,
+      "loss": 0.8638,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937072218551752e-05,
+      "loss": 0.9746,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936371823244764e-05,
+      "loss": 0.9392,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935667564168008e-05,
+      "loss": 0.9261,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993495944159534e-05,
+      "loss": 0.9113,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934247455802115e-05,
+      "loss": 0.9067,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933531607065193e-05,
+      "loss": 0.8814,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993281189566292e-05,
+      "loss": 0.9195,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932088321875175e-05,
+      "loss": 0.951,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99313608859833e-05,
+      "loss": 0.9142,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930629588270177e-05,
+      "loss": 0.8602,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992989442902016e-05,
+      "loss": 1.0023,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929155408519124e-05,
+      "loss": 0.8959,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928412527054434e-05,
+      "loss": 0.9305,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992766578491496e-05,
+      "loss": 0.9086,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926915182391074e-05,
+      "loss": 0.9621,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926160719774656e-05,
+      "loss": 0.9244,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925402397359066e-05,
+      "loss": 0.9292,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992464021543919e-05,
+      "loss": 0.9144,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923874174311394e-05,
+      "loss": 0.9041,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923104274273564e-05,
+      "loss": 0.9005,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992233051562507e-05,
+      "loss": 0.8992,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921552898666793e-05,
+      "loss": 0.9709,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920771423701108e-05,
+      "loss": 0.9355,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919986091031885e-05,
+      "loss": 0.9014,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919196900964517e-05,
+      "loss": 0.9472,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991840385380587e-05,
+      "loss": 0.9154,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917606949864326e-05,
+      "loss": 0.9302,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991680618944976e-05,
+      "loss": 0.9218,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991600157287355e-05,
+      "loss": 0.9002,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991519310044857e-05,
+      "loss": 0.9319,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914380772489195e-05,
+      "loss": 0.9606,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99135645893113e-05,
+      "loss": 0.9307,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912744551232268e-05,
+      "loss": 0.9445,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991192065857096e-05,
+      "loss": 0.9788,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911092911647752e-05,
+      "loss": 0.9706,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910261310784515e-05,
+      "loss": 0.9621,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909425856304615e-05,
+      "loss": 0.9592,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990858654853292e-05,
+      "loss": 0.3369,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99077433877958e-05,
+      "loss": 0.8484,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906896374421116e-05,
+      "loss": 0.8951,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990604550873823e-05,
+      "loss": 0.904,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905190791077998e-05,
+      "loss": 0.9725,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904332221772788e-05,
+      "loss": 0.8955,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990346980115645e-05,
+      "loss": 0.91,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902603529564338e-05,
+      "loss": 0.9298,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901733407333303e-05,
+      "loss": 0.3098,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900859434801686e-05,
+      "loss": 0.8831,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899981612309344e-05,
+      "loss": 0.9436,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989909994019761e-05,
+      "loss": 0.9165,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989821441880933e-05,
+      "loss": 0.3223,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897325048488833e-05,
+      "loss": 0.8934,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989643182958196e-05,
+      "loss": 0.9448,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895534762436032e-05,
+      "loss": 0.9691,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894633847399883e-05,
+      "loss": 0.8683,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893729084823826e-05,
+      "loss": 0.9007,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892820475059687e-05,
+      "loss": 0.9022,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891908018460775e-05,
+      "loss": 0.8674,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98909917153819e-05,
+      "loss": 0.9657,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989007156617937e-05,
+      "loss": 0.9093,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889147571210984e-05,
+      "loss": 0.92,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888219730836038e-05,
+      "loss": 0.9695,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887288045415327e-05,
+      "loss": 0.3204,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886352515311134e-05,
+      "loss": 0.9266,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885413140887244e-05,
+      "loss": 0.2994,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884469922508934e-05,
+      "loss": 0.8913,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988352286054297e-05,
+      "loss": 0.9076,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988257195535763e-05,
+      "loss": 1.0125,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988161720732266e-05,
+      "loss": 0.8998,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880658616809324e-05,
+      "loss": 0.8945,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987969618419037e-05,
+      "loss": 0.9698,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878729909840043e-05,
+      "loss": 0.9192,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877759794134072e-05,
+      "loss": 0.8592,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876785837449696e-05,
+      "loss": 0.9541,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987580804016563e-05,
+      "loss": 0.3329,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874826402662105e-05,
+      "loss": 0.8888,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873840925320823e-05,
+      "loss": 0.962,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872851608524987e-05,
+      "loss": 0.8731,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871858452659293e-05,
+      "loss": 0.9315,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987086145810994e-05,
+      "loss": 0.9004,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869860625264602e-05,
+      "loss": 0.9573,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868855954512456e-05,
+      "loss": 0.9731,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867847446244172e-05,
+      "loss": 0.8945,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866835100851903e-05,
+      "loss": 0.9791,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986581891872931e-05,
+      "loss": 0.877,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986479890027153e-05,
+      "loss": 0.9302,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863775045875206e-05,
+      "loss": 0.9113,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986274735593845e-05,
+      "loss": 0.9437,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861715830860894e-05,
+      "loss": 0.902,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860680471043644e-05,
+      "loss": 0.9636,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98596412768893e-05,
+      "loss": 0.8819,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858598248801953e-05,
+      "loss": 0.8944,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985755138718719e-05,
+      "loss": 0.9192,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985650069245208e-05,
+      "loss": 0.333,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855446165005185e-05,
+      "loss": 0.3395,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854387805256566e-05,
+      "loss": 0.9123,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985332561361776e-05,
+      "loss": 0.953,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852259590501807e-05,
+      "loss": 0.974,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851189736323228e-05,
+      "loss": 0.951,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850116051498043e-05,
+      "loss": 0.3411,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849038536443746e-05,
+      "loss": 0.9468,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984795719157934e-05,
+      "loss": 0.9394,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98468720173253e-05,
+      "loss": 0.9042,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98457830141036e-05,
+      "loss": 0.8927,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98446901823377e-05,
+      "loss": 0.9282,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843593522452546e-05,
+      "loss": 0.9195,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984249303487458e-05,
+      "loss": 0.8814,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841388720031727e-05,
+      "loss": 0.9093,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840280578353397e-05,
+      "loss": 0.9111,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839168610270498e-05,
+      "loss": 0.9371,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838052816215416e-05,
+      "loss": 0.9311,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836933196622033e-05,
+      "loss": 0.8913,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835809751925708e-05,
+      "loss": 0.8834,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834682482563302e-05,
+      "loss": 0.9307,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833551388973147e-05,
+      "loss": 0.3145,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832416471595074e-05,
+      "loss": 0.9579,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831277730870396e-05,
+      "loss": 0.9425,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830135167241916e-05,
+      "loss": 0.8771,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828988781153916e-05,
+      "loss": 0.9582,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827838573052176e-05,
+      "loss": 0.8441,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982668454338395e-05,
+      "loss": 0.9754,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982552669259799e-05,
+      "loss": 0.9089,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982436502114452e-05,
+      "loss": 0.9226,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823199529475265e-05,
+      "loss": 0.8787,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982203021804342e-05,
+      "loss": 0.9562,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982085708730368e-05,
+      "loss": 0.2879,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819680137712214e-05,
+      "loss": 0.35,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818499369726684e-05,
+      "loss": 0.989,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817314783806226e-05,
+      "loss": 0.9409,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816126380411478e-05,
+      "loss": 0.9047,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814934160004543e-05,
+      "loss": 0.8565,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813738123049022e-05,
+      "loss": 0.9428,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812538270009995e-05,
+      "loss": 0.9269,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811334601354023e-05,
+      "loss": 0.9519,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810127117549158e-05,
+      "loss": 0.9447,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808915819064926e-05,
+      "loss": 0.9368,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980770070637235e-05,
+      "loss": 0.9246,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806481779943922e-05,
+      "loss": 0.9612,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805259040253624e-05,
+      "loss": 0.8641,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980403248777692e-05,
+      "loss": 0.8996,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802802122990758e-05,
+      "loss": 0.9209,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801567946373564e-05,
+      "loss": 0.9616,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800329958405254e-05,
+      "loss": 0.3289,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799088159567213e-05,
+      "loss": 0.3258,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979784255034232e-05,
+      "loss": 0.8599,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796593131214935e-05,
+      "loss": 0.8828,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795339902670893e-05,
+      "loss": 0.9342,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979408286519751e-05,
+      "loss": 0.9237,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792822019283594e-05,
+      "loss": 0.8947,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979155736541942e-05,
+      "loss": 0.3306,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790288904096754e-05,
+      "loss": 0.8377,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789016635808836e-05,
+      "loss": 0.8891,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978774056105039e-05,
+      "loss": 0.9015,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978646068031762e-05,
+      "loss": 0.885,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785176994108207e-05,
+      "loss": 0.8985,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783889502921315e-05,
+      "loss": 0.9202,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782598207257592e-05,
+      "loss": 0.9464,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978130310761915e-05,
+      "loss": 0.9708,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780004204509594e-05,
+      "loss": 0.9323,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778701498434007e-05,
+      "loss": 0.8968,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777394989898947e-05,
+      "loss": 0.9066,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776084679412443e-05,
+      "loss": 0.2967,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977477056748402e-05,
+      "loss": 0.9934,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773452654624675e-05,
+      "loss": 0.9435,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772130941346866e-05,
+      "loss": 0.8613,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770805428164558e-05,
+      "loss": 0.9679,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976947611559317e-05,
+      "loss": 0.2944,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768143004149607e-05,
+      "loss": 0.9105,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766806094352252e-05,
+      "loss": 0.3178,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765465386720963e-05,
+      "loss": 0.8613,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764120881777077e-05,
+      "loss": 0.3236,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976277258004341e-05,
+      "loss": 0.9392,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976142048204424e-05,
+      "loss": 0.8393,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760064588305347e-05,
+      "loss": 0.8781,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758704899353954e-05,
+      "loss": 0.8902,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757341415718794e-05,
+      "loss": 0.2885,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755974137930053e-05,
+      "loss": 0.8268,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754603066519395e-05,
+      "loss": 0.9048,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753228202019966e-05,
+      "loss": 0.2879,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751849544966385e-05,
+      "loss": 0.919,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750467095894738e-05,
+      "loss": 0.8974,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97490808553426e-05,
+      "loss": 0.9634,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747690823849012e-05,
+      "loss": 0.9004,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974629700195448e-05,
+      "loss": 0.8954,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744899390201006e-05,
+      "loss": 0.8987,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743497989132045e-05,
+      "loss": 0.8864,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974209279929253e-05,
+      "loss": 0.9171,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974068382122888e-05,
+      "loss": 0.9104,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739271055488972e-05,
+      "loss": 0.9247,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737854502622165e-05,
+      "loss": 0.9647,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736434163179284e-05,
+      "loss": 0.9334,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735010037712628e-05,
+      "loss": 0.9028,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733582126775978e-05,
+      "loss": 0.8699,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973215043092457e-05,
+      "loss": 0.9586,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730714950715125e-05,
+      "loss": 0.9047,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729275686705832e-05,
+      "loss": 0.8837,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727832639456345e-05,
+      "loss": 0.9438,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726385809527803e-05,
+      "loss": 0.9803,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724935197482798e-05,
+      "loss": 0.9027,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972348080388541e-05,
+      "loss": 0.31,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722022629301182e-05,
+      "loss": 0.9012,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720560674297118e-05,
+      "loss": 0.9086,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719094939441707e-05,
+      "loss": 0.9099,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717625425304903e-05,
+      "loss": 0.9629,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971615213245813e-05,
+      "loss": 0.8824,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714675061474274e-05,
+      "loss": 0.898,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713194212927694e-05,
+      "loss": 0.9043,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711709587394226e-05,
+      "loss": 0.8808,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710221185451166e-05,
+      "loss": 0.93,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708729007677282e-05,
+      "loss": 0.8678,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970723305465281e-05,
+      "loss": 0.3275,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705733326959448e-05,
+      "loss": 0.9482,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970422982518037e-05,
+      "loss": 0.9454,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702722549900214e-05,
+      "loss": 0.868,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701211501705083e-05,
+      "loss": 0.9062,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699696681182557e-05,
+      "loss": 0.9297,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969817808892167e-05,
+      "loss": 0.8995,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696655725512933e-05,
+      "loss": 0.9003,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695129591548315e-05,
+      "loss": 0.8993,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693599687621253e-05,
+      "loss": 0.9329,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692066014326655e-05,
+      "loss": 0.9043,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690528572260894e-05,
+      "loss": 0.3308,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688987362021802e-05,
+      "loss": 0.9179,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687442384208682e-05,
+      "loss": 0.911,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96858936394223e-05,
+      "loss": 0.9174,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684341128264884e-05,
+      "loss": 0.9178,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682784851340137e-05,
+      "loss": 0.9507,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681224809253217e-05,
+      "loss": 0.85,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679661002610743e-05,
+      "loss": 0.9042,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967809343202081e-05,
+      "loss": 0.9706,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676522098092965e-05,
+      "loss": 0.8951,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967494700143822e-05,
+      "loss": 0.8918,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673368142669066e-05,
+      "loss": 0.9113,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671785522399428e-05,
+      "loss": 0.9292,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967019914124472e-05,
+      "loss": 0.968,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668608999821806e-05,
+      "loss": 0.95,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667015098749013e-05,
+      "loss": 0.8177,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665417438646132e-05,
+      "loss": 0.9626,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966381602013441e-05,
+      "loss": 0.8781,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662210843836574e-05,
+      "loss": 0.8867,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660601910376784e-05,
+      "loss": 0.8869,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658989220380686e-05,
+      "loss": 0.8628,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965737277447537e-05,
+      "loss": 0.929,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655752573289396e-05,
+      "loss": 0.9017,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965412861745278e-05,
+      "loss": 0.9482,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652500907596998e-05,
+      "loss": 0.9202,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965086944435499e-05,
+      "loss": 0.8418,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649234228361148e-05,
+      "loss": 0.9401,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964759526025133e-05,
+      "loss": 0.9097,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645952540662855e-05,
+      "loss": 0.9277,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644306070234492e-05,
+      "loss": 0.869,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642655849606468e-05,
+      "loss": 0.9223,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641001879420482e-05,
+      "loss": 0.9154,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639344160319678e-05,
+      "loss": 0.8473,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637682692948666e-05,
+      "loss": 0.8664,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636017477953502e-05,
+      "loss": 0.9175,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963434851598171e-05,
+      "loss": 0.912,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963267580768227e-05,
+      "loss": 0.9017,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630999353705617e-05,
+      "loss": 0.9178,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962931915470364e-05,
+      "loss": 0.3294,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627635211329686e-05,
+      "loss": 0.9636,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625947524238564e-05,
+      "loss": 0.9262,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624256094086523e-05,
+      "loss": 0.9401,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622560921531285e-05,
+      "loss": 0.9628,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620862007232015e-05,
+      "loss": 0.9206,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961915935184934e-05,
+      "loss": 0.9237,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617452956045344e-05,
+      "loss": 0.9396,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615742820483554e-05,
+      "loss": 0.8277,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961402894582896e-05,
+      "loss": 0.9056,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612311332748008e-05,
+      "loss": 0.9275,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610589981908593e-05,
+      "loss": 0.9114,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608864893980056e-05,
+      "loss": 0.9162,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607136069633212e-05,
+      "loss": 0.878,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605403509540308e-05,
+      "loss": 0.8981,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603667214375053e-05,
+      "loss": 0.8907,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960192718481261e-05,
+      "loss": 0.9339,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960018342152959e-05,
+      "loss": 0.8789,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959843592520406e-05,
+      "loss": 0.9071,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596684696515533e-05,
+      "loss": 0.8472,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 0.8771,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593171044774807e-05,
+      "loss": 0.8756,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95914086230889e-05,
+      "loss": 0.8748,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589642471772565e-05,
+      "loss": 0.8996,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587872591512583e-05,
+      "loss": 0.8837,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586098982997168e-05,
+      "loss": 0.9413,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584321646915988e-05,
+      "loss": 0.3134,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958254058396017e-05,
+      "loss": 0.872,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580755794822278e-05,
+      "loss": 0.9513,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578967280196327e-05,
+      "loss": 0.3528,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957717504077778e-05,
+      "loss": 0.3559,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575379077263562e-05,
+      "loss": 0.8821,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573579390352028e-05,
+      "loss": 0.8887,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957177598074299e-05,
+      "loss": 0.9207,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569968849137705e-05,
+      "loss": 0.9216,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568157996238884e-05,
+      "loss": 0.912,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956634342275067e-05,
+      "loss": 0.8404,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956452512937867e-05,
+      "loss": 0.9446,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562703116829923e-05,
+      "loss": 0.8754,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560877385812924e-05,
+      "loss": 0.8601,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559047937037613e-05,
+      "loss": 0.9551,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955721477121537e-05,
+      "loss": 0.2967,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555377889059022e-05,
+      "loss": 0.9001,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553537291282846e-05,
+      "loss": 0.936,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551692978602557e-05,
+      "loss": 0.9123,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549844951735328e-05,
+      "loss": 0.8691,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547993211399753e-05,
+      "loss": 0.8922,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954613775831589e-05,
+      "loss": 0.9049,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954427859320524e-05,
+      "loss": 0.317,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542415716790727e-05,
+      "loss": 0.9282,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540549129796745e-05,
+      "loss": 0.8582,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538678832949113e-05,
+      "loss": 0.93,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536804826975105e-05,
+      "loss": 0.9261,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534927112603417e-05,
+      "loss": 0.9258,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533045690564214e-05,
+      "loss": 0.3147,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953116056158908e-05,
+      "loss": 0.8442,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529271726411055e-05,
+      "loss": 0.3093,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527379185764613e-05,
+      "loss": 0.8665,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952548294038567e-05,
+      "loss": 0.9183,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523582991011585e-05,
+      "loss": 0.9581,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952167933838115e-05,
+      "loss": 0.8841,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951977198323461e-05,
+      "loss": 0.8895,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517860926313644e-05,
+      "loss": 0.9653,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515946168361357e-05,
+      "loss": 0.9019,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514027710122313e-05,
+      "loss": 0.9226,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512105552342507e-05,
+      "loss": 0.8052,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510179695769367e-05,
+      "loss": 0.9782,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508250141151773e-05,
+      "loss": 0.9225,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506316889240027e-05,
+      "loss": 0.9431,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950437994078588e-05,
+      "loss": 0.3288,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502439296542516e-05,
+      "loss": 0.8851,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500494957264556e-05,
+      "loss": 0.912,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949854692370806e-05,
+      "loss": 0.8534,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949659519663052e-05,
+      "loss": 0.8485,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949463977679087e-05,
+      "loss": 0.8965,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949268066494948e-05,
+      "loss": 0.8528,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490717861868146e-05,
+      "loss": 0.863,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488751368310113e-05,
+      "loss": 0.8399,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486781185040052e-05,
+      "loss": 0.8217,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484807312824066e-05,
+      "loss": 0.9055,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482829752429703e-05,
+      "loss": 0.9028,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480848504625943e-05,
+      "loss": 0.9509,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947886357018319e-05,
+      "loss": 0.8888,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476874949873286e-05,
+      "loss": 0.8821,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474882644469517e-05,
+      "loss": 0.9697,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472886654746588e-05,
+      "loss": 0.9194,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470886981480642e-05,
+      "loss": 0.8667,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468883625449256e-05,
+      "loss": 0.9006,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466876587431436e-05,
+      "loss": 0.9269,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946486586820762e-05,
+      "loss": 0.9133,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946285146855968e-05,
+      "loss": 0.8983,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946083338927092e-05,
+      "loss": 0.8916,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945881163112607e-05,
+      "loss": 0.8983,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456786194911292e-05,
+      "loss": 0.8713,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945475708141418e-05,
+      "loss": 0.907,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452724291423756e-05,
+      "loss": 0.87,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450687825730476e-05,
+      "loss": 0.8694,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448647685126222e-05,
+      "loss": 0.9175,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94466038704043e-05,
+      "loss": 0.8631,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444556382359454e-05,
+      "loss": 0.9301,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944250522178785e-05,
+      "loss": 0.3297,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944045038948709e-05,
+      "loss": 0.9069,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943839188625619e-05,
+      "loss": 0.9039,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436329712895605e-05,
+      "loss": 0.9207,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943426387020721e-05,
+      "loss": 0.7671,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943219435899432e-05,
+      "loss": 0.9211,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430121180061654e-05,
+      "loss": 0.9221,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942804433421538e-05,
+      "loss": 0.8734,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425963822263077e-05,
+      "loss": 0.9348,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423879645013758e-05,
+      "loss": 0.8906,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942179180327785e-05,
+      "loss": 1.0078,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419700297867227e-05,
+      "loss": 0.8942,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941760512959516e-05,
+      "loss": 0.8357,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941550629927636e-05,
+      "loss": 0.8963,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413403807726964e-05,
+      "loss": 0.8919,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411297655764526e-05,
+      "loss": 0.9529,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409187844208026e-05,
+      "loss": 0.8649,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407074373877868e-05,
+      "loss": 0.8373,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404957245595872e-05,
+      "loss": 0.9045,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402836460185294e-05,
+      "loss": 0.8559,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400712018470795e-05,
+      "loss": 0.8496,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398583921278473e-05,
+      "loss": 0.9144,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939645216943584e-05,
+      "loss": 0.9174,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939431676377183e-05,
+      "loss": 0.8561,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392177705116796e-05,
+      "loss": 0.8751,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390034994302516e-05,
+      "loss": 0.9414,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387888632162185e-05,
+      "loss": 0.9299,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385738619530415e-05,
+      "loss": 0.8678,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383584957243245e-05,
+      "loss": 0.8798,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938142764613813e-05,
+      "loss": 0.9092,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937926668705394e-05,
+      "loss": 0.8347,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937710208083096e-05,
+      "loss": 0.8676,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374933828310917e-05,
+      "loss": 0.9424,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372761930336925e-05,
+      "loss": 0.8984,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370586387753532e-05,
+      "loss": 0.8925,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93684072014067e-05,
+      "loss": 0.8471,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936622437214381e-05,
+      "loss": 0.9831,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936403790081366e-05,
+      "loss": 0.9157,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361847788266455e-05,
+      "loss": 0.9186,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359654035353825e-05,
+      "loss": 0.3153,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357456642928817e-05,
+      "loss": 0.8549,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355255611845885e-05,
+      "loss": 0.8419,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935305094296091e-05,
+      "loss": 0.8921,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350842637131172e-05,
+      "loss": 0.8532,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934863069521537e-05,
+      "loss": 0.8834,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346415118073634e-05,
+      "loss": 0.8525,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344195906567488e-05,
+      "loss": 0.9247,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341973061559866e-05,
+      "loss": 0.8323,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339746583915135e-05,
+      "loss": 0.9091,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337516474499058e-05,
+      "loss": 0.8332,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335282734178816e-05,
+      "loss": 0.8467,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333045363823005e-05,
+      "loss": 0.9453,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933080436430163e-05,
+      "loss": 0.906,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93285597364861e-05,
+      "loss": 0.3344,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932631148124924e-05,
+      "loss": 0.9018,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324059599465296e-05,
+      "loss": 0.9587,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932180409200991e-05,
+      "loss": 0.3127,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319544959760135e-05,
+      "loss": 0.8722,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317282203594438e-05,
+      "loss": 0.9133,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315015824392703e-05,
+      "loss": 0.8696,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312745823036204e-05,
+      "loss": 0.8795,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931047220040764e-05,
+      "loss": 0.904,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308194957391108e-05,
+      "loss": 0.8587,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930591409487212e-05,
+      "loss": 0.9236,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303629613737592e-05,
+      "loss": 0.8841,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930134151487584e-05,
+      "loss": 0.9237,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299049799176602e-05,
+      "loss": 0.9318,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296754467531015e-05,
+      "loss": 0.8816,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929445552083161e-05,
+      "loss": 0.9545,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929215295997235e-05,
+      "loss": 0.9354,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928984678584858e-05,
+      "loss": 0.8912,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928753699935706e-05,
+      "loss": 0.9357,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285223601395953e-05,
+      "loss": 0.8599,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282906592864826e-05,
+      "loss": 0.8928,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280585974664653e-05,
+      "loss": 0.9081,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278261747697804e-05,
+      "loss": 0.925,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275933912868065e-05,
+      "loss": 0.9822,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927360247108061e-05,
+      "loss": 0.9067,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271267423242028e-05,
+      "loss": 0.9604,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268928770260298e-05,
+      "loss": 0.921,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266586513044818e-05,
+      "loss": 0.9018,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926424065250637e-05,
+      "loss": 0.9294,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261891189557146e-05,
+      "loss": 0.9085,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925953812511074e-05,
+      "loss": 0.9494,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925718146008214e-05,
+      "loss": 0.8778,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254821195387744e-05,
+      "loss": 0.9528,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252457331945336e-05,
+      "loss": 0.9089,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250089870674115e-05,
+      "loss": 0.8768,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247718812494662e-05,
+      "loss": 0.8957,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245344158328975e-05,
+      "loss": 0.8747,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924296590910043e-05,
+      "loss": 0.8946,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924058406573383e-05,
+      "loss": 0.9125,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238198629155337e-05,
+      "loss": 0.8749,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235809600292542e-05,
+      "loss": 0.8234,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923341698007442e-05,
+      "loss": 0.9414,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231020769431344e-05,
+      "loss": 0.9384,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922862096929508e-05,
+      "loss": 0.8475,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226217580598804e-05,
+      "loss": 0.8653,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223810604277063e-05,
+      "loss": 0.8698,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221400041265818e-05,
+      "loss": 1.0105,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921898589250242e-05,
+      "loss": 0.9253,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921656815892561e-05,
+      "loss": 0.9164,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214146841475536e-05,
+      "loss": 0.3583,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211721941093722e-05,
+      "loss": 0.9056,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209293458723093e-05,
+      "loss": 0.925,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206861395307973e-05,
+      "loss": 0.9387,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204425751794064e-05,
+      "loss": 0.8424,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201986529128477e-05,
+      "loss": 0.8522,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.91995437282597e-05,
+      "loss": 0.9293,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197097350137625e-05,
+      "loss": 0.8588,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194647395713525e-05,
+      "loss": 0.8799,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919219386594007e-05,
+      "loss": 0.9014,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189736761771318e-05,
+      "loss": 0.3627,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187276084162714e-05,
+      "loss": 0.8935,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184811834071097e-05,
+      "loss": 0.8921,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918234401245469e-05,
+      "loss": 0.9154,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179872620273116e-05,
+      "loss": 0.8847,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177397658487376e-05,
+      "loss": 0.871,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174919128059856e-05,
+      "loss": 0.8782,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172437029954342e-05,
+      "loss": 0.8341,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169951365136e-05,
+      "loss": 0.891,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916746213457138e-05,
+      "loss": 0.8761,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164969339228424e-05,
+      "loss": 0.8424,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916247298007646e-05,
+      "loss": 0.9145,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159973058086198e-05,
+      "loss": 0.9742,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9157469574229736e-05,
+      "loss": 0.8846,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915496252948056e-05,
+      "loss": 0.8704,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152451924813532e-05,
+      "loss": 0.9198,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149937761204907e-05,
+      "loss": 0.8942,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914742003963232e-05,
+      "loss": 0.3535,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144898761074785e-05,
+      "loss": 0.8421,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142373926512713e-05,
+      "loss": 0.803,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139845536927882e-05,
+      "loss": 0.9114,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137313593303465e-05,
+      "loss": 0.8684,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134778096624e-05,
+      "loss": 0.3676,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132239047875433e-05,
+      "loss": 0.9095,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129696448045067e-05,
+      "loss": 0.9018,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127150298121596e-05,
+      "loss": 0.9308,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124600599095094e-05,
+      "loss": 0.8331,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122047351957012e-05,
+      "loss": 0.8519,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119490557700192e-05,
+      "loss": 0.8833,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9116930217318834e-05,
+      "loss": 0.8741,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911436633180854e-05,
+      "loss": 0.9333,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911179890216627e-05,
+      "loss": 0.9668,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109227929390378e-05,
+      "loss": 0.8922,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9106653414480593e-05,
+      "loss": 0.8983,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910407535843801e-05,
+      "loss": 0.8608,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101493762265116e-05,
+      "loss": 0.9287,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909890862696576e-05,
+      "loss": 0.9242,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096319953545186e-05,
+      "loss": 0.9058,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093727743009993e-05,
+      "loss": 0.9634,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909113199636817e-05,
+      "loss": 0.877,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908853271462907e-05,
+      "loss": 0.9151,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9085929898803437e-05,
+      "loss": 0.9089,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908332354990337e-05,
+      "loss": 0.9434,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080713668942356e-05,
+      "loss": 0.9527,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078100256935247e-05,
+      "loss": 0.9727,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075483314898275e-05,
+      "loss": 0.9042,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072862843849036e-05,
+      "loss": 0.8832,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070238844806507e-05,
+      "loss": 0.9048,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906761131879103e-05,
+      "loss": 0.9461,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064980266824323e-05,
+      "loss": 0.8537,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906234568992948e-05,
+      "loss": 0.8919,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9059707589130946e-05,
+      "loss": 0.7777,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057065965454562e-05,
+      "loss": 0.8761,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905442081992752e-05,
+      "loss": 0.888,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905177215357839e-05,
+      "loss": 0.947,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049119967437108e-05,
+      "loss": 0.8899,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046464262534978e-05,
+      "loss": 0.847,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9043805039904677e-05,
+      "loss": 0.8682,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041142300580247e-05,
+      "loss": 0.8792,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038476045597093e-05,
+      "loss": 0.8756,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035806275991996e-05,
+      "loss": 0.9159,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90331329928031e-05,
+      "loss": 0.8413,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903045619706991e-05,
+      "loss": 0.8522,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90277758898333e-05,
+      "loss": 0.8654,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025092072135514e-05,
+      "loss": 0.8385,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022404745020165e-05,
+      "loss": 0.8885,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019713909532207e-05,
+      "loss": 0.3603,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017019566717984e-05,
+      "loss": 0.8713,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014321717625198e-05,
+      "loss": 0.8701,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011620363302898e-05,
+      "loss": 0.8562,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008915504801524e-05,
+      "loss": 0.8881,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900620714317285e-05,
+      "loss": 0.933,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003495279470035e-05,
+      "loss": 0.96,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900077991474759e-05,
+      "loss": 0.8647,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899806105006138e-05,
+      "loss": 0.877,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8995338686468648e-05,
+      "loss": 0.8826,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992612825027978e-05,
+      "loss": 0.885,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989883466799335e-05,
+      "loss": 0.898,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987150612844028e-05,
+      "loss": 0.9399,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984414264224727e-05,
+      "loss": 0.868,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981674422005472e-05,
+      "loss": 0.908,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978931087251648e-05,
+      "loss": 0.8519,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976184261030007e-05,
+      "loss": 0.7749,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897343394440865e-05,
+      "loss": 0.864,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970680138457052e-05,
+      "loss": 0.9146,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967922844246025e-05,
+      "loss": 0.9059,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965162062847743e-05,
+      "loss": 0.9132,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896239779533575e-05,
+      "loss": 0.9046,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959630042784928e-05,
+      "loss": 0.3484,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895685880627152e-05,
+      "loss": 0.8991,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954084086873124e-05,
+      "loss": 0.922,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951305885668695e-05,
+      "loss": 0.8879,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948524203738545e-05,
+      "loss": 0.9363,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945739042164325e-05,
+      "loss": 0.8976,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894295040202905e-05,
+      "loss": 0.8362,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940158284417093e-05,
+      "loss": 0.9014,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937362690414165e-05,
+      "loss": 0.9054,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893456362110734e-05,
+      "loss": 0.8829,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931761077585037e-05,
+      "loss": 0.925,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892895506093703e-05,
+      "loss": 0.8357,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926145572254437e-05,
+      "loss": 0.8532,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892333261262974e-05,
+      "loss": 0.8667,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920516183156753e-05,
+      "loss": 0.9792,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8917696284930654e-05,
+      "loss": 0.8818,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914872919047957e-05,
+      "loss": 0.9872,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912046086606537e-05,
+      "loss": 0.8876,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890921578870561e-05,
+      "loss": 0.9407,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890638202644574e-05,
+      "loss": 0.9298,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903544800928842e-05,
+      "loss": 0.912,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900704113258166e-05,
+      "loss": 0.8957,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897859964538323e-05,
+      "loss": 0.9206,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895012355875262e-05,
+      "loss": 0.841,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892161288376282e-05,
+      "loss": 0.8872,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889306763150023e-05,
+      "loss": 0.8475,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886448781306464e-05,
+      "loss": 0.9074,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888358734395694e-05,
+      "loss": 0.89,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880722452214123e-05,
+      "loss": 0.8846,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887785410719203e-05,
+      "loss": 0.8549,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874982310006024e-05,
+      "loss": 0.9086,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88721070617728e-05,
+      "loss": 0.859,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8869228363610406e-05,
+      "loss": 0.8681,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866346216638226e-05,
+      "loss": 0.8406,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863460621976985e-05,
+      "loss": 0.9152,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886057158074875e-05,
+      "loss": 0.8989,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885767909407693e-05,
+      "loss": 0.8512,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885478316308627e-05,
+      "loss": 0.8344,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885188378890286e-05,
+      "loss": 0.8949,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848980972654117e-05,
+      "loss": 0.8536,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846074715468813e-05,
+      "loss": 0.3276,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8843165018477045e-05,
+      "loss": 0.8819,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884025188281025e-05,
+      "loss": 0.8016,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8837335309601214e-05,
+      "loss": 0.9053,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834415299984035e-05,
+      "loss": 0.9696,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831491855094176e-05,
+      "loss": 0.8401,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828564976068414e-05,
+      "loss": 0.8922,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882563466404487e-05,
+      "loss": 0.8365,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8822700920163002e-05,
+      "loss": 0.9004,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88197637455636e-05,
+      "loss": 0.3288,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881682314138878e-05,
+      "loss": 0.8673,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8813879108782014e-05,
+      "loss": 0.9055,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881093164888808e-05,
+      "loss": 0.9087,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807980762853108e-05,
+      "loss": 0.8853,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805026451824547e-05,
+      "loss": 0.9035,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802068716951187e-05,
+      "loss": 0.8478,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879910755938315e-05,
+      "loss": 0.9009,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796142980271887e-05,
+      "loss": 0.8606,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793174980770167e-05,
+      "loss": 0.9304,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8790203562032114e-05,
+      "loss": 0.8955,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787228725213155e-05,
+      "loss": 0.8633,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878425047147007e-05,
+      "loss": 0.8894,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8781268801960947e-05,
+      "loss": 0.8823,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778283717845216e-05,
+      "loss": 0.9057,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775295220283633e-05,
+      "loss": 0.8404,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772303310438275e-05,
+      "loss": 0.8915,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8769307989472548e-05,
+      "loss": 0.8368,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8766309258551195e-05,
+      "loss": 0.9312,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763307118840264e-05,
+      "loss": 0.7447,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876030157150715e-05,
+      "loss": 0.8692,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757292617720558e-05,
+      "loss": 0.8495,
+      "step": 964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754280258650522e-05,
+      "loss": 0.8847,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751264495468412e-05,
+      "loss": 0.883,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748245329346895e-05,
+      "loss": 0.8415,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8745222761459995e-05,
+      "loss": 0.9591,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742196792983023e-05,
+      "loss": 0.8656,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739167425092644e-05,
+      "loss": 0.8617,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736134658966823e-05,
+      "loss": 0.9235,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8733098495784864e-05,
+      "loss": 0.8597,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730058936727367e-05,
+      "loss": 0.9243,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727015982976287e-05,
+      "loss": 0.8252,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723969635714863e-05,
+      "loss": 0.9024,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720919896127686e-05,
+      "loss": 0.8676,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717866765400638e-05,
+      "loss": 0.8752,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871481024472094e-05,
+      "loss": 0.8719,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871175033527712e-05,
+      "loss": 0.8499,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870868703825903e-05,
+      "loss": 0.8807,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705620354857833e-05,
+      "loss": 0.8967,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702550286266016e-05,
+      "loss": 0.8885,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869947683367738e-05,
+      "loss": 0.9571,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696399998287035e-05,
+      "loss": 0.9387,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869331978129141e-05,
+      "loss": 0.3255,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690236183888266e-05,
+      "loss": 0.782,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868714920727665e-05,
+      "loss": 0.8811,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868405885265694e-05,
+      "loss": 0.8535,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680965121230822e-05,
+      "loss": 0.8207,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86778680142013e-05,
+      "loss": 0.8853,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674767532772688e-05,
+      "loss": 0.8801,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671663678150605e-05,
+      "loss": 0.8092,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668556451542e-05,
+      "loss": 0.313,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665445854155114e-05,
+      "loss": 0.3428,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662331887199508e-05,
+      "loss": 0.885,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659214551886052e-05,
+      "loss": 0.8236,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656093849426927e-05,
+      "loss": 0.8598,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652969781035617e-05,
+      "loss": 0.7997,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8649842347926922e-05,
+      "loss": 0.8746,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646711551316953e-05,
+      "loss": 0.8648,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643577392423117e-05,
+      "loss": 0.908,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640439872464134e-05,
+      "loss": 0.8814,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8637298992660042e-05,
+      "loss": 0.3489,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634154754232166e-05,
+      "loss": 0.8872,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631007158403155e-05,
+      "loss": 0.9506,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862785620639695e-05,
+      "loss": 0.7797,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8624701899438802e-05,
+      "loss": 0.895,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862154423875527e-05,
+      "loss": 0.9094,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861838322557421e-05,
+      "loss": 0.8803,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615218861124787e-05,
+      "loss": 0.9088,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8612051146637477e-05,
+      "loss": 0.9306,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608880083344033e-05,
+      "loss": 0.8875,
+      "step": 1012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860570567247754e-05,
+      "loss": 0.8907,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860252791527236e-05,
+      "loss": 0.8227,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859934681296418e-05,
+      "loss": 0.8657,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859616236678997e-05,
+      "loss": 0.8903,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592974577988e-05,
+      "loss": 0.9113,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589783447797852e-05,
+      "loss": 0.9278,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85865889774604e-05,
+      "loss": 0.9245,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583391168217814e-05,
+      "loss": 0.9009,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858019002131357e-05,
+      "loss": 0.8893,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857698553799244e-05,
+      "loss": 0.9106,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857377771950048e-05,
+      "loss": 0.8742,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570566567085067e-05,
+      "loss": 0.9436,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567352081994852e-05,
+      "loss": 0.8631,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564134265479798e-05,
+      "loss": 0.8872,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856091311879115e-05,
+      "loss": 0.842,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855768864318146e-05,
+      "loss": 0.9502,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8554460839904566e-05,
+      "loss": 0.8599,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551229710215604e-05,
+      "loss": 0.8545,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547995255371007e-05,
+      "loss": 0.9069,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544757476628487e-05,
+      "loss": 0.948,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8541516375247063e-05,
+      "loss": 0.8791,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8538271952487044e-05,
+      "loss": 0.871,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535024209610026e-05,
+      "loss": 0.8344,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531773147878895e-05,
+      "loss": 0.8699,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852851876855783e-05,
+      "loss": 0.8736,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525261072912305e-05,
+      "loss": 0.8292,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522000062209078e-05,
+      "loss": 0.8455,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8518735737716188e-05,
+      "loss": 0.8451,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515468100702983e-05,
+      "loss": 0.8955,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851219715244008e-05,
+      "loss": 0.8758,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8508922894199393e-05,
+      "loss": 0.8992,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850564532725412e-05,
+      "loss": 0.8583,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502364452878755e-05,
+      "loss": 0.8644,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499080272349057e-05,
+      "loss": 0.9177,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849579278694209e-05,
+      "loss": 0.9296,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492501997936196e-05,
+      "loss": 0.8578,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489207906611003e-05,
+      "loss": 0.9054,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8485910514247424e-05,
+      "loss": 0.852,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8482609822127647e-05,
+      "loss": 0.8914,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8479305831535156e-05,
+      "loss": 0.9805,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8475998543754713e-05,
+      "loss": 0.8775,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847268796007235e-05,
+      "loss": 0.9081,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8469374081775405e-05,
+      "loss": 0.8875,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466056910152476e-05,
+      "loss": 0.889,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8462736446493452e-05,
+      "loss": 0.8794,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459412692089497e-05,
+      "loss": 0.849,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8456085648233054e-05,
+      "loss": 0.8583,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452755316217854e-05,
+      "loss": 0.3452,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8449421697338896e-05,
+      "loss": 0.9165,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446084792892464e-05,
+      "loss": 0.9162,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844274460417612e-05,
+      "loss": 0.8356,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843940113248869e-05,
+      "loss": 0.8346,
+      "step": 1064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8436054379130296e-05,
+      "loss": 0.8202,
+      "step": 1065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432704345402324e-05,
+      "loss": 0.8668,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429351032607444e-05,
+      "loss": 0.8695,
+      "step": 1067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425994442049585e-05,
+      "loss": 0.855,
+      "step": 1068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842263457503397e-05,
+      "loss": 0.9092,
+      "step": 1069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841927143286708e-05,
+      "loss": 0.8081,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8415905016856682e-05,
+      "loss": 0.8797,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 0.9232,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8409162368542778e-05,
+      "loss": 0.8341,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840578613886115e-05,
+      "loss": 0.9463,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8402406640579788e-05,
+      "loss": 0.9217,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399023875012808e-05,
+      "loss": 0.9009,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8395637843475607e-05,
+      "loss": 0.9436,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392248547284844e-05,
+      "loss": 0.9138,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388855987758455e-05,
+      "loss": 0.8997,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838546016621564e-05,
+      "loss": 0.8582,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838206108397686e-05,
+      "loss": 0.8775,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8378658742363862e-05,
+      "loss": 0.8659,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8375253142699646e-05,
+      "loss": 0.8635,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8371844286308486e-05,
+      "loss": 0.8694,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8368432174515915e-05,
+      "loss": 0.9014,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8365016808648743e-05,
+      "loss": 0.898,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836159819003503e-05,
+      "loss": 0.8828,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835817632000412e-05,
+      "loss": 0.8402,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8354751199886608e-05,
+      "loss": 0.8264,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835132283101435e-05,
+      "loss": 0.9217,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8347891214720477e-05,
+      "loss": 0.9211,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8344456352339377e-05,
+      "loss": 0.8517,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8341018245206696e-05,
+      "loss": 0.7747,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337576894659353e-05,
+      "loss": 0.9139,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334132302035515e-05,
+      "loss": 0.8864,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833068446867462e-05,
+      "loss": 0.9048,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832723339591736e-05,
+      "loss": 0.8134,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8323779085105684e-05,
+      "loss": 0.8775,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320321537582815e-05,
+      "loss": 0.8272,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8316860754693215e-05,
+      "loss": 0.9122,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831339673778262e-05,
+      "loss": 0.914,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8309929488198012e-05,
+      "loss": 0.9156,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8306459007287637e-05,
+      "loss": 0.9374,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302985296401e-05,
+      "loss": 0.8675,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829950835688885e-05,
+      "loss": 0.8819,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8296028190103205e-05,
+      "loss": 0.8788,
+      "step": 1106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8292544797397327e-05,
+      "loss": 0.851,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828905818012574e-05,
+      "loss": 0.885,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828556833964422e-05,
+      "loss": 0.8699,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8282075277309792e-05,
+      "loss": 0.8889,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827857899448074e-05,
+      "loss": 0.8832,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275079492516597e-05,
+      "loss": 0.8559,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271576772778154e-05,
+      "loss": 0.8345,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826807083662744e-05,
+      "loss": 0.9121,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826456168542774e-05,
+      "loss": 0.8733,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82610493205436e-05,
+      "loss": 0.9328,
+      "step": 1116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8257533743340806e-05,
+      "loss": 0.854,
+      "step": 1117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254014955186386e-05,
+      "loss": 0.8961,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8250492957448634e-05,
+      "loss": 0.9239,
+      "step": 1119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8246967751497083e-05,
+      "loss": 0.9109,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8243439338702504e-05,
+      "loss": 0.8832,
+      "step": 1121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8239907720436935e-05,
+      "loss": 0.8628,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236372898073642e-05,
+      "loss": 0.8437,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232834872987147e-05,
+      "loss": 0.8453,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8229293646553224e-05,
+      "loss": 0.3452,
+      "step": 1125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822574922014887e-05,
+      "loss": 0.8088,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822220159515234e-05,
+      "loss": 0.8802,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8218650772943144e-05,
+      "loss": 0.8218,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215096754902014e-05,
+      "loss": 0.9027,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8211539542410938e-05,
+      "loss": 0.9391,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207979136853136e-05,
+      "loss": 0.943,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204415539613085e-05,
+      "loss": 0.9503,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200848752076486e-05,
+      "loss": 0.9458,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197278775630295e-05,
+      "loss": 0.8602,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193705611662697e-05,
+      "loss": 0.8544,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819012926156312e-05,
+      "loss": 0.8705,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186549726722234e-05,
+      "loss": 0.9038,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182967008531948e-05,
+      "loss": 0.8641,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81793811083854e-05,
+      "loss": 0.8631,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817579202767698e-05,
+      "loss": 0.876,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172199767802298e-05,
+      "loss": 0.9418,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816860433015821e-05,
+      "loss": 0.871,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165005716142813e-05,
+      "loss": 0.8519,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161403927155424e-05,
+      "loss": 0.8914,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157798964596606e-05,
+      "loss": 0.8597,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154190829868152e-05,
+      "loss": 0.8678,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815057952437309e-05,
+      "loss": 0.8463,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8146965049515684e-05,
+      "loss": 0.3597,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8143347406701418e-05,
+      "loss": 0.8871,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139726597337026e-05,
+      "loss": 0.8819,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136102622830458e-05,
+      "loss": 0.8406,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132475484590905e-05,
+      "loss": 0.917,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812884518402878e-05,
+      "loss": 0.9036,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812521172255573e-05,
+      "loss": 0.897,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8121575101584632e-05,
+      "loss": 0.8538,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811793532252959e-05,
+      "loss": 0.8818,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114292386805935e-05,
+      "loss": 0.9074,
+      "step": 1157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8110646295830233e-05,
+      "loss": 0.8497,
+      "step": 1158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810699705102026e-05,
+      "loss": 0.9528,
+      "step": 1159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8103344653795042e-05,
+      "loss": 0.9144,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099689105574805e-05,
+      "loss": 0.821,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809603040778102e-05,
+      "loss": 0.8602,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8092368561836377e-05,
+      "loss": 0.3713,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808870356916479e-05,
+      "loss": 0.9162,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085035431191388e-05,
+      "loss": 0.8921,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8081364149342535e-05,
+      "loss": 0.8437,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8077689725045818e-05,
+      "loss": 0.8573,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074012159730034e-05,
+      "loss": 0.9126,
+      "step": 1168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807033145482521e-05,
+      "loss": 0.8953,
+      "step": 1169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8066647611762596e-05,
+      "loss": 0.8228,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062960631974657e-05,
+      "loss": 0.8725,
+      "step": 1171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805927051689508e-05,
+      "loss": 0.8447,
+      "step": 1172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805557726795877e-05,
+      "loss": 0.8506,
+      "step": 1173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051880886601845e-05,
+      "loss": 0.8828,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8048181374261656e-05,
+      "loss": 0.8808,
+      "step": 1175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804447873237676e-05,
+      "loss": 0.9648,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040772962386934e-05,
+      "loss": 0.8492,
+      "step": 1177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803706406573317e-05,
+      "loss": 0.8945,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033352043857677e-05,
+      "loss": 0.8506,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029636898203877e-05,
+      "loss": 0.8374,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802591863021641e-05,
+      "loss": 0.3273,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8022197241341134e-05,
+      "loss": 0.8809,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8018472733025108e-05,
+      "loss": 0.8952,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8014745106716612e-05,
+      "loss": 0.8805,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011014363865143e-05,
+      "loss": 0.9495,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8007280505921404e-05,
+      "loss": 0.8676,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8003543534337303e-05,
+      "loss": 0.3264,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7999803450565974e-05,
+      "loss": 0.884,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799606025606175e-05,
+      "loss": 0.8878,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992313952280175e-05,
+      "loss": 0.8822,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7988564540678e-05,
+      "loss": 0.8568,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79848120227132e-05,
+      "loss": 0.9043,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981056399844934e-05,
+      "loss": 0.842,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977297673533586e-05,
+      "loss": 0.8639,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7973535845240743e-05,
+      "loss": 0.8966,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796977091642919e-05,
+      "loss": 0.8972,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796600288856293e-05,
+      "loss": 0.8256,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962231763107158e-05,
+      "loss": 0.8743,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7958457541528292e-05,
+      "loss": 0.8877,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795468022529393e-05,
+      "loss": 0.9018,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7950899815872894e-05,
+      "loss": 0.8942,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7947116314735197e-05,
+      "loss": 0.9097,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943329723352057e-05,
+      "loss": 0.8847,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7939540043195898e-05,
+      "loss": 0.9033,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793574727574034e-05,
+      "loss": 0.8935,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7931951422460203e-05,
+      "loss": 0.8196,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7928152484831512e-05,
+      "loss": 0.9102,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792435046433149e-05,
+      "loss": 0.8529,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792054536243855e-05,
+      "loss": 0.8808,
+      "step": 1209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7916737180632316e-05,
+      "loss": 0.3679,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79129259203936e-05,
+      "loss": 0.9047,
+      "step": 1211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790911158320442e-05,
+      "loss": 0.8966,
+      "step": 1212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905294170547983e-05,
+      "loss": 0.8925,
+      "step": 1213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7901473683908694e-05,
+      "loss": 0.8265,
+      "step": 1214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897650124772154e-05,
+      "loss": 0.7964,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7893823494625158e-05,
+      "loss": 0.8366,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7889993794955695e-05,
+      "loss": 0.8563,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788616102725295e-05,
+      "loss": 0.9071,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.78823251930073e-05,
+      "loss": 0.8182,
+      "step": 1219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7878486293710308e-05,
+      "loss": 0.8485,
+      "step": 1220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7874644330854742e-05,
+      "loss": 0.8492,
+      "step": 1221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7870799305934547e-05,
+      "loss": 0.8702,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786695122044487e-05,
+      "loss": 0.8774,
+      "step": 1223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786310007588204e-05,
+      "loss": 0.8838,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7859245873743586e-05,
+      "loss": 0.8191,
+      "step": 1225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7855388615528213e-05,
+      "loss": 0.8632,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785152830273582e-05,
+      "loss": 0.8764,
+      "step": 1227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847664936867493e-05,
+      "loss": 0.9229,
+      "step": 1228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843798519425514e-05,
+      "loss": 0.8528,
+      "step": 1229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7839929051913336e-05,
+      "loss": 0.8455,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7836056535835607e-05,
+      "loss": 0.9237,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7832180972698166e-05,
+      "loss": 0.8623,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828302364008024e-05,
+      "loss": 0.9137,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782442071127338e-05,
+      "loss": 0.8517,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7820536016003625e-05,
+      "loss": 0.8676,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816648279709327e-05,
+      "loss": 0.8539,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7812757503902234e-05,
+      "loss": 0.8828,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7808863690095283e-05,
+      "loss": 0.91,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780496683980258e-05,
+      "loss": 0.8553,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7801066954539424e-05,
+      "loss": 0.8267,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797164035822295e-05,
+      "loss": 0.9493,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793258085168844e-05,
+      "loss": 0.8726,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77893491040979e-05,
+      "loss": 0.8722,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778543709412948e-05,
+      "loss": 0.9271,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778152205678477e-05,
+      "loss": 0.3291,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777760399358614e-05,
+      "loss": 0.8594,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773682906057133e-05,
+      "loss": 0.9093,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776975879572247e-05,
+      "loss": 0.8177,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776583166410804e-05,
+      "loss": 0.9041,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7761901512740918e-05,
+      "loss": 0.9276,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7757968343149348e-05,
+      "loss": 0.8019,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754032156862744e-05,
+      "loss": 0.8619,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750092955411697e-05,
+      "loss": 0.8921,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7746150740327978e-05,
+      "loss": 0.8524,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7742205513144507e-05,
+      "loss": 0.8145,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7738257275395404e-05,
+      "loss": 0.9272,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773430602861594e-05,
+      "loss": 0.8449,
+      "step": 1257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773035177434256e-05,
+      "loss": 0.8742,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7726394514112885e-05,
+      "loss": 0.9175,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722434249465695e-05,
+      "loss": 0.8867,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7718470981940953e-05,
+      "loss": 0.8729,
+      "step": 1261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771450471307977e-05,
+      "loss": 0.824,
+      "step": 1262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771053544442444e-05,
+      "loss": 0.3842,
+      "step": 1263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7706563177518414e-05,
+      "loss": 0.915,
+      "step": 1264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702587913906314e-05,
+      "loss": 0.862,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7698609655133925e-05,
+      "loss": 0.9035,
+      "step": 1266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7694628402748203e-05,
+      "loss": 0.8669,
+      "step": 1267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7690644158297258e-05,
+      "loss": 0.8527,
+      "step": 1268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7686656923330366e-05,
+      "loss": 0.9582,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7682666699397974e-05,
+      "loss": 0.9491,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7678673488051677e-05,
+      "loss": 0.8804,
+      "step": 1271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7674677290844245e-05,
+      "loss": 0.8213,
+      "step": 1272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767067810932961e-05,
+      "loss": 0.8223,
+      "step": 1273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7666675945062848e-05,
+      "loss": 0.8155,
+      "step": 1274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7662670799600207e-05,
+      "loss": 0.8789,
+      "step": 1275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76586626744991e-05,
+      "loss": 0.9145,
+      "step": 1276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765465157131808e-05,
+      "loss": 0.8696,
+      "step": 1277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765063749161688e-05,
+      "loss": 0.9154,
+      "step": 1278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7646620436956372e-05,
+      "loss": 0.8387,
+      "step": 1279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76426004088986e-05,
+      "loss": 0.9008,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638577409006746e-05,
+      "loss": 0.9311,
+      "step": 1281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7634551438845166e-05,
+      "loss": 0.8498,
+      "step": 1282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7630522499979358e-05,
+      "loss": 0.8708,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7626490593975987e-05,
+      "loss": 0.9374,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7622455722402857e-05,
+      "loss": 0.8784,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7618417886828934e-05,
+      "loss": 0.9406,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761437708882434e-05,
+      "loss": 0.9362,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761033332996034e-05,
+      "loss": 0.8486,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7606286611809353e-05,
+      "loss": 0.909,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7602236935944953e-05,
+      "loss": 0.8862,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7598184303941862e-05,
+      "loss": 0.8479,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7594128717375945e-05,
+      "loss": 0.8866,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7590070177824228e-05,
+      "loss": 0.7915,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7586008686864874e-05,
+      "loss": 0.8346,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7581944246077202e-05,
+      "loss": 0.8467,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7577876857041672e-05,
+      "loss": 0.8412,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75738065213399e-05,
+      "loss": 0.9268,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756973324055463e-05,
+      "loss": 0.8754,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7565657016269767e-05,
+      "loss": 0.8162,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7561577850070355e-05,
+      "loss": 0.8257,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7557495743542586e-05,
+      "loss": 0.9044,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7553410698273786e-05,
+      "loss": 0.8714,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754932271585243e-05,
+      "loss": 0.8708,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7545231797868144e-05,
+      "loss": 0.9053,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541137945911676e-05,
+      "loss": 0.8088,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753704116157493e-05,
+      "loss": 0.8597,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7532941446450943e-05,
+      "loss": 0.8905,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752883880213389e-05,
+      "loss": 0.8956,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7524733230219102e-05,
+      "loss": 0.921,
+      "step": 1309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520624732303024e-05,
+      "loss": 0.8267,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7516513309983253e-05,
+      "loss": 0.8896,
+      "step": 1311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7512398964858523e-05,
+      "loss": 0.8563,
+      "step": 1312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75082816985287e-05,
+      "loss": 0.8704,
+      "step": 1313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7504161512594782e-05,
+      "loss": 0.8906,
+      "step": 1314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7500038408658922e-05,
+      "loss": 0.8656,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749591238832438e-05,
+      "loss": 0.8989,
+      "step": 1316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749178345319557e-05,
+      "loss": 0.8539,
+      "step": 1317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7487651604878037e-05,
+      "loss": 0.8747,
+      "step": 1318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7483516844978445e-05,
+      "loss": 0.346,
+      "step": 1319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747937917510461e-05,
+      "loss": 0.8327,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747523859686546e-05,
+      "loss": 0.929,
+      "step": 1321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7471095111871076e-05,
+      "loss": 0.8625,
+      "step": 1322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7466948721732646e-05,
+      "loss": 0.9112,
+      "step": 1323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7462799428062505e-05,
+      "loss": 0.881,
+      "step": 1324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7458647232474106e-05,
+      "loss": 0.8492,
+      "step": 1325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454492136582042e-05,
+      "loss": 0.912,
+      "step": 1326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7450334142002022e-05,
+      "loss": 0.8684,
+      "step": 1327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7446173250350883e-05,
+      "loss": 0.8774,
+      "step": 1328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74420094632466e-05,
+      "loss": 0.8604,
+      "step": 1329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7437842782308262e-05,
+      "loss": 0.3318,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743367320915609e-05,
+      "loss": 0.8511,
+      "step": 1331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7429500745411426e-05,
+      "loss": 0.8309,
+      "step": 1332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7425325392696742e-05,
+      "loss": 0.8567,
+      "step": 1333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421147152635622e-05,
+      "loss": 0.8301,
+      "step": 1334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7416966026852784e-05,
+      "loss": 0.8478,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741278201697406e-05,
+      "loss": 0.8457,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740859512462641e-05,
+      "loss": 0.8884,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404405351437913e-05,
+      "loss": 0.8588,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740021269903776e-05,
+      "loss": 0.885,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7396017169056278e-05,
+      "loss": 0.3233,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7391818763124902e-05,
+      "loss": 0.9127,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738761748287618e-05,
+      "loss": 0.3269,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738341332994379e-05,
+      "loss": 0.7983,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7379206305962525e-05,
+      "loss": 0.8707,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7374996412568286e-05,
+      "loss": 0.9309,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7370783651398094e-05,
+      "loss": 0.8687,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7366568024090095e-05,
+      "loss": 0.3177,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736234953228353e-05,
+      "loss": 0.9639,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358128177618766e-05,
+      "loss": 0.883,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735390396173729e-05,
+      "loss": 0.8128,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734967688628168e-05,
+      "loss": 0.8856,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734544695289565e-05,
+      "loss": 0.8174,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7341214163224016e-05,
+      "loss": 0.8117,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336978518912695e-05,
+      "loss": 0.8593,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7332740021608722e-05,
+      "loss": 0.8682,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7328498672960253e-05,
+      "loss": 0.8687,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7324254474616528e-05,
+      "loss": 0.9189,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7320007428227918e-05,
+      "loss": 0.8795,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7315757535445884e-05,
+      "loss": 0.8381,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7311504797923006e-05,
+      "loss": 0.8675,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730724921731297e-05,
+      "loss": 0.3503,
+      "step": 1361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7302990795270557e-05,
+      "loss": 0.9103,
+      "step": 1362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729872953345166e-05,
+      "loss": 0.8519,
+      "step": 1363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294465433513278e-05,
+      "loss": 0.8458,
+      "step": 1364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7290198497113513e-05,
+      "loss": 0.871,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7285928725911562e-05,
+      "loss": 0.8795,
+      "step": 1366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728165612156774e-05,
+      "loss": 0.9048,
+      "step": 1367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7277380685743444e-05,
+      "loss": 0.9084,
+      "step": 1368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273102420101187e-05,
+      "loss": 0.8921,
+      "step": 1369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726882132630458e-05,
+      "loss": 0.8532,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264537406018323e-05,
+      "loss": 0.8638,
+      "step": 1371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7260250660908233e-05,
+      "loss": 0.835,
+      "step": 1372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7255961092641212e-05,
+      "loss": 0.8041,
+      "step": 1373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7251668702885262e-05,
+      "loss": 0.8284,
+      "step": 1374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7247373493309484e-05,
+      "loss": 0.8433,
+      "step": 1375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724307546558408e-05,
+      "loss": 0.905,
+      "step": 1376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238774621380335e-05,
+      "loss": 0.82,
+      "step": 1377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234470962370646e-05,
+      "loss": 0.8552,
+      "step": 1378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723016449022849e-05,
+      "loss": 0.9112,
+      "step": 1379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722585520662844e-05,
+      "loss": 0.8576,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722154311324618e-05,
+      "loss": 0.8487,
+      "step": 1381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721722821175846e-05,
+      "loss": 0.8989,
+      "step": 1382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212910503843135e-05,
+      "loss": 0.8468,
+      "step": 1383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7208589991179157e-05,
+      "loss": 0.859,
+      "step": 1384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720426667544656e-05,
+      "loss": 0.8783,
+      "step": 1385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719994055832647e-05,
+      "loss": 0.82,
+      "step": 1386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7195611641501103e-05,
+      "loss": 0.876,
+      "step": 1387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719127992665376e-05,
+      "loss": 0.8759,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7186945415468837e-05,
+      "loss": 0.8104,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7182608109631816e-05,
+      "loss": 0.8901,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7178268010829256e-05,
+      "loss": 0.9304,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173925120748816e-05,
+      "loss": 0.8656,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716957944107923e-05,
+      "loss": 0.8671,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7165230973510323e-05,
+      "loss": 0.3197,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160879719733002e-05,
+      "loss": 0.8534,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7156525681439252e-05,
+      "loss": 0.8849,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7152168860322156e-05,
+      "loss": 0.8597,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7147809258075862e-05,
+      "loss": 0.8464,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71434468763956e-05,
+      "loss": 0.8813,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7139081716977704e-05,
+      "loss": 0.9098,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7134713781519556e-05,
+      "loss": 0.9042,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713034307171964e-05,
+      "loss": 0.9052,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712596958927751e-05,
+      "loss": 0.9497,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71215933358938e-05,
+      "loss": 0.8387,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117214313270224e-05,
+      "loss": 0.8076,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711283252310956e-05,
+      "loss": 0.8915,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7108447967115686e-05,
+      "loss": 0.8185,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7104060646993535e-05,
+      "loss": 0.8396,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709967056444912e-05,
+      "loss": 0.8419,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709527772118953e-05,
+      "loss": 0.8474,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709088211892293e-05,
+      "loss": 0.8596,
+      "step": 1411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7086483759358556e-05,
+      "loss": 0.8426,
+      "step": 1412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7082082644206708e-05,
+      "loss": 0.8527,
+      "step": 1413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7077678775178774e-05,
+      "loss": 0.8588,
+      "step": 1414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7073272153987196e-05,
+      "loss": 0.8884,
+      "step": 1415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70688627823455e-05,
+      "loss": 0.8857,
+      "step": 1416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706445066196827e-05,
+      "loss": 0.8791,
+      "step": 1417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706003579457117e-05,
+      "loss": 0.8642,
+      "step": 1418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705561818187092e-05,
+      "loss": 0.9424,
+      "step": 1419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7051197825585316e-05,
+      "loss": 0.9505,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7046774727433223e-05,
+      "loss": 0.9084,
+      "step": 1421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704234888913456e-05,
+      "loss": 0.9076,
+      "step": 1422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037920312410325e-05,
+      "loss": 0.8742,
+      "step": 1423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7033488998982573e-05,
+      "loss": 0.8114,
+      "step": 1424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702905495057443e-05,
+      "loss": 0.8748,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7024618168910075e-05,
+      "loss": 0.8475,
+      "step": 1426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7020178655714757e-05,
+      "loss": 0.8709,
+      "step": 1427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7015736412714786e-05,
+      "loss": 0.8194,
+      "step": 1428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7011291441637537e-05,
+      "loss": 0.9424,
+      "step": 1429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7006843744211437e-05,
+      "loss": 0.897,
+      "step": 1430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7002393322165982e-05,
+      "loss": 0.9053,
+      "step": 1431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6997940177231722e-05,
+      "loss": 0.8309,
+      "step": 1432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699348431114027e-05,
+      "loss": 0.8581,
+      "step": 1433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698902572562429e-05,
+      "loss": 0.8265,
+      "step": 1434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698456442241751e-05,
+      "loss": 0.9032,
+      "step": 1435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6980100403254714e-05,
+      "loss": 0.8452,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975633669871748e-05,
+      "loss": 0.3438,
+      "step": 1437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6971164224005492e-05,
+      "loss": 0.8522,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6966692067393905e-05,
+      "loss": 0.8393,
+      "step": 1439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6962217201775986e-05,
+      "loss": 0.8897,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6957739628891797e-05,
+      "loss": 0.8948,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6953259350482444e-05,
+      "loss": 0.9439,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948776368290085e-05,
+      "loss": 0.8481,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694429068405794e-05,
+      "loss": 0.8477,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6939802299530266e-05,
+      "loss": 0.8556,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6935311216452382e-05,
+      "loss": 0.8517,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6930817436570644e-05,
+      "loss": 0.8172,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6926320961632475e-05,
+      "loss": 0.8949,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6921821793386324e-05,
+      "loss": 0.8886,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69173199335817e-05,
+      "loss": 0.8759,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6912815383969155e-05,
+      "loss": 0.8236,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908308146300297e-05,
+      "loss": 0.8646,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6903798222327765e-05,
+      "loss": 0.8316,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6899285613805246e-05,
+      "loss": 0.8915,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894770322487474e-05,
+      "loss": 0.8913,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689025235013023e-05,
+      "loss": 0.8624,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6885731698490322e-05,
+      "loss": 0.8416,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6881208369325622e-05,
+      "loss": 0.8556,
+      "step": 1458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687668236439503e-05,
+      "loss": 0.8027,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6872153685458484e-05,
+      "loss": 0.8888,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6867622334276966e-05,
+      "loss": 0.8676,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68630883126125e-05,
+      "loss": 0.8161,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6858551622228144e-05,
+      "loss": 0.9144,
+      "step": 1463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6854012264888e-05,
+      "loss": 0.9201,
+      "step": 1464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6849470242357197e-05,
+      "loss": 0.9102,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6844925556401906e-05,
+      "loss": 0.8687,
+      "step": 1466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840378208789334e-05,
+      "loss": 0.9208,
+      "step": 1467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6835828201287723e-05,
+      "loss": 0.8435,
+      "step": 1468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683127553566635e-05,
+      "loss": 0.815,
+      "step": 1469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682672021369552e-05,
+      "loss": 0.8673,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6822162237146585e-05,
+      "loss": 0.8248,
+      "step": 1471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681760160779191e-05,
+      "loss": 0.9112,
+      "step": 1472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6813038327404903e-05,
+      "loss": 0.8957,
+      "step": 1473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680847239776e-05,
+      "loss": 0.8165,
+      "step": 1474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803903820632674e-05,
+      "loss": 0.8321,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6799332597799413e-05,
+      "loss": 0.8407,
+      "step": 1476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679475873103775e-05,
+      "loss": 0.8868,
+      "step": 1477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6790182222126237e-05,
+      "loss": 0.8999,
+      "step": 1478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6785603072844446e-05,
+      "loss": 0.8811,
+      "step": 1479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6781021284972998e-05,
+      "loss": 0.8459,
+      "step": 1480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677643686029352e-05,
+      "loss": 0.8945,
+      "step": 1481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6771849800588665e-05,
+      "loss": 0.8707,
+      "step": 1482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6767260107642128e-05,
+      "loss": 0.9163,
+      "step": 1483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676266778323861e-05,
+      "loss": 0.8638,
+      "step": 1484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6758072829163843e-05,
+      "loss": 0.8651,
+      "step": 1485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6753475247204577e-05,
+      "loss": 0.8984,
+      "step": 1486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6748875039148592e-05,
+      "loss": 0.8627,
+      "step": 1487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6744272206784685e-05,
+      "loss": 0.8836,
+      "step": 1488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673966675190267e-05,
+      "loss": 0.899,
+      "step": 1489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673505867629338e-05,
+      "loss": 0.8875,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6730447981748672e-05,
+      "loss": 0.9329,
+      "step": 1491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6725834670061427e-05,
+      "loss": 0.8809,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672121874302553e-05,
+      "loss": 0.8236,
+      "step": 1493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671660020243589e-05,
+      "loss": 0.881,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6711979050088434e-05,
+      "loss": 0.8542,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67073552877801e-05,
+      "loss": 0.8261,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6702728917308847e-05,
+      "loss": 0.8269,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6698099940473644e-05,
+      "loss": 0.8628,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669346835907447e-05,
+      "loss": 0.8637,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6688834174912326e-05,
+      "loss": 0.8556,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6684197389789214e-05,
+      "loss": 0.3505,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679558005508157e-05,
+      "loss": 0.8031,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6674916023873186e-05,
+      "loss": 0.8395,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667027144668934e-05,
+      "loss": 0.9056,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6665624275762663e-05,
+      "loss": 0.8647,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6660974512900226e-05,
+      "loss": 0.8661,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665632215991008e-05,
+      "loss": 0.8574,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6651667218601306e-05,
+      "loss": 0.8964,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664700969078398e-05,
+      "loss": 0.8316,
+      "step": 1509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6642349578269192e-05,
+      "loss": 0.9567,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637686882869027e-05,
+      "loss": 0.8678,
+      "step": 1511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663302160639658e-05,
+      "loss": 0.83,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6628353750665954e-05,
+      "loss": 0.3299,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623683317492247e-05,
+      "loss": 0.8221,
+      "step": 1514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6619010308691558e-05,
+      "loss": 0.3422,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6614334726080997e-05,
+      "loss": 0.8906,
+      "step": 1516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6609656571478673e-05,
+      "loss": 0.9111,
+      "step": 1517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6604975846703685e-05,
+      "loss": 0.9043,
+      "step": 1518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6600292553576142e-05,
+      "loss": 0.8785,
+      "step": 1519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659560669391714e-05,
+      "loss": 0.8978,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6590918269548798e-05,
+      "loss": 0.8827,
+      "step": 1521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65862272822942e-05,
+      "loss": 0.8791,
+      "step": 1522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6581533733977448e-05,
+      "loss": 0.8692,
+      "step": 1523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657683762642363e-05,
+      "loss": 0.8988,
+      "step": 1524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6572138961458842e-05,
+      "loss": 0.872,
+      "step": 1525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6567437740910156e-05,
+      "loss": 0.8493,
+      "step": 1526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6562733966605648e-05,
+      "loss": 0.8933,
+      "step": 1527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655802764037439e-05,
+      "loss": 0.8372,
+      "step": 1528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655331876404644e-05,
+      "loss": 0.8652,
+      "step": 1529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.8647,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6543893368425664e-05,
+      "loss": 0.9396,
+      "step": 1531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.653917685279792e-05,
+      "loss": 0.8073,
+      "step": 1532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653445779440363e-05,
+      "loss": 0.8213,
+      "step": 1533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6529736195077814e-05,
+      "loss": 0.8545,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6525012056656463e-05,
+      "loss": 0.9015,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6520285380976576e-05,
+      "loss": 0.8359,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6515556169876115e-05,
+      "loss": 0.8905,
+      "step": 1537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6510824425194043e-05,
+      "loss": 0.8064,
+      "step": 1538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6506090148770306e-05,
+      "loss": 0.3621,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650135334244583e-05,
+      "loss": 0.9363,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6496614008062533e-05,
+      "loss": 0.863,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6491872147463307e-05,
+      "loss": 0.8532,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648712776249203e-05,
+      "loss": 0.8268,
+      "step": 1543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482380854993558e-05,
+      "loss": 0.8129,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6477631426813737e-05,
+      "loss": 0.8939,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647287947979939e-05,
+      "loss": 0.8382,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6468125015798307e-05,
+      "loss": 0.7867,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646336803665928e-05,
+      "loss": 0.8657,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6458608544232052e-05,
+      "loss": 0.912,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645384654036737e-05,
+      "loss": 0.8512,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6449082026916934e-05,
+      "loss": 0.8688,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644431500573344e-05,
+      "loss": 0.8115,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439545478670543e-05,
+      "loss": 0.8953,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6434773447582882e-05,
+      "loss": 0.8501,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6429998914326064e-05,
+      "loss": 0.8759,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6425221880756677e-05,
+      "loss": 0.8777,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642044234873227e-05,
+      "loss": 0.8925,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415660320111376e-05,
+      "loss": 0.9027,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641087579675349e-05,
+      "loss": 0.8392,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640608878051907e-05,
+      "loss": 0.8749,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6401299273269572e-05,
+      "loss": 0.7925,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6396507276867386e-05,
+      "loss": 0.8345,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639171279317589e-05,
+      "loss": 0.8484,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386915824059427e-05,
+      "loss": 0.8928,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63821163713833e-05,
+      "loss": 0.3469,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6377314437013786e-05,
+      "loss": 0.8984,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637251002281812e-05,
+      "loss": 0.8215,
+      "step": 1567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63677031306645e-05,
+      "loss": 0.863,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63628937624221e-05,
+      "loss": 0.8213,
+      "step": 1569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6358081919961044e-05,
+      "loss": 0.8839,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353267605152424e-05,
+      "loss": 0.8493,
+      "step": 1571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634845081986829e-05,
+      "loss": 0.8123,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6343631565981654e-05,
+      "loss": 0.8231,
+      "step": 1573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6338809845366493e-05,
+      "loss": 0.8497,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6333985659897737e-05,
+      "loss": 0.9347,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6329159011451273e-05,
+      "loss": 0.8793,
+      "step": 1576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6324329901903946e-05,
+      "loss": 0.7979,
+      "step": 1577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6319498333133572e-05,
+      "loss": 0.9107,
+      "step": 1578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6314664307018902e-05,
+      "loss": 0.8239,
+      "step": 1579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630982782543966e-05,
+      "loss": 0.8558,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6304988890276515e-05,
+      "loss": 0.9166,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630014750341109e-05,
+      "loss": 0.8933,
+      "step": 1582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6295303666725963e-05,
+      "loss": 0.883,
+      "step": 1583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629045738210467e-05,
+      "loss": 0.8824,
+      "step": 1584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6285608651431692e-05,
+      "loss": 0.8843,
+      "step": 1585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6280757476592467e-05,
+      "loss": 0.8508,
+      "step": 1586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6275903859473374e-05,
+      "loss": 0.8428,
+      "step": 1587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627104780196175e-05,
+      "loss": 0.8815,
+      "step": 1588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6266189305945878e-05,
+      "loss": 0.8719,
+      "step": 1589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6261328373314995e-05,
+      "loss": 0.9065,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6256465005959277e-05,
+      "loss": 0.9172,
+      "step": 1591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6251599205769845e-05,
+      "loss": 0.8765,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624673097463878e-05,
+      "loss": 0.8608,
+      "step": 1593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6241860314459096e-05,
+      "loss": 0.8713,
+      "step": 1594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6236987227124748e-05,
+      "loss": 0.8669,
+      "step": 1595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623211171453065e-05,
+      "loss": 0.9071,
+      "step": 1596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622723377857265e-05,
+      "loss": 0.8802,
+      "step": 1597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622235342114754e-05,
+      "loss": 0.8594,
+      "step": 1598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6217470644153045e-05,
+      "loss": 0.7979,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6212585449487842e-05,
+      "loss": 0.8357,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620769783905155e-05,
+      "loss": 0.9287,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620280781474472e-05,
+      "loss": 0.8657,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619791537846884e-05,
+      "loss": 0.8874,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6193020532126342e-05,
+      "loss": 0.9067,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618812327762059e-05,
+      "loss": 0.9014,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6183223616855888e-05,
+      "loss": 0.8994,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617832155173748e-05,
+      "loss": 0.9208,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6173417084171537e-05,
+      "loss": 0.8678,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6168510216065162e-05,
+      "loss": 0.8488,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163600949326405e-05,
+      "loss": 0.916,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615868928586423e-05,
+      "loss": 0.9051,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6153775227588547e-05,
+      "loss": 0.8706,
+      "step": 1612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61488587764102e-05,
+      "loss": 0.8982,
+      "step": 1613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614393993424095e-05,
+      "loss": 0.8755,
+      "step": 1614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6139018702993495e-05,
+      "loss": 0.9343,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613409508458146e-05,
+      "loss": 0.8584,
+      "step": 1616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6129169080919405e-05,
+      "loss": 0.9413,
+      "step": 1617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6124240693922808e-05,
+      "loss": 0.8973,
+      "step": 1618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611930992550808e-05,
+      "loss": 0.8661,
+      "step": 1619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6114376777592556e-05,
+      "loss": 0.8641,
+      "step": 1620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6109441252094495e-05,
+      "loss": 0.8625,
+      "step": 1621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610450335093308e-05,
+      "loss": 0.8527,
+      "step": 1622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6099563076028427e-05,
+      "loss": 0.902,
+      "step": 1623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609462042930156e-05,
+      "loss": 0.3733,
+      "step": 1624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6089675412674436e-05,
+      "loss": 0.3275,
+      "step": 1625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6084728028069935e-05,
+      "loss": 0.8519,
+      "step": 1626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6079778277411843e-05,
+      "loss": 0.825,
+      "step": 1627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6074826162624883e-05,
+      "loss": 0.9137,
+      "step": 1628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606987168563469e-05,
+      "loss": 0.9153,
+      "step": 1629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064914848367818e-05,
+      "loss": 0.9168,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605995565275174e-05,
+      "loss": 0.8409,
+      "step": 1631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605499410071484e-05,
+      "loss": 0.3482,
+      "step": 1632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605003019418643e-05,
+      "loss": 0.8155,
+      "step": 1633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604506393509673e-05,
+      "loss": 0.8665,
+      "step": 1634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6040095325376867e-05,
+      "loss": 0.8306,
+      "step": 1635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60351243669589e-05,
+      "loss": 0.956,
+      "step": 1636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603015106177579e-05,
+      "loss": 0.894,
+      "step": 1637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602517541176141e-05,
+      "loss": 0.791,
+      "step": 1638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602019741885055e-05,
+      "loss": 0.7888,
+      "step": 1639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6015217084978908e-05,
+      "loss": 0.8593,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601023441208309e-05,
+      "loss": 0.8814,
+      "step": 1641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6005249402100612e-05,
+      "loss": 0.8434,
+      "step": 1642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600026205696991e-05,
+      "loss": 0.8628,
+      "step": 1643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5995272378630313e-05,
+      "loss": 0.9369,
+      "step": 1644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5990280369022064e-05,
+      "loss": 0.8182,
+      "step": 1645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5985286030086315e-05,
+      "loss": 0.3493,
+      "step": 1646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5980289363765113e-05,
+      "loss": 0.9266,
+      "step": 1647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5975290372001423e-05,
+      "loss": 0.8619,
+      "step": 1648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597028905673911e-05,
+      "loss": 0.8368,
+      "step": 1649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5965285419922936e-05,
+      "loss": 0.8581,
+      "step": 1650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5960279463498576e-05,
+      "loss": 0.8915,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5955271189412596e-05,
+      "loss": 0.8626,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5950260599612478e-05,
+      "loss": 0.8624,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594524769604659e-05,
+      "loss": 0.8923,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940232480664205e-05,
+      "loss": 0.8598,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5935214955415496e-05,
+      "loss": 0.8535,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593019512225154e-05,
+      "loss": 0.9271,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5925172983124292e-05,
+      "loss": 0.9127,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5920148539986623e-05,
+      "loss": 0.8695,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5915121794792302e-05,
+      "loss": 0.8568,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5910092749495972e-05,
+      "loss": 0.894,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5905061406053186e-05,
+      "loss": 0.8898,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5900027766420396e-05,
+      "loss": 0.8856,
+      "step": 1663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5894991832554926e-05,
+      "loss": 0.8433,
+      "step": 1664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5889953606415014e-05,
+      "loss": 0.9103,
+      "step": 1665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5884913089959778e-05,
+      "loss": 0.8759,
+      "step": 1666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5879870285149232e-05,
+      "loss": 0.883,
+      "step": 1667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5874825193944264e-05,
+      "loss": 0.399,
+      "step": 1668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586977781830668e-05,
+      "loss": 0.8442,
+      "step": 1669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5864728160199145e-05,
+      "loss": 0.8543,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5859676221585233e-05,
+      "loss": 0.8892,
+      "step": 1671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5854622004429392e-05,
+      "loss": 0.3599,
+      "step": 1672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584956551069696e-05,
+      "loss": 0.8727,
+      "step": 1673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5844506742354163e-05,
+      "loss": 0.9238,
+      "step": 1674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5839445701368104e-05,
+      "loss": 0.8465,
+      "step": 1675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583438238970678e-05,
+      "loss": 0.8147,
+      "step": 1676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5829316809339068e-05,
+      "loss": 0.8407,
+      "step": 1677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5824248962234715e-05,
+      "loss": 0.8273,
+      "step": 1678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581917885036437e-05,
+      "loss": 0.821,
+      "step": 1679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5814106475699546e-05,
+      "loss": 0.8451,
+      "step": 1680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580903184021264e-05,
+      "loss": 0.8766,
+      "step": 1681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5803954945876933e-05,
+      "loss": 0.8427,
+      "step": 1682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5798875794666582e-05,
+      "loss": 0.9255,
+      "step": 1683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5793794388556617e-05,
+      "loss": 0.8402,
+      "step": 1684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5788710729522953e-05,
+      "loss": 0.8532,
+      "step": 1685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578362481954238e-05,
+      "loss": 0.8554,
+      "step": 1686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5778536660592545e-05,
+      "loss": 0.8639,
+      "step": 1687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5773446254652002e-05,
+      "loss": 0.8629,
+      "step": 1688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5768353603700152e-05,
+      "loss": 0.9332,
+      "step": 1689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5763258709717282e-05,
+      "loss": 0.7796,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5758161574684544e-05,
+      "loss": 0.3823,
+      "step": 1691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5753062200583967e-05,
+      "loss": 0.853,
+      "step": 1692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5747960589398447e-05,
+      "loss": 0.8624,
+      "step": 1693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5742856743111753e-05,
+      "loss": 0.8816,
+      "step": 1694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5737750663708526e-05,
+      "loss": 0.865,
+      "step": 1695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5732642353174257e-05,
+      "loss": 0.8528,
+      "step": 1696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727531813495335e-05,
+      "loss": 0.8739,
+      "step": 1697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572241904665899e-05,
+      "loss": 0.8364,
+      "step": 1698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5717304054653327e-05,
+      "loss": 0.8119,
+      "step": 1699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571218683946732e-05,
+      "loss": 0.9,
+      "step": 1700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707067403090803e-05,
+      "loss": 0.8798,
+      "step": 1701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5701945747514476e-05,
+      "loss": 0.8995,
+      "step": 1702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56968218747299e-05,
+      "loss": 0.8936,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5691695786729498e-05,
+      "loss": 0.8859,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5686567485506552e-05,
+      "loss": 0.7876,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5681436973055214e-05,
+      "loss": 0.9048,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567630425137049e-05,
+      "loss": 0.8359,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567116932244824e-05,
+      "loss": 0.8505,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5666032188285188e-05,
+      "loss": 0.8418,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566089285087892e-05,
+      "loss": 0.8943,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5655751312227863e-05,
+      "loss": 0.853,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565060757433132e-05,
+      "loss": 0.8404,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5645461639189437e-05,
+      "loss": 0.8133,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5640313508803217e-05,
+      "loss": 0.8986,
+      "step": 1714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5635163185174512e-05,
+      "loss": 0.861,
+      "step": 1715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5630010670306037e-05,
+      "loss": 0.3475,
+      "step": 1716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562485596620135e-05,
+      "loss": 0.9164,
+      "step": 1717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5619699074864864e-05,
+      "loss": 0.8476,
+      "step": 1718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561453999830184e-05,
+      "loss": 0.8536,
+      "step": 1719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56093787385184e-05,
+      "loss": 0.901,
+      "step": 1720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5604215297521498e-05,
+      "loss": 0.8655,
+      "step": 1721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5599049677318944e-05,
+      "loss": 0.9198,
+      "step": 1722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5593881879919403e-05,
+      "loss": 0.8915,
+      "step": 1723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5588711907332368e-05,
+      "loss": 0.841,
+      "step": 1724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5583539761568194e-05,
+      "loss": 0.8562,
+      "step": 1725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5578365444638078e-05,
+      "loss": 0.8752,
+      "step": 1726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5573188958554056e-05,
+      "loss": 0.8947,
+      "step": 1727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5568010305329012e-05,
+      "loss": 0.8968,
+      "step": 1728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5562829486976675e-05,
+      "loss": 0.863,
+      "step": 1729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5557646505511605e-05,
+      "loss": 0.9126,
+      "step": 1730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5552461362949214e-05,
+      "loss": 0.8492,
+      "step": 1731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5547274061305753e-05,
+      "loss": 0.8122,
+      "step": 1732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5542084602598305e-05,
+      "loss": 0.8145,
+      "step": 1733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55368929888448e-05,
+      "loss": 0.842,
+      "step": 1734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5531699222064005e-05,
+      "loss": 0.8591,
+      "step": 1735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5526503304275518e-05,
+      "loss": 0.8101,
+      "step": 1736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5521305237499785e-05,
+      "loss": 0.9186,
+      "step": 1737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551610502375807e-05,
+      "loss": 0.845,
+      "step": 1738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551090266507249e-05,
+      "loss": 0.8851,
+      "step": 1739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5505698163465986e-05,
+      "loss": 0.8464,
+      "step": 1740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5500491520962333e-05,
+      "loss": 0.799,
+      "step": 1741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549528273958614e-05,
+      "loss": 0.8755,
+      "step": 1742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5490071821362855e-05,
+      "loss": 0.8309,
+      "step": 1743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5484858768318738e-05,
+      "loss": 0.8491,
+      "step": 1744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54796435824809e-05,
+      "loss": 0.8472,
+      "step": 1745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5474426265877266e-05,
+      "loss": 0.8436,
+      "step": 1746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5469206820536595e-05,
+      "loss": 0.8813,
+      "step": 1747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5463985248488484e-05,
+      "loss": 0.3381,
+      "step": 1748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545876155176334e-05,
+      "loss": 0.8878,
+      "step": 1749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5453535732392397e-05,
+      "loss": 0.8982,
+      "step": 1750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5448307792407737e-05,
+      "loss": 0.8777,
+      "step": 1751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544307773384224e-05,
+      "loss": 0.9129,
+      "step": 1752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543784555872962e-05,
+      "loss": 0.8971,
+      "step": 1753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5432611269104416e-05,
+      "loss": 0.3624,
+      "step": 1754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5427374867001993e-05,
+      "loss": 0.8466,
+      "step": 1755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5422136354458522e-05,
+      "loss": 0.8618,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541689573351101e-05,
+      "loss": 0.7868,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5411653006197284e-05,
+      "loss": 0.805,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5406408174555978e-05,
+      "loss": 0.8878,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540116124062655e-05,
+      "loss": 0.8616,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539591220644928e-05,
+      "loss": 0.8781,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5390661074065257e-05,
+      "loss": 0.8426,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5385407845516395e-05,
+      "loss": 0.8507,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5380152522845416e-05,
+      "loss": 0.9023,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5374895108095858e-05,
+      "loss": 0.8337,
+      "step": 1765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5369635603312073e-05,
+      "loss": 0.8233,
+      "step": 1766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5364374010539225e-05,
+      "loss": 0.8378,
+      "step": 1767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535911033182329e-05,
+      "loss": 0.8866,
+      "step": 1768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5353844569211056e-05,
+      "loss": 0.891,
+      "step": 1769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5348576724750123e-05,
+      "loss": 0.8493,
+      "step": 1770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5343306800488896e-05,
+      "loss": 0.832,
+      "step": 1771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533803479847659e-05,
+      "loss": 0.3465,
+      "step": 1772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5332760720763232e-05,
+      "loss": 0.8266,
+      "step": 1773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5327484569399655e-05,
+      "loss": 0.3624,
+      "step": 1774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5322206346437492e-05,
+      "loss": 0.7877,
+      "step": 1775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5316926053929192e-05,
+      "loss": 0.8275,
+      "step": 1776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5311643693927997e-05,
+      "loss": 0.8834,
+      "step": 1777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5306359268487958e-05,
+      "loss": 0.8329,
+      "step": 1778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5301072779663936e-05,
+      "loss": 0.8615,
+      "step": 1779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5295784229511583e-05,
+      "loss": 0.8222,
+      "step": 1780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5290493620087362e-05,
+      "loss": 0.8702,
+      "step": 1781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528520095344853e-05,
+      "loss": 0.9025,
+      "step": 1782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5279906231653146e-05,
+      "loss": 0.8776,
+      "step": 1783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5274609456760073e-05,
+      "loss": 0.917,
+      "step": 1784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5269310630828963e-05,
+      "loss": 0.8748,
+      "step": 1785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5264009755920276e-05,
+      "loss": 0.3601,
+      "step": 1786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5258706834095256e-05,
+      "loss": 0.8585,
+      "step": 1787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5253401867415958e-05,
+      "loss": 0.778,
+      "step": 1788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524809485794522e-05,
+      "loss": 0.8286,
+      "step": 1789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5242785807746678e-05,
+      "loss": 0.3317,
+      "step": 1790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5237474718884767e-05,
+      "loss": 0.9186,
+      "step": 1791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5232161593424704e-05,
+      "loss": 0.8805,
+      "step": 1792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522684643343251e-05,
+      "loss": 0.9107,
+      "step": 1793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5221529240974987e-05,
+      "loss": 0.8445,
+      "step": 1794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5216210018119735e-05,
+      "loss": 0.8549,
+      "step": 1795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5210888766935138e-05,
+      "loss": 0.9064,
+      "step": 1796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5205565489490369e-05,
+      "loss": 0.8377,
+      "step": 1797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5200240187855392e-05,
+      "loss": 0.9088,
+      "step": 1798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519491286410096e-05,
+      "loss": 0.8105,
+      "step": 1799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5189583520298605e-05,
+      "loss": 0.8298,
+      "step": 1800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5184252158520647e-05,
+      "loss": 0.8312,
+      "step": 1801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5178918780840198e-05,
+      "loss": 0.8026,
+      "step": 1802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173583389331144e-05,
+      "loss": 0.8783,
+      "step": 1803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516824598606816e-05,
+      "loss": 0.3489,
+      "step": 1804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5162906573126698e-05,
+      "loss": 0.8839,
+      "step": 1805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5157565152583002e-05,
+      "loss": 0.9337,
+      "step": 1806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5152221726514082e-05,
+      "loss": 0.8586,
+      "step": 1807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5146876296997739e-05,
+      "loss": 0.8221,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5141528866112551e-05,
+      "loss": 0.9069,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136179435937871e-05,
+      "loss": 0.8727,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513082800855383e-05,
+      "loss": 0.9059,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5125474586041337e-05,
+      "loss": 0.8709,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5120119170482079e-05,
+      "loss": 0.9025,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114761763958513e-05,
+      "loss": 0.8478,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5109402368553876e-05,
+      "loss": 0.8544,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5104040986352173e-05,
+      "loss": 0.3489,
+      "step": 1816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5098677619438183e-05,
+      "loss": 0.9143,
+      "step": 1817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5093312269897458e-05,
+      "loss": 0.8388,
+      "step": 1818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5087944939816322e-05,
+      "loss": 0.8195,
+      "step": 1819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082575631281862e-05,
+      "loss": 0.8488,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5077204346381946e-05,
+      "loss": 0.8813,
+      "step": 1821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5071831087205205e-05,
+      "loss": 0.3726,
+      "step": 1822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5066455855841028e-05,
+      "loss": 0.8447,
+      "step": 1823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5061078654379593e-05,
+      "loss": 0.8497,
+      "step": 1824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5055699484911822e-05,
+      "loss": 0.8963,
+      "step": 1825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5050318349529416e-05,
+      "loss": 0.8214,
+      "step": 1826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5044935250324832e-05,
+      "loss": 0.337,
+      "step": 1827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50395501893913e-05,
+      "loss": 0.8652,
+      "step": 1828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5034163168822797e-05,
+      "loss": 0.3521,
+      "step": 1829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5028774190714082e-05,
+      "loss": 0.8838,
+      "step": 1830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5023383257160662e-05,
+      "loss": 0.8384,
+      "step": 1831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5017990370258806e-05,
+      "loss": 0.8777,
+      "step": 1832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5012595532105547e-05,
+      "loss": 0.8443,
+      "step": 1833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5007198744798673e-05,
+      "loss": 0.7931,
+      "step": 1834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5001800010436731e-05,
+      "loss": 0.8799,
+      "step": 1835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4996399331119026e-05,
+      "loss": 0.7597,
+      "step": 1836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4990996708945618e-05,
+      "loss": 0.8571,
+      "step": 1837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4985592146017322e-05,
+      "loss": 0.8455,
+      "step": 1838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4980185644435712e-05,
+      "loss": 0.8394,
+      "step": 1839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4974777206303106e-05,
+      "loss": 0.8783,
+      "step": 1840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4969366833722587e-05,
+      "loss": 0.8247,
+      "step": 1841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4963954528797983e-05,
+      "loss": 0.8545,
+      "step": 1842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4958540293633875e-05,
+      "loss": 0.9171,
+      "step": 1843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4953124130335594e-05,
+      "loss": 0.8497,
+      "step": 1844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4947706041009223e-05,
+      "loss": 0.7993,
+      "step": 1845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494228602776159e-05,
+      "loss": 0.8682,
+      "step": 1846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4936864092700278e-05,
+      "loss": 0.8408,
+      "step": 1847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4931440237933605e-05,
+      "loss": 0.877,
+      "step": 1848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4926014465570652e-05,
+      "loss": 0.8635,
+      "step": 1849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4920586777721231e-05,
+      "loss": 0.9223,
+      "step": 1850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491515717649591e-05,
+      "loss": 0.8842,
+      "step": 1851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4909725664005987e-05,
+      "loss": 0.8432,
+      "step": 1852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4904292242363521e-05,
+      "loss": 0.8515,
+      "step": 1853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48988569136813e-05,
+      "loss": 0.7853,
+      "step": 1854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4893419680072856e-05,
+      "loss": 0.8728,
+      "step": 1855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4887980543652467e-05,
+      "loss": 0.8487,
+      "step": 1856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488253950653515e-05,
+      "loss": 0.7933,
+      "step": 1857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4877096570836651e-05,
+      "loss": 0.9095,
+      "step": 1858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4871651738673469e-05,
+      "loss": 0.8194,
+      "step": 1859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486620501216283e-05,
+      "loss": 0.8598,
+      "step": 1860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4860756393422699e-05,
+      "loss": 0.8356,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4855305884571782e-05,
+      "loss": 0.7821,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4849853487729512e-05,
+      "loss": 0.8928,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4844399205016062e-05,
+      "loss": 0.8037,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483894303855233e-05,
+      "loss": 0.8697,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4833484990459963e-05,
+      "loss": 0.882,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482802506286132e-05,
+      "loss": 0.8342,
+      "step": 1867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4822563257879508e-05,
+      "loss": 0.3507,
+      "step": 1868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481709957763835e-05,
+      "loss": 0.8202,
+      "step": 1869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4811634024262409e-05,
+      "loss": 0.3453,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4806166599876967e-05,
+      "loss": 0.8592,
+      "step": 1871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4800697306608043e-05,
+      "loss": 0.8106,
+      "step": 1872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4795226146582378e-05,
+      "loss": 0.916,
+      "step": 1873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4789753121927435e-05,
+      "loss": 0.8437,
+      "step": 1874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4784278234771407e-05,
+      "loss": 0.8214,
+      "step": 1875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4778801487243214e-05,
+      "loss": 0.8848,
+      "step": 1876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.477332288147249e-05,
+      "loss": 0.3585,
+      "step": 1877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.47678424195896e-05,
+      "loss": 0.8107,
+      "step": 1878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476236010372563e-05,
+      "loss": 0.7975,
+      "step": 1879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4756875936012385e-05,
+      "loss": 0.837,
+      "step": 1880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4751389918582387e-05,
+      "loss": 0.9202,
+      "step": 1881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4745902053568882e-05,
+      "loss": 0.857,
+      "step": 1882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740412343105828e-05,
+      "loss": 0.7867,
+      "step": 1883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4734920789327913e-05,
+      "loss": 0.7834,
+      "step": 1884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4729427394370533e-05,
+      "loss": 0.7964,
+      "step": 1885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4723932160369794e-05,
+      "loss": 0.8146,
+      "step": 1886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4718435089462528e-05,
+      "loss": 0.8159,
+      "step": 1887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4712936183786281e-05,
+      "loss": 0.793,
+      "step": 1888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4707435445479306e-05,
+      "loss": 0.7957,
+      "step": 1889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470193287668057e-05,
+      "loss": 0.9027,
+      "step": 1890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4696428479529758e-05,
+      "loss": 0.8208,
+      "step": 1891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4690922256167256e-05,
+      "loss": 0.879,
+      "step": 1892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4685414208734168e-05,
+      "loss": 0.8636,
+      "step": 1893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4679904339372301e-05,
+      "loss": 0.8507,
+      "step": 1894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467439265022418e-05,
+      "loss": 0.818,
+      "step": 1895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4668879143433031e-05,
+      "loss": 0.8427,
+      "step": 1896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4663363821142784e-05,
+      "loss": 0.8401,
+      "step": 1897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4657846685498079e-05,
+      "loss": 0.8818,
+      "step": 1898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465232773864426e-05,
+      "loss": 0.9104,
+      "step": 1899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464680698272738e-05,
+      "loss": 0.8636,
+      "step": 1900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464128441989419e-05,
+      "loss": 0.8562,
+      "step": 1901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463576005229214e-05,
+      "loss": 0.8784,
+      "step": 1902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4630233882069396e-05,
+      "loss": 0.904,
+      "step": 1903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4624705911374808e-05,
+      "loss": 0.8388,
+      "step": 1904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4619176142357936e-05,
+      "loss": 0.8296,
+      "step": 1905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4613644577169035e-05,
+      "loss": 0.8867,
+      "step": 1906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4608111217959067e-05,
+      "loss": 0.833,
+      "step": 1907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460257606687968e-05,
+      "loss": 0.8388,
+      "step": 1908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4597039126083224e-05,
+      "loss": 0.8315,
+      "step": 1909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4591500397722746e-05,
+      "loss": 0.8787,
+      "step": 1910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458595988395199e-05,
+      "loss": 0.3211,
+      "step": 1911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4580417586925384e-05,
+      "loss": 0.8136,
+      "step": 1912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4574873508798063e-05,
+      "loss": 0.8759,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4569327651725847e-05,
+      "loss": 0.8697,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4563780017865248e-05,
+      "loss": 0.8237,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4558230609373469e-05,
+      "loss": 0.3732,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4552679428408402e-05,
+      "loss": 0.8484,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4547126477128637e-05,
+      "loss": 0.8539,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454157175769344e-05,
+      "loss": 0.877,
+      "step": 1919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4536015272262773e-05,
+      "loss": 0.8156,
+      "step": 1920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4530457022997282e-05,
+      "loss": 0.8678,
+      "step": 1921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4524897012058297e-05,
+      "loss": 0.8018,
+      "step": 1922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4519335241607833e-05,
+      "loss": 0.8459,
+      "step": 1923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513771713808594e-05,
+      "loss": 0.8622,
+      "step": 1924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4508206430823963e-05,
+      "loss": 0.8637,
+      "step": 1925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4502639394818007e-05,
+      "loss": 0.3605,
+      "step": 1926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4497070607955477e-05,
+      "loss": 0.8365,
+      "step": 1927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4491500072401796e-05,
+      "loss": 0.8619,
+      "step": 1928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448592779032308e-05,
+      "loss": 0.8505,
+      "step": 1929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448035376388611e-05,
+      "loss": 0.8142,
+      "step": 1930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4474777995258363e-05,
+      "loss": 0.8194,
+      "step": 1931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4469200486607972e-05,
+      "loss": 0.8612,
+      "step": 1932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4463621240103768e-05,
+      "loss": 0.8807,
+      "step": 1933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445804025791524e-05,
+      "loss": 0.8447,
+      "step": 1934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452457542212561e-05,
+      "loss": 0.9231,
+      "step": 1935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4446873095166578e-05,
+      "loss": 0.7727,
+      "step": 1936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4441286918948811e-05,
+      "loss": 0.8266,
+      "step": 1937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4435699015731449e-05,
+      "loss": 0.912,
+      "step": 1938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4430109387687352e-05,
+      "loss": 0.3528,
+      "step": 1939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4424518036990059e-05,
+      "loss": 0.8434,
+      "step": 1940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4418924965813767e-05,
+      "loss": 0.8204,
+      "step": 1941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4413330176333356e-05,
+      "loss": 0.8806,
+      "step": 1942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440773367072436e-05,
+      "loss": 0.9053,
+      "step": 1943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440213545116299e-05,
+      "loss": 0.7967,
+      "step": 1944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439653551982612e-05,
+      "loss": 0.8923,
+      "step": 1945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4390933878891286e-05,
+      "loss": 0.8679,
+      "step": 1946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.43853305305367e-05,
+      "loss": 0.912,
+      "step": 1947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4379725476941222e-05,
+      "loss": 0.8355,
+      "step": 1948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4374118720284388e-05,
+      "loss": 0.8231,
+      "step": 1949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4368510262746393e-05,
+      "loss": 0.8693,
+      "step": 1950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4362900106508094e-05,
+      "loss": 0.3395,
+      "step": 1951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4357288253750996e-05,
+      "loss": 0.9154,
+      "step": 1952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4351674706657288e-05,
+      "loss": 0.8992,
+      "step": 1953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4346059467409798e-05,
+      "loss": 0.8811,
+      "step": 1954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340442538192018e-05,
+      "loss": 0.921,
+      "step": 1955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4334823921188098e-05,
+      "loss": 0.8344,
+      "step": 1956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4329203618582845e-05,
+      "loss": 0.8487,
+      "step": 1957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4323581632561716e-05,
+      "loss": 0.8421,
+      "step": 1958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4317957965310834e-05,
+      "loss": 0.9141,
+      "step": 1959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4312332619016964e-05,
+      "loss": 0.321,
+      "step": 1960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4306705595867528e-05,
+      "loss": 0.8569,
+      "step": 1961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4301076898050601e-05,
+      "loss": 0.893,
+      "step": 1962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429544652775491e-05,
+      "loss": 0.8078,
+      "step": 1963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428981448716983e-05,
+      "loss": 0.8245,
+      "step": 1964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4284180778485388e-05,
+      "loss": 0.809,
+      "step": 1965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4278545403892257e-05,
+      "loss": 0.8677,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4272908365581758e-05,
+      "loss": 0.8156,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4267269665745862e-05,
+      "loss": 0.8261,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4261629306577182e-05,
+      "loss": 0.3241,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425598729026898e-05,
+      "loss": 0.8288,
+      "step": 1970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4250343619015162e-05,
+      "loss": 0.8765,
+      "step": 1971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4244698295010268e-05,
+      "loss": 0.7631,
+      "step": 1972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4239051320449501e-05,
+      "loss": 0.8891,
+      "step": 1973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4233402697528685e-05,
+      "loss": 0.8812,
+      "step": 1974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4227752428444294e-05,
+      "loss": 0.8388,
+      "step": 1975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4222100515393447e-05,
+      "loss": 0.8814,
+      "step": 1976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4216446960573893e-05,
+      "loss": 0.8665,
+      "step": 1977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4210791766184025e-05,
+      "loss": 0.7879,
+      "step": 1978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420513493442287e-05,
+      "loss": 0.8686,
+      "step": 1979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4199476467490097e-05,
+      "loss": 0.8985,
+      "step": 1980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4193816367586e-05,
+      "loss": 0.8175,
+      "step": 1981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4188154636911524e-05,
+      "loss": 0.8733,
+      "step": 1982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4182491277668233e-05,
+      "loss": 0.9354,
+      "step": 1983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4176826292058338e-05,
+      "loss": 0.7566,
+      "step": 1984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417115968228467e-05,
+      "loss": 0.817,
+      "step": 1985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41654914505507e-05,
+      "loss": 0.8439,
+      "step": 1986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4159821599060518e-05,
+      "loss": 0.8375,
+      "step": 1987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 0.8471,
+      "step": 1988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4148477045631094e-05,
+      "loss": 0.9243,
+      "step": 1989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4142802348103184e-05,
+      "loss": 0.784,
+      "step": 1990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137126039641755e-05,
+      "loss": 0.8472,
+      "step": 1991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4131448122454042e-05,
+      "loss": 0.8289,
+      "step": 1992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412576859874791e-05,
+      "loss": 0.7914,
+      "step": 1993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4120087470731854e-05,
+      "loss": 0.7988,
+      "step": 1994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411440474061498e-05,
+      "loss": 0.3274,
+      "step": 1995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4108720410607028e-05,
+      "loss": 0.7806,
+      "step": 1996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4103034482918359e-05,
+      "loss": 0.8058,
+      "step": 1997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4097346959759943e-05,
+      "loss": 0.9262,
+      "step": 1998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4091657843343392e-05,
+      "loss": 0.8576,
+      "step": 1999
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4085967135880915e-05,
+      "loss": 0.8594,
+      "step": 2000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408027483958536e-05,
+      "loss": 0.3709,
+      "step": 2001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4074580956670176e-05,
+      "loss": 0.839,
+      "step": 2002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4068885489349437e-05,
+      "loss": 0.9026,
+      "step": 2003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4063188439837831e-05,
+      "loss": 0.8113,
+      "step": 2004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4057489810350668e-05,
+      "loss": 0.7986,
+      "step": 2005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4051789603103865e-05,
+      "loss": 0.8512,
+      "step": 2006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4046087820313949e-05,
+      "loss": 0.769,
+      "step": 2007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4040384464198072e-05,
+      "loss": 0.7986,
+      "step": 2008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4034679536973985e-05,
+      "loss": 0.8041,
+      "step": 2009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4028973040860056e-05,
+      "loss": 0.8364,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4023264978075266e-05,
+      "loss": 0.3574,
+      "step": 2011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40175553508392e-05,
+      "loss": 0.8459,
+      "step": 2012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4011844161372054e-05,
+      "loss": 0.8485,
+      "step": 2013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4006131411894627e-05,
+      "loss": 0.8264,
+      "step": 2014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4000417104628332e-05,
+      "loss": 0.7969,
+      "step": 2015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3994701241795188e-05,
+      "loss": 0.83,
+      "step": 2016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3988983825617806e-05,
+      "loss": 0.8356,
+      "step": 2017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3983264858319416e-05,
+      "loss": 0.9146,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977544342123846e-05,
+      "loss": 0.9047,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3971822279255518e-05,
+      "loss": 0.8168,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396609867193947e-05,
+      "loss": 0.8603,
+      "step": 2021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396037352240133e-05,
+      "loss": 0.8519,
+      "step": 2022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954646832867332e-05,
+      "loss": 0.8983,
+      "step": 2023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3948918605564302e-05,
+      "loss": 0.8698,
+      "step": 2024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3943188842719669e-05,
+      "loss": 0.844,
+      "step": 2025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393745754656146e-05,
+      "loss": 0.8383,
+      "step": 2026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3931724719318293e-05,
+      "loss": 0.9192,
+      "step": 2027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3925990363219391e-05,
+      "loss": 0.8565,
+      "step": 2028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3920254480494558e-05,
+      "loss": 0.9146,
+      "step": 2029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39145170733742e-05,
+      "loss": 0.8646,
+      "step": 2030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3908778144089318e-05,
+      "loss": 0.8296,
+      "step": 2031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39030376948715e-05,
+      "loss": 0.3334,
+      "step": 2032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3897295727952922e-05,
+      "loss": 0.8568,
+      "step": 2033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3891552245566357e-05,
+      "loss": 0.8876,
+      "step": 2034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3885807249945166e-05,
+      "loss": 0.8592,
+      "step": 2035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38800607433233e-05,
+      "loss": 0.384,
+      "step": 2036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3874312727935293e-05,
+      "loss": 0.8162,
+      "step": 2037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3868563206016263e-05,
+      "loss": 0.8122,
+      "step": 2038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3862812179801922e-05,
+      "loss": 0.8449,
+      "step": 2039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3857059651528564e-05,
+      "loss": 0.8396,
+      "step": 2040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3851305623433064e-05,
+      "loss": 0.8347,
+      "step": 2041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3845550097752884e-05,
+      "loss": 0.7583,
+      "step": 2042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3839793076726069e-05,
+      "loss": 0.8551,
+      "step": 2043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3834034562591236e-05,
+      "loss": 0.822,
+      "step": 2044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38282745575876e-05,
+      "loss": 0.8617,
+      "step": 2045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3822513063954934e-05,
+      "loss": 0.8399,
+      "step": 2046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816750083933612e-05,
+      "loss": 0.8393,
+      "step": 2047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3810985619764573e-05,
+      "loss": 0.8012,
+      "step": 2048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3805219673689333e-05,
+      "loss": 0.8406,
+      "step": 2049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3799452247949985e-05,
+      "loss": 0.8954,
+      "step": 2050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3793683344789206e-05,
+      "loss": 0.8738,
+      "step": 2051
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3787912966450234e-05,
+      "loss": 0.8228,
+      "step": 2052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3782141115176893e-05,
+      "loss": 0.8599,
+      "step": 2053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377636779321357e-05,
+      "loss": 0.8145,
+      "step": 2054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3770593002805228e-05,
+      "loss": 0.7837,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3764816746197405e-05,
+      "loss": 0.8702,
+      "step": 2056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3759039025636199e-05,
+      "loss": 0.8294,
+      "step": 2057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3753259843368287e-05,
+      "loss": 0.8233,
+      "step": 2058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3747479201640914e-05,
+      "loss": 0.8541,
+      "step": 2059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3741697102701884e-05,
+      "loss": 0.7853,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3735913548799576e-05,
+      "loss": 0.9132,
+      "step": 2061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3730128542182926e-05,
+      "loss": 0.8232,
+      "step": 2062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372434208510145e-05,
+      "loss": 0.9015,
+      "step": 2063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.371855417980521e-05,
+      "loss": 0.7865,
+      "step": 2064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712764828544844e-05,
+      "loss": 0.8547,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3706974033571548e-05,
+      "loss": 0.8897,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3701181797137077e-05,
+      "loss": 0.8863,
+      "step": 2067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369538812149375e-05,
+      "loss": 0.8502,
+      "step": 2068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3689593008894448e-05,
+      "loss": 0.8703,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3683796461592604e-05,
+      "loss": 0.8593,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3677998481842213e-05,
+      "loss": 0.8856,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3672199071897825e-05,
+      "loss": 0.8419,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3666398234014546e-05,
+      "loss": 0.908,
+      "step": 2073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3660595970448044e-05,
+      "loss": 0.9032,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3654792283454529e-05,
+      "loss": 0.8801,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3648987175290777e-05,
+      "loss": 0.8463,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3643180648214105e-05,
+      "loss": 0.8528,
+      "step": 2077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3637372704482397e-05,
+      "loss": 0.8574,
+      "step": 2078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3631563346354072e-05,
+      "loss": 0.8684,
+      "step": 2079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3625752576088105e-05,
+      "loss": 0.842,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3619940395944027e-05,
+      "loss": 0.8719,
+      "step": 2081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614126808181904e-05,
+      "loss": 0.9065,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3608311815062361e-05,
+      "loss": 0.8722,
+      "step": 2083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3602495418846564e-05,
+      "loss": 0.8661,
+      "step": 2084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3596677621796221e-05,
+      "loss": 0.8274,
+      "step": 2085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3590858426173595e-05,
+      "loss": 0.8765,
+      "step": 2086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3585037834241485e-05,
+      "loss": 0.9132,
+      "step": 2087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3579215848263234e-05,
+      "loss": 0.8146,
+      "step": 2088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3573392470502729e-05,
+      "loss": 0.3226,
+      "step": 2089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3567567703224398e-05,
+      "loss": 0.8334,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3561741548693208e-05,
+      "loss": 0.861,
+      "step": 2091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3555914009174665e-05,
+      "loss": 0.7691,
+      "step": 2092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3550085086934815e-05,
+      "loss": 0.8946,
+      "step": 2093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3544254784240243e-05,
+      "loss": 0.9347,
+      "step": 2094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3538423103358066e-05,
+      "loss": 0.9008,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3532590046555939e-05,
+      "loss": 0.9608,
+      "step": 2096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3526755616102058e-05,
+      "loss": 0.8297,
+      "step": 2097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520919814265146e-05,
+      "loss": 0.8345,
+      "step": 2098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3515082643314457e-05,
+      "loss": 0.8838,
+      "step": 2099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3509244105519788e-05,
+      "loss": 0.8198,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3503404203151458e-05,
+      "loss": 0.9006,
+      "step": 2101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349756293848032e-05,
+      "loss": 0.898,
+      "step": 2102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3491720313777756e-05,
+      "loss": 0.7546,
+      "step": 2103
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3485876331315682e-05,
+      "loss": 0.9148,
+      "step": 2104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348003099336653e-05,
+      "loss": 0.8897,
+      "step": 2105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3474184302203275e-05,
+      "loss": 0.869,
+      "step": 2106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3468336260099401e-05,
+      "loss": 0.9289,
+      "step": 2107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346248686932893e-05,
+      "loss": 0.8406,
+      "step": 2108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3456636132166406e-05,
+      "loss": 0.7921,
+      "step": 2109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450784050886891e-05,
+      "loss": 0.8437,
+      "step": 2110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3444930627765978e-05,
+      "loss": 0.8391,
+      "step": 2111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3439075865079777e-05,
+      "loss": 0.7895,
+      "step": 2112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3433219765104913e-05,
+      "loss": 0.7828,
+      "step": 2113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3427362330118542e-05,
+      "loss": 0.8508,
+      "step": 2114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3421503562398339e-05,
+      "loss": 0.9453,
+      "step": 2115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3415643464222487e-05,
+      "loss": 0.8453,
+      "step": 2116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3409782037869692e-05,
+      "loss": 0.8531,
+      "step": 2117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3403919285619179e-05,
+      "loss": 0.837,
+      "step": 2118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339805520975068e-05,
+      "loss": 0.8502,
+      "step": 2119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3392189812544454e-05,
+      "loss": 0.8539,
+      "step": 2120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3386323096281268e-05,
+      "loss": 0.8493,
+      "step": 2121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3380455063242399e-05,
+      "loss": 0.8124,
+      "step": 2122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3374585715709638e-05,
+      "loss": 0.8792,
+      "step": 2123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3368715055965289e-05,
+      "loss": 0.8611,
+      "step": 2124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336284308629216e-05,
+      "loss": 0.8689,
+      "step": 2125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3356969808973583e-05,
+      "loss": 0.8024,
+      "step": 2126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3351095226293383e-05,
+      "loss": 0.8284,
+      "step": 2127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3345219340535897e-05,
+      "loss": 0.3417,
+      "step": 2128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339342153985973e-05,
+      "loss": 0.8187,
+      "step": 2129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3333463668928958e-05,
+      "loss": 0.8596,
+      "step": 2130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332758388765071e-05,
+      "loss": 0.8903,
+      "step": 2131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3321702812437591e-05,
+      "loss": 0.7892,
+      "step": 2132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3315820445576461e-05,
+      "loss": 0.8365,
+      "step": 2133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3309936789354687e-05,
+      "loss": 0.802,
+      "step": 2134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3304051846060135e-05,
+      "loss": 0.891,
+      "step": 2135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3298165617981171e-05,
+      "loss": 0.8064,
+      "step": 2136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3292278107406662e-05,
+      "loss": 0.8911,
+      "step": 2137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3286389316625974e-05,
+      "loss": 0.8098,
+      "step": 2138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3280499247928971e-05,
+      "loss": 0.8735,
+      "step": 2139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3274607903606011e-05,
+      "loss": 0.8228,
+      "step": 2140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326871528594795e-05,
+      "loss": 0.3304,
+      "step": 2141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3262821397246142e-05,
+      "loss": 0.8042,
+      "step": 2142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3256926239792432e-05,
+      "loss": 0.8744,
+      "step": 2143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3251029815879157e-05,
+      "loss": 0.788,
+      "step": 2144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3245132127799147e-05,
+      "loss": 0.8566,
+      "step": 2145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3239233177845732e-05,
+      "loss": 0.8758,
+      "step": 2146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3233332968312715e-05,
+      "loss": 0.8233,
+      "step": 2147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3227431501494406e-05,
+      "loss": 0.9063,
+      "step": 2148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3221528779685597e-05,
+      "loss": 0.8634,
+      "step": 2149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3215624805181568e-05,
+      "loss": 0.8344,
+      "step": 2150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3209719580278085e-05,
+      "loss": 0.7816,
+      "step": 2151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3203813107271401e-05,
+      "loss": 0.874,
+      "step": 2152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319790538845826e-05,
+      "loss": 0.7986,
+      "step": 2153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3191996426135876e-05,
+      "loss": 0.8754,
+      "step": 2154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3186086222601965e-05,
+      "loss": 0.8402,
+      "step": 2155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3180174780154712e-05,
+      "loss": 0.8245,
+      "step": 2156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3174262101092792e-05,
+      "loss": 0.8088,
+      "step": 2157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3168348187715353e-05,
+      "loss": 0.7835,
+      "step": 2158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3162433042322026e-05,
+      "loss": 0.8748,
+      "step": 2159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3156516667212929e-05,
+      "loss": 0.3646,
+      "step": 2160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3150599064688643e-05,
+      "loss": 0.8892,
+      "step": 2161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314468023705024e-05,
+      "loss": 0.8119,
+      "step": 2162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3138760186599262e-05,
+      "loss": 0.8929,
+      "step": 2163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3132838915637727e-05,
+      "loss": 0.8974,
+      "step": 2164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3126916426468128e-05,
+      "loss": 0.8616,
+      "step": 2165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120992721393433e-05,
+      "loss": 0.8438,
+      "step": 2166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311506780271708e-05,
+      "loss": 0.865,
+      "step": 2167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3109141672742983e-05,
+      "loss": 0.86,
+      "step": 2168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3103214333775522e-05,
+      "loss": 0.8509,
+      "step": 2169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309728578811955e-05,
+      "loss": 0.8449,
+      "step": 2170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309135603808039e-05,
+      "loss": 0.8823,
+      "step": 2171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308542508596383e-05,
+      "loss": 0.3795,
+      "step": 2172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307949293407613e-05,
+      "loss": 0.8347,
+      "step": 2173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3073559584724014e-05,
+      "loss": 0.8995,
+      "step": 2174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3067625040214673e-05,
+      "loss": 0.8382,
+      "step": 2175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3061689302855758e-05,
+      "loss": 0.3564,
+      "step": 2176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3055752374955389e-05,
+      "loss": 0.7935,
+      "step": 2177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3049814258822147e-05,
+      "loss": 0.8752,
+      "step": 2178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3043874956765077e-05,
+      "loss": 0.8157,
+      "step": 2179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3037934471093683e-05,
+      "loss": 0.7868,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303199280411793e-05,
+      "loss": 0.3585,
+      "step": 2181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3026049958148236e-05,
+      "loss": 0.8326,
+      "step": 2182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3020105935495496e-05,
+      "loss": 0.8424,
+      "step": 2183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3014160738471037e-05,
+      "loss": 0.3448,
+      "step": 2184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3008214369386667e-05,
+      "loss": 0.8468,
+      "step": 2185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300226683055463e-05,
+      "loss": 0.8744,
+      "step": 2186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2996318124287638e-05,
+      "loss": 0.8856,
+      "step": 2187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299036825289885e-05,
+      "loss": 0.8113,
+      "step": 2188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2984417218701883e-05,
+      "loss": 0.718,
+      "step": 2189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2978465024010803e-05,
+      "loss": 0.84,
+      "step": 2190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2972511671140127e-05,
+      "loss": 0.3543,
+      "step": 2191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2966557162404823e-05,
+      "loss": 0.8958,
+      "step": 2192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2960601500120307e-05,
+      "loss": 0.7898,
+      "step": 2193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2954644686602449e-05,
+      "loss": 0.8361,
+      "step": 2194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2948686724167559e-05,
+      "loss": 0.8468,
+      "step": 2195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29427276151324e-05,
+      "loss": 0.3299,
+      "step": 2196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293676736181418e-05,
+      "loss": 0.7764,
+      "step": 2197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2930805966530548e-05,
+      "loss": 0.9029,
+      "step": 2198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29248434315996e-05,
+      "loss": 0.3524,
+      "step": 2199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2918879759339874e-05,
+      "loss": 0.8303,
+      "step": 2200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2912914952070352e-05,
+      "loss": 0.8532,
+      "step": 2201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2906949012110456e-05,
+      "loss": 0.3371,
+      "step": 2202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2900981941780049e-05,
+      "loss": 0.8362,
+      "step": 2203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2895013743399432e-05,
+      "loss": 0.8149,
+      "step": 2204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2889044419289347e-05,
+      "loss": 0.8417,
+      "step": 2205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2883073971770971e-05,
+      "loss": 0.7742,
+      "step": 2206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2877102403165922e-05,
+      "loss": 0.8429,
+      "step": 2207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2871129715796253e-05,
+      "loss": 0.857,
+      "step": 2208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2865155911984445e-05,
+      "loss": 0.8326,
+      "step": 2209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2859180994053424e-05,
+      "loss": 0.8147,
+      "step": 2210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2853204964326536e-05,
+      "loss": 0.7957,
+      "step": 2211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2847227825127578e-05,
+      "loss": 0.8452,
+      "step": 2212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2841249578780756e-05,
+      "loss": 0.8763,
+      "step": 2213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283527022761073e-05,
+      "loss": 0.7642,
+      "step": 2214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2829289773942567e-05,
+      "loss": 0.8617,
+      "step": 2215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2823308220101777e-05,
+      "loss": 0.811,
+      "step": 2216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817325568414299e-05,
+      "loss": 0.8294,
+      "step": 2217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2811341821206487e-05,
+      "loss": 0.8481,
+      "step": 2218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2805356980805132e-05,
+      "loss": 0.3599,
+      "step": 2219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2799371049537448e-05,
+      "loss": 0.8702,
+      "step": 2220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2793384029731066e-05,
+      "loss": 0.8375,
+      "step": 2221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2787395923714049e-05,
+      "loss": 0.7968,
+      "step": 2222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2781406733814879e-05,
+      "loss": 0.9235,
+      "step": 2223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2775416462362458e-05,
+      "loss": 0.8636,
+      "step": 2224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.276942511168611e-05,
+      "loss": 0.8524,
+      "step": 2225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2763432684115582e-05,
+      "loss": 0.8852,
+      "step": 2226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2757439181981033e-05,
+      "loss": 0.8456,
+      "step": 2227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2751444607613044e-05,
+      "loss": 0.8737,
+      "step": 2228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2745448963342612e-05,
+      "loss": 0.8553,
+      "step": 2229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2739452251501152e-05,
+      "loss": 0.821,
+      "step": 2230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2733454474420493e-05,
+      "loss": 0.3743,
+      "step": 2231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2727455634432873e-05,
+      "loss": 0.8295,
+      "step": 2232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2721455733870952e-05,
+      "loss": 0.8782,
+      "step": 2233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2715454775067798e-05,
+      "loss": 0.843,
+      "step": 2234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2709452760356884e-05,
+      "loss": 0.8489,
+      "step": 2235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2703449692072112e-05,
+      "loss": 0.864,
+      "step": 2236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697445572547773e-05,
+      "loss": 0.8445,
+      "step": 2237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2691440404118582e-05,
+      "loss": 0.8405,
+      "step": 2238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2685434189119652e-05,
+      "loss": 0.8404,
+      "step": 2239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2679426929886507e-05,
+      "loss": 0.8573,
+      "step": 2240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673418628755078e-05,
+      "loss": 0.832,
+      "step": 2241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26674092880617e-05,
+      "loss": 0.8835,
+      "step": 2242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2661398910143112e-05,
+      "loss": 0.8787,
+      "step": 2243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2655387497336456e-05,
+      "loss": 0.8682,
+      "step": 2244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2649375051979276e-05,
+      "loss": 0.8633,
+      "step": 2245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2643361576409517e-05,
+      "loss": 0.8506,
+      "step": 2246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263734707296553e-05,
+      "loss": 0.8998,
+      "step": 2247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263133154398606e-05,
+      "loss": 0.8286,
+      "step": 2248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2625314991810254e-05,
+      "loss": 0.9122,
+      "step": 2249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2619297418777651e-05,
+      "loss": 0.8301,
+      "step": 2250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2613278827228197e-05,
+      "loss": 0.8471,
+      "step": 2251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2607259219502222e-05,
+      "loss": 0.7783,
+      "step": 2252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2601238597940462e-05,
+      "loss": 0.8649,
+      "step": 2253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2595216964884043e-05,
+      "loss": 0.7944,
+      "step": 2254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258919432267448e-05,
+      "loss": 0.8303,
+      "step": 2255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2583170673653688e-05,
+      "loss": 0.8215,
+      "step": 2256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2577146020163968e-05,
+      "loss": 0.8633,
+      "step": 2257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2571120364548015e-05,
+      "loss": 0.8402,
+      "step": 2258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2565093709148913e-05,
+      "loss": 0.8752,
+      "step": 2259
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2559066056310132e-05,
+      "loss": 0.7535,
+      "step": 2260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2553037408375532e-05,
+      "loss": 0.8516,
+      "step": 2261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2547007767689361e-05,
+      "loss": 0.8566,
+      "step": 2262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.254097713659625e-05,
+      "loss": 0.9231,
+      "step": 2263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2534945517441217e-05,
+      "loss": 0.7779,
+      "step": 2264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2528912912569666e-05,
+      "loss": 0.8529,
+      "step": 2265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2522879324327382e-05,
+      "loss": 0.8622,
+      "step": 2266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2516844755060533e-05,
+      "loss": 0.8618,
+      "step": 2267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2510809207115666e-05,
+      "loss": 0.8484,
+      "step": 2268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2504772682839713e-05,
+      "loss": 0.3326,
+      "step": 2269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2498735184579982e-05,
+      "loss": 0.9023,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492696714684165e-05,
+      "loss": 0.8484,
+      "step": 2271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486657275500322e-05,
+      "loss": 0.8506,
+      "step": 2272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2480616869376903e-05,
+      "loss": 0.8633,
+      "step": 2273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2474575498662722e-05,
+      "loss": 0.804,
+      "step": 2274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2468533165706971e-05,
+      "loss": 0.787,
+      "step": 2275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2462489872859225e-05,
+      "loss": 0.8429,
+      "step": 2276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2456445622469422e-05,
+      "loss": 0.819,
+      "step": 2277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2450400416887872e-05,
+      "loss": 0.8526,
+      "step": 2278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2444354258465268e-05,
+      "loss": 0.8139,
+      "step": 2279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2438307149552659e-05,
+      "loss": 0.8387,
+      "step": 2280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432259092501474e-05,
+      "loss": 0.8171,
+      "step": 2281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2426210089663507e-05,
+      "loss": 0.8346,
+      "step": 2282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2420160143390918e-05,
+      "loss": 0.2954,
+      "step": 2283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2414109256036237e-05,
+      "loss": 0.8484,
+      "step": 2284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240805742995236e-05,
+      "loss": 0.8754,
+      "step": 2285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240200466749254e-05,
+      "loss": 0.8503,
+      "step": 2286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2395950971010408e-05,
+      "loss": 0.8783,
+      "step": 2287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238989634285995e-05,
+      "loss": 0.8251,
+      "step": 2288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2383840785395511e-05,
+      "loss": 0.8798,
+      "step": 2289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2377784300971807e-05,
+      "loss": 0.9033,
+      "step": 2290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2371726891943905e-05,
+      "loss": 0.8212,
+      "step": 2291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2365668560667237e-05,
+      "loss": 0.9113,
+      "step": 2292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2359609309497591e-05,
+      "loss": 0.8514,
+      "step": 2293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2353549140791116e-05,
+      "loss": 0.8735,
+      "step": 2294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2347488056904316e-05,
+      "loss": 0.8397,
+      "step": 2295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2341426060194049e-05,
+      "loss": 0.8712,
+      "step": 2296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335363153017528e-05,
+      "loss": 0.82,
+      "step": 2297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232929933773232e-05,
+      "loss": 0.8292,
+      "step": 2298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2323234616696353e-05,
+      "loss": 0.7791,
+      "step": 2299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2317168992267897e-05,
+      "loss": 0.8814,
+      "step": 2300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2311102466805578e-05,
+      "loss": 0.88,
+      "step": 2301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2305035042668372e-05,
+      "loss": 0.7877,
+      "step": 2302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2298966722215604e-05,
+      "loss": 0.8222,
+      "step": 2303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2292897507806942e-05,
+      "loss": 0.8617,
+      "step": 2304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2286827401802418e-05,
+      "loss": 0.8271,
+      "step": 2305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2280756406562388e-05,
+      "loss": 0.7635,
+      "step": 2306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2274684524447575e-05,
+      "loss": 0.9002,
+      "step": 2307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2268611757819034e-05,
+      "loss": 0.864,
+      "step": 2308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2262538109038162e-05,
+      "loss": 0.8277,
+      "step": 2309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2256463580466713e-05,
+      "loss": 0.8615,
+      "step": 2310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225038817446677e-05,
+      "loss": 0.8931,
+      "step": 2311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2244311893400761e-05,
+      "loss": 0.8876,
+      "step": 2312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223823473963146e-05,
+      "loss": 0.843,
+      "step": 2313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223215671552197e-05,
+      "loss": 0.9115,
+      "step": 2314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2226077823435738e-05,
+      "loss": 0.8264,
+      "step": 2315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2219998065736554e-05,
+      "loss": 0.7367,
+      "step": 2316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2213917444788532e-05,
+      "loss": 0.8051,
+      "step": 2317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220783596295613e-05,
+      "loss": 0.8788,
+      "step": 2318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2201753622604145e-05,
+      "loss": 0.8972,
+      "step": 2319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2195670426097692e-05,
+      "loss": 0.7847,
+      "step": 2320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218958637580224e-05,
+      "loss": 0.8959,
+      "step": 2321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218350147408357e-05,
+      "loss": 0.8796,
+      "step": 2322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2177415723307808e-05,
+      "loss": 0.8333,
+      "step": 2323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2171329125841404e-05,
+      "loss": 0.8468,
+      "step": 2324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.216524168405114e-05,
+      "loss": 0.8889,
+      "step": 2325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2159153400304124e-05,
+      "loss": 0.8753,
+      "step": 2326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215306427696779e-05,
+      "loss": 0.8576,
+      "step": 2327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2146974316409902e-05,
+      "loss": 0.863,
+      "step": 2328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2140883520998552e-05,
+      "loss": 0.8701,
+      "step": 2329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213479189310215e-05,
+      "loss": 0.7653,
+      "step": 2330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2128699435089429e-05,
+      "loss": 0.7794,
+      "step": 2331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2122606149329452e-05,
+      "loss": 0.8276,
+      "step": 2332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2116512038191599e-05,
+      "loss": 0.8362,
+      "step": 2333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2110417104045575e-05,
+      "loss": 0.3608,
+      "step": 2334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2104321349261401e-05,
+      "loss": 0.8457,
+      "step": 2335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2098224776209417e-05,
+      "loss": 0.8865,
+      "step": 2336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2092127387260283e-05,
+      "loss": 0.8187,
+      "step": 2337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2086029184784978e-05,
+      "loss": 0.8904,
+      "step": 2338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2079930171154792e-05,
+      "loss": 0.7816,
+      "step": 2339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2073830348741335e-05,
+      "loss": 0.8383,
+      "step": 2340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206772971991653e-05,
+      "loss": 0.8687,
+      "step": 2341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2061628287052611e-05,
+      "loss": 0.8681,
+      "step": 2342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2055526052522132e-05,
+      "loss": 0.8784,
+      "step": 2343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204942301869795e-05,
+      "loss": 0.9069,
+      "step": 2344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2043319187953242e-05,
+      "loss": 0.8529,
+      "step": 2345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2037214562661483e-05,
+      "loss": 0.8118,
+      "step": 2346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2031109145196469e-05,
+      "loss": 0.8756,
+      "step": 2347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2025002937932298e-05,
+      "loss": 0.8862,
+      "step": 2348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018895943243372e-05,
+      "loss": 0.3473,
+      "step": 2349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2012788163504405e-05,
+      "loss": 0.8781,
+      "step": 2350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2006679601090415e-05,
+      "loss": 0.8721,
+      "step": 2351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2000570258376724e-05,
+      "loss": 0.7751,
+      "step": 2352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994460137738954e-05,
+      "loss": 0.8971,
+      "step": 2353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1988349241553038e-05,
+      "loss": 0.9059,
+      "step": 2354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1982237572195203e-05,
+      "loss": 0.8441,
+      "step": 2355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1976125132041974e-05,
+      "loss": 0.3253,
+      "step": 2356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1970011923470189e-05,
+      "loss": 0.8185,
+      "step": 2357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196389794885697e-05,
+      "loss": 0.7801,
+      "step": 2358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.195778321057974e-05,
+      "loss": 0.8567,
+      "step": 2359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1951667711016233e-05,
+      "loss": 0.8023,
+      "step": 2360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1945551452544458e-05,
+      "loss": 0.8784,
+      "step": 2361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1939434437542734e-05,
+      "loss": 0.3477,
+      "step": 2362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1933316668389669e-05,
+      "loss": 0.8263,
+      "step": 2363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1927198147464163e-05,
+      "loss": 0.8518,
+      "step": 2364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1921078877145414e-05,
+      "loss": 0.8739,
+      "step": 2365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1914958859812901e-05,
+      "loss": 0.3361,
+      "step": 2366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1908838097846404e-05,
+      "loss": 0.888,
+      "step": 2367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1902716593625991e-05,
+      "loss": 0.8408,
+      "step": 2368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1896594349532014e-05,
+      "loss": 0.8507,
+      "step": 2369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1890471367945108e-05,
+      "loss": 0.8388,
+      "step": 2370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1884347651246214e-05,
+      "loss": 0.8986,
+      "step": 2371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1878223201816534e-05,
+      "loss": 0.8052,
+      "step": 2372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1872098022037575e-05,
+      "loss": 0.3527,
+      "step": 2373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1865972114291116e-05,
+      "loss": 0.8909,
+      "step": 2374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1859845480959229e-05,
+      "loss": 0.8331,
+      "step": 2375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1853718124424258e-05,
+      "loss": 0.8473,
+      "step": 2376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1847590047068834e-05,
+      "loss": 0.8404,
+      "step": 2377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1841461251275868e-05,
+      "loss": 0.867,
+      "step": 2378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1835331739428545e-05,
+      "loss": 0.8131,
+      "step": 2379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1829201513910339e-05,
+      "loss": 0.7862,
+      "step": 2380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1823070577104989e-05,
+      "loss": 0.7682,
+      "step": 2381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1816938931396525e-05,
+      "loss": 0.7892,
+      "step": 2382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1810806579169235e-05,
+      "loss": 0.7732,
+      "step": 2383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1804673522807699e-05,
+      "loss": 0.8361,
+      "step": 2384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1798539764696756e-05,
+      "loss": 0.8417,
+      "step": 2385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1792405307221535e-05,
+      "loss": 0.8244,
+      "step": 2386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178627015276742e-05,
+      "loss": 0.8666,
+      "step": 2387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1780134303720074e-05,
+      "loss": 0.8278,
+      "step": 2388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177399776246543e-05,
+      "loss": 0.8535,
+      "step": 2389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1767860531389686e-05,
+      "loss": 0.8466,
+      "step": 2390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1761722612879317e-05,
+      "loss": 0.8881,
+      "step": 2391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1755584009321055e-05,
+      "loss": 0.3435,
+      "step": 2392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1749444723101907e-05,
+      "loss": 0.9213,
+      "step": 2393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1743304756609133e-05,
+      "loss": 0.8597,
+      "step": 2394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173716411223028e-05,
+      "loss": 0.8353,
+      "step": 2395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173102279235313e-05,
+      "loss": 0.8252,
+      "step": 2396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1724880799365757e-05,
+      "loss": 0.8843,
+      "step": 2397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1718738135656471e-05,
+      "loss": 0.3367,
+      "step": 2398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171259480361386e-05,
+      "loss": 0.3537,
+      "step": 2399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706450805626762e-05,
+      "loss": 0.7682,
+      "step": 2400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1700306144084279e-05,
+      "loss": 0.8849,
+      "step": 2401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1694160821375768e-05,
+      "loss": 0.786,
+      "step": 2402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168801483989085e-05,
+      "loss": 0.8328,
+      "step": 2403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1681868202019397e-05,
+      "loss": 0.8212,
+      "step": 2404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1675720910151529e-05,
+      "loss": 0.8446,
+      "step": 2405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1669572966677634e-05,
+      "loss": 0.8177,
+      "step": 2406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663424373988341e-05,
+      "loss": 0.8046,
+      "step": 2407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1657275134474547e-05,
+      "loss": 0.8359,
+      "step": 2408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1651125250527385e-05,
+      "loss": 0.8089,
+      "step": 2409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1644974724538243e-05,
+      "loss": 0.8254,
+      "step": 2410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1638823558898762e-05,
+      "loss": 0.8599,
+      "step": 2411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1632671756000831e-05,
+      "loss": 0.8458,
+      "step": 2412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1626519318236582e-05,
+      "loss": 0.8782,
+      "step": 2413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1620366247998404e-05,
+      "loss": 0.8512,
+      "step": 2414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1614212547678923e-05,
+      "loss": 0.8874,
+      "step": 2415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1608058219671005e-05,
+      "loss": 0.8703,
+      "step": 2416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160190326636778e-05,
+      "loss": 0.8253,
+      "step": 2417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1595747690162601e-05,
+      "loss": 0.8929,
+      "step": 2418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1589591493449073e-05,
+      "loss": 0.845,
+      "step": 2419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158343467862104e-05,
+      "loss": 0.7885,
+      "step": 2420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577277248072586e-05,
+      "loss": 0.9013,
+      "step": 2421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1571119204198038e-05,
+      "loss": 0.7435,
+      "step": 2422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1564960549391957e-05,
+      "loss": 0.8742,
+      "step": 2423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155880128604914e-05,
+      "loss": 0.8571,
+      "step": 2424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1552641416564626e-05,
+      "loss": 0.9127,
+      "step": 2425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1546480943333687e-05,
+      "loss": 0.8192,
+      "step": 2426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154031986875183e-05,
+      "loss": 0.8708,
+      "step": 2427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15341581952148e-05,
+      "loss": 0.8574,
+      "step": 2428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1527995925118567e-05,
+      "loss": 0.802,
+      "step": 2429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1521833060859338e-05,
+      "loss": 0.8586,
+      "step": 2430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1515669604833546e-05,
+      "loss": 0.8392,
+      "step": 2431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1509505559437866e-05,
+      "loss": 0.8325,
+      "step": 2432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1503340927069189e-05,
+      "loss": 0.8677,
+      "step": 2433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149717571012464e-05,
+      "loss": 0.7825,
+      "step": 2434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1491009911001572e-05,
+      "loss": 0.8399,
+      "step": 2435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1484843532097558e-05,
+      "loss": 0.8624,
+      "step": 2436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.147867657581041e-05,
+      "loss": 0.8189,
+      "step": 2437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.147250904453815e-05,
+      "loss": 0.7865,
+      "step": 2438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1466340940679033e-05,
+      "loss": 0.8791,
+      "step": 2439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1460172266631532e-05,
+      "loss": 0.8071,
+      "step": 2440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1454003024794344e-05,
+      "loss": 0.8395,
+      "step": 2441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1447833217566383e-05,
+      "loss": 0.8741,
+      "step": 2442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1441662847346794e-05,
+      "loss": 0.8303,
+      "step": 2443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1435491916534919e-05,
+      "loss": 0.8466,
+      "step": 2444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1429320427530343e-05,
+      "loss": 0.841,
+      "step": 2445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 0.81,
+      "step": 2446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1416975784542454e-05,
+      "loss": 0.8624,
+      "step": 2447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1410802635359373e-05,
+      "loss": 0.8266,
+      "step": 2448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404628937584036e-05,
+      "loss": 0.8341,
+      "step": 2449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1398454693617107e-05,
+      "loss": 0.7849,
+      "step": 2450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1392279905859438e-05,
+      "loss": 0.3103,
+      "step": 2451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1386104576712105e-05,
+      "loss": 0.8174,
+      "step": 2452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1379928708576394e-05,
+      "loss": 0.7707,
+      "step": 2453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1373752303853791e-05,
+      "loss": 0.8554,
+      "step": 2454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1367575364946006e-05,
+      "loss": 0.786,
+      "step": 2455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1361397894254949e-05,
+      "loss": 0.8051,
+      "step": 2456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1355219894182733e-05,
+      "loss": 0.8317,
+      "step": 2457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1349041367131677e-05,
+      "loss": 0.8436,
+      "step": 2458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1342862315504317e-05,
+      "loss": 0.8486,
+      "step": 2459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133668274170338e-05,
+      "loss": 0.8298,
+      "step": 2460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13305026481318e-05,
+      "loss": 0.8439,
+      "step": 2461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1324322037192716e-05,
+      "loss": 0.8641,
+      "step": 2462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1318140911289465e-05,
+      "loss": 0.8574,
+      "step": 2463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1311959272825586e-05,
+      "loss": 0.8798,
+      "step": 2464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1305777124204812e-05,
+      "loss": 0.891,
+      "step": 2465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1299594467831079e-05,
+      "loss": 0.8067,
+      "step": 2466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1293411306108526e-05,
+      "loss": 0.834,
+      "step": 2467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1287227641441478e-05,
+      "loss": 0.8061,
+      "step": 2468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1281043476234461e-05,
+      "loss": 0.7527,
+      "step": 2469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1274858812892198e-05,
+      "loss": 0.8826,
+      "step": 2470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1268673653819594e-05,
+      "loss": 0.8544,
+      "step": 2471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1262488001421765e-05,
+      "loss": 0.8476,
+      "step": 2472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1256301858104007e-05,
+      "loss": 0.7973,
+      "step": 2473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1250115226271804e-05,
+      "loss": 0.8232,
+      "step": 2474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.124392810833084e-05,
+      "loss": 0.8167,
+      "step": 2475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1237740506686983e-05,
+      "loss": 0.8674,
+      "step": 2476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1231552423746284e-05,
+      "loss": 0.8359,
+      "step": 2477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.122536386191499e-05,
+      "loss": 0.8136,
+      "step": 2478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1219174823599532e-05,
+      "loss": 0.8083,
+      "step": 2479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1212985311206522e-05,
+      "loss": 0.8371,
+      "step": 2480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120679532714276e-05,
+      "loss": 0.7736,
+      "step": 2481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1200604873815226e-05,
+      "loss": 0.8216,
+      "step": 2482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119441395363109e-05,
+      "loss": 0.7933,
+      "step": 2483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1188222568997697e-05,
+      "loss": 0.8262,
+      "step": 2484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1182030722322571e-05,
+      "loss": 0.8344,
+      "step": 2485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1175838416013425e-05,
+      "loss": 0.7809,
+      "step": 2486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1169645652478138e-05,
+      "loss": 0.7468,
+      "step": 2487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163452434124773e-05,
+      "loss": 0.7857,
+      "step": 2488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1157258763361577e-05,
+      "loss": 0.8282,
+      "step": 2489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1151064642596958e-05,
+      "loss": 0.882,
+      "step": 2490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1144870074239514e-05,
+      "loss": 0.8662,
+      "step": 2491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1138675060698007e-05,
+      "loss": 0.8643,
+      "step": 2492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1132479604381377e-05,
+      "loss": 0.8967,
+      "step": 2493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.112628370769873e-05,
+      "loss": 0.8342,
+      "step": 2494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1120087373059351e-05,
+      "loss": 0.8638,
+      "step": 2495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1113890602872695e-05,
+      "loss": 0.8358,
+      "step": 2496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1107693399548378e-05,
+      "loss": 0.8132,
+      "step": 2497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.110149576549619e-05,
+      "loss": 0.822,
+      "step": 2498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1095297703126094e-05,
+      "loss": 0.8517,
+      "step": 2499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1089099214848202e-05,
+      "loss": 0.8551,
+      "step": 2500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1082900303072814e-05,
+      "loss": 0.9041,
+      "step": 2501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1076700970210383e-05,
+      "loss": 0.8402,
+      "step": 2502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1070501218671526e-05,
+      "loss": 0.7989,
+      "step": 2503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1064301050867022e-05,
+      "loss": 0.8339,
+      "step": 2504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1058100469207811e-05,
+      "loss": 0.8041,
+      "step": 2505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1051899476105001e-05,
+      "loss": 0.8501,
+      "step": 2506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1045698073969855e-05,
+      "loss": 0.8872,
+      "step": 2507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1039496265213792e-05,
+      "loss": 0.8724,
+      "step": 2508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1033294052248396e-05,
+      "loss": 0.8402,
+      "step": 2509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1027091437485404e-05,
+      "loss": 0.7996,
+      "step": 2510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1020888423336706e-05,
+      "loss": 0.8167,
+      "step": 2511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1014685012214356e-05,
+      "loss": 0.8304,
+      "step": 2512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1008481206530559e-05,
+      "loss": 0.8566,
+      "step": 2513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1002277008697668e-05,
+      "loss": 0.7913,
+      "step": 2514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0996072421128196e-05,
+      "loss": 0.8163,
+      "step": 2515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0989867446234801e-05,
+      "loss": 0.8157,
+      "step": 2516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0983662086430294e-05,
+      "loss": 0.8378,
+      "step": 2517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0977456344127639e-05,
+      "loss": 0.3465,
+      "step": 2518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0971250221739947e-05,
+      "loss": 0.8235,
+      "step": 2519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0965043721680472e-05,
+      "loss": 0.876,
+      "step": 2520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958836846362622e-05,
+      "loss": 0.8673,
+      "step": 2521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0952629598199946e-05,
+      "loss": 0.773,
+      "step": 2522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0946421979606141e-05,
+      "loss": 0.8466,
+      "step": 2523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0940213992995048e-05,
+      "loss": 0.8494,
+      "step": 2524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934005640780648e-05,
+      "loss": 0.8309,
+      "step": 2525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0927796925377064e-05,
+      "loss": 0.8806,
+      "step": 2526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0921587849198565e-05,
+      "loss": 0.853,
+      "step": 2527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0915378414659557e-05,
+      "loss": 0.8401,
+      "step": 2528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0909168624174584e-05,
+      "loss": 0.8529,
+      "step": 2529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0902958480158335e-05,
+      "loss": 0.3697,
+      "step": 2530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0896747985025626e-05,
+      "loss": 0.8524,
+      "step": 2531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0890537141191417e-05,
+      "loss": 0.8674,
+      "step": 2532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0884325951070804e-05,
+      "loss": 0.8031,
+      "step": 2533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0878114417079013e-05,
+      "loss": 0.831,
+      "step": 2534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0871902541631409e-05,
+      "loss": 0.797,
+      "step": 2535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0865690327143488e-05,
+      "loss": 0.8496,
+      "step": 2536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0859477776030872e-05,
+      "loss": 0.8558,
+      "step": 2537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085326489070932e-05,
+      "loss": 0.7918,
+      "step": 2538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0847051673594724e-05,
+      "loss": 0.3656,
+      "step": 2539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0840838127103093e-05,
+      "loss": 0.8982,
+      "step": 2540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0834624253650579e-05,
+      "loss": 0.8446,
+      "step": 2541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0828410055653451e-05,
+      "loss": 0.3332,
+      "step": 2542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0822195535528106e-05,
+      "loss": 0.8997,
+      "step": 2543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0815980695691068e-05,
+      "loss": 0.8078,
+      "step": 2544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080976553855899e-05,
+      "loss": 0.8773,
+      "step": 2545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0803550066548634e-05,
+      "loss": 0.8408,
+      "step": 2546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0797334282076898e-05,
+      "loss": 0.8967,
+      "step": 2547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0791118187560801e-05,
+      "loss": 0.8065,
+      "step": 2548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0784901785417474e-05,
+      "loss": 0.8212,
+      "step": 2549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0778685078064172e-05,
+      "loss": 0.8193,
+      "step": 2550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.077246806791827e-05,
+      "loss": 0.828,
+      "step": 2551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0766250757397256e-05,
+      "loss": 0.7797,
+      "step": 2552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760033148918747e-05,
+      "loss": 0.8188,
+      "step": 2553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0753815244900459e-05,
+      "loss": 0.8615,
+      "step": 2554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0747597047760236e-05,
+      "loss": 0.8306,
+      "step": 2555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0741378559916029e-05,
+      "loss": 0.8856,
+      "step": 2556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0735159783785905e-05,
+      "loss": 0.7826,
+      "step": 2557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0728940721788041e-05,
+      "loss": 0.9048,
+      "step": 2558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.072272137634073e-05,
+      "loss": 0.3125,
+      "step": 2559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0716501749862369e-05,
+      "loss": 0.8522,
+      "step": 2560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0710281844771469e-05,
+      "loss": 0.9117,
+      "step": 2561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0704061663486645e-05,
+      "loss": 0.8266,
+      "step": 2562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0697841208426617e-05,
+      "loss": 0.8381,
+      "step": 2563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0691620482010227e-05,
+      "loss": 0.8411,
+      "step": 2564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0685399486656407e-05,
+      "loss": 0.8334,
+      "step": 2565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0679178224784197e-05,
+      "loss": 0.8593,
+      "step": 2566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0672956698812741e-05,
+      "loss": 0.8446,
+      "step": 2567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0666734911161292e-05,
+      "loss": 0.8068,
+      "step": 2568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066051286424919e-05,
+      "loss": 0.8037,
+      "step": 2569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0654290560495897e-05,
+      "loss": 0.8807,
+      "step": 2570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0648068002320953e-05,
+      "loss": 0.8357,
+      "step": 2571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0641845192144013e-05,
+      "loss": 0.8717,
+      "step": 2572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0635622132384821e-05,
+      "loss": 0.8069,
+      "step": 2573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0629398825463223e-05,
+      "loss": 0.786,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0623175273799158e-05,
+      "loss": 0.839,
+      "step": 2575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0616951479812658e-05,
+      "loss": 0.8697,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0610727445923858e-05,
+      "loss": 0.8605,
+      "step": 2577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0604503174552982e-05,
+      "loss": 0.7773,
+      "step": 2578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059827866812034e-05,
+      "loss": 0.8669,
+      "step": 2579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059205392904634e-05,
+      "loss": 0.3101,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0585828959751478e-05,
+      "loss": 0.812,
+      "step": 2581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0579603762656347e-05,
+      "loss": 0.8024,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0573378340181614e-05,
+      "loss": 0.8318,
+      "step": 2583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0567152694748046e-05,
+      "loss": 0.8304,
+      "step": 2584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056092682877649e-05,
+      "loss": 0.8346,
+      "step": 2585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0554700744687885e-05,
+      "loss": 0.8442,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0548474444903247e-05,
+      "loss": 0.8133,
+      "step": 2587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0542247931843684e-05,
+      "loss": 0.8189,
+      "step": 2588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0536021207930381e-05,
+      "loss": 0.3073,
+      "step": 2589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0529794275584606e-05,
+      "loss": 0.8065,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0523567137227708e-05,
+      "loss": 0.8037,
+      "step": 2591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0517339795281118e-05,
+      "loss": 0.7755,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051111225216635e-05,
+      "loss": 0.8337,
+      "step": 2593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0504884510304982e-05,
+      "loss": 0.8123,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0498656572118687e-05,
+      "loss": 0.7894,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04924284400292e-05,
+      "loss": 0.8972,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0486200116458343e-05,
+      "loss": 0.8154,
+      "step": 2597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0479971603828001e-05,
+      "loss": 0.8506,
+      "step": 2598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0473742904560143e-05,
+      "loss": 0.8513,
+      "step": 2599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0467514021076805e-05,
+      "loss": 0.796,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0461284955800094e-05,
+      "loss": 0.8715,
+      "step": 2601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0455055711152194e-05,
+      "loss": 0.7965,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0448826289555345e-05,
+      "loss": 0.8258,
+      "step": 2603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0442596693431872e-05,
+      "loss": 0.8785,
+      "step": 2604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043636692520416e-05,
+      "loss": 0.7619,
+      "step": 2605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0430136987294661e-05,
+      "loss": 0.8579,
+      "step": 2606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042390688212589e-05,
+      "loss": 0.7689,
+      "step": 2607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0417676612120438e-05,
+      "loss": 0.8618,
+      "step": 2608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0411446179700944e-05,
+      "loss": 0.8575,
+      "step": 2609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0405215587290126e-05,
+      "loss": 0.8348,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0398984837310755e-05,
+      "loss": 0.8468,
+      "step": 2611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0392753932185667e-05,
+      "loss": 0.849,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0386522874337752e-05,
+      "loss": 0.8061,
+      "step": 2613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0380291666189968e-05,
+      "loss": 0.8905,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0374060310165324e-05,
+      "loss": 0.9047,
+      "step": 2615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0367828808686892e-05,
+      "loss": 0.8533,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0361597164177802e-05,
+      "loss": 0.9113,
+      "step": 2617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.035536537906123e-05,
+      "loss": 0.8232,
+      "step": 2618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034913345576042e-05,
+      "loss": 0.8616,
+      "step": 2619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0342901396698658e-05,
+      "loss": 0.8981,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0336669204299286e-05,
+      "loss": 0.8679,
+      "step": 2621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0330436880985704e-05,
+      "loss": 0.8513,
+      "step": 2622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0324204429181355e-05,
+      "loss": 0.8322,
+      "step": 2623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0317971851309736e-05,
+      "loss": 0.7858,
+      "step": 2624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0311739149794394e-05,
+      "loss": 0.8831,
+      "step": 2625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0305506327058916e-05,
+      "loss": 0.8588,
+      "step": 2626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0299273385526949e-05,
+      "loss": 0.3416,
+      "step": 2627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0293040327622175e-05,
+      "loss": 0.864,
+      "step": 2628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0286807155768331e-05,
+      "loss": 0.8183,
+      "step": 2629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0280573872389187e-05,
+      "loss": 0.8239,
+      "step": 2630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0274340479908568e-05,
+      "loss": 0.8773,
+      "step": 2631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0268106980750334e-05,
+      "loss": 0.8016,
+      "step": 2632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0261873377338387e-05,
+      "loss": 0.8153,
+      "step": 2633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0255639672096673e-05,
+      "loss": 0.909,
+      "step": 2634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0249405867449176e-05,
+      "loss": 0.8391,
+      "step": 2635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0243171965819914e-05,
+      "loss": 0.7664,
+      "step": 2636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0236937969632951e-05,
+      "loss": 0.8374,
+      "step": 2637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023070388131238e-05,
+      "loss": 0.8487,
+      "step": 2638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224469703282336e-05,
+      "loss": 0.3256,
+      "step": 2639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021823543796699e-05,
+      "loss": 0.8924,
+      "step": 2640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0212001087790538e-05,
+      "loss": 0.8896,
+      "step": 2641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0205766655177217e-05,
+      "loss": 0.8801,
+      "step": 2642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019953214255129e-05,
+      "loss": 0.8571,
+      "step": 2643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0193297552337055e-05,
+      "loss": 0.8137,
+      "step": 2644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0187062886958845e-05,
+      "loss": 0.8809,
+      "step": 2645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0180828148841013e-05,
+      "loss": 0.835,
+      "step": 2646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0174593340407945e-05,
+      "loss": 0.879,
+      "step": 2647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0168358464084052e-05,
+      "loss": 0.7794,
+      "step": 2648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0162123522293775e-05,
+      "loss": 0.7655,
+      "step": 2649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0155888517461575e-05,
+      "loss": 0.8286,
+      "step": 2650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0149653452011943e-05,
+      "loss": 0.8151,
+      "step": 2651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0143418328369394e-05,
+      "loss": 0.8619,
+      "step": 2652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0137183148958462e-05,
+      "loss": 0.8379,
+      "step": 2653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0130947916203701e-05,
+      "loss": 0.8218,
+      "step": 2654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012471263252969e-05,
+      "loss": 0.8082,
+      "step": 2655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0118477300361027e-05,
+      "loss": 0.786,
+      "step": 2656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0112241922122328e-05,
+      "loss": 0.856,
+      "step": 2657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0106006500238225e-05,
+      "loss": 0.751,
+      "step": 2658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099771037133372e-05,
+      "loss": 0.8616,
+      "step": 2659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0093535535232435e-05,
+      "loss": 0.8721,
+      "step": 2660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087299996960096e-05,
+      "loss": 0.839,
+      "step": 2661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0081064424741052e-05,
+      "loss": 0.893,
+      "step": 2662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0074828821000012e-05,
+      "loss": 0.8877,
+      "step": 2663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0068593188161698e-05,
+      "loss": 0.8828,
+      "step": 2664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062357528650841e-05,
+      "loss": 0.7838,
+      "step": 2665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0056121844892188e-05,
+      "loss": 0.8186,
+      "step": 2666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004988613931049e-05,
+      "loss": 0.8182,
+      "step": 2667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043650414330508e-05,
+      "loss": 0.813,
+      "step": 2668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0037414672377013e-05,
+      "loss": 0.8621,
+      "step": 2669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0031178915874775e-05,
+      "loss": 0.8651,
+      "step": 2670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0024943147248582e-05,
+      "loss": 0.7551,
+      "step": 2671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0018707368923217e-05,
+      "loss": 0.8019,
+      "step": 2672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001247158332347e-05,
+      "loss": 0.7899,
+      "step": 2673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0006235792874133e-05,
+      "loss": 0.8423,
+      "step": 2674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 0.7848,
+      "step": 2675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993764207125868e-06,
+      "loss": 0.359,
+      "step": 2676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987528416676534e-06,
+      "loss": 0.8207,
+      "step": 2677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.981292631076785e-06,
+      "loss": 0.7922,
+      "step": 2678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.975056852751421e-06,
+      "loss": 0.8703,
+      "step": 2679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.968821084125228e-06,
+      "loss": 0.3457,
+      "step": 2680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.962585327622992e-06,
+      "loss": 0.3249,
+      "step": 2681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956349585669495e-06,
+      "loss": 0.8189,
+      "step": 2682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.950113860689512e-06,
+      "loss": 0.876,
+      "step": 2683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943878155107817e-06,
+      "loss": 0.8035,
+      "step": 2684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937642471349162e-06,
+      "loss": 0.7983,
+      "step": 2685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931406811838307e-06,
+      "loss": 0.8274,
+      "step": 2686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925171178999992e-06,
+      "loss": 0.8096,
+      "step": 2687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.918935575258948e-06,
+      "loss": 0.8238,
+      "step": 2688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912700003039907e-06,
+      "loss": 0.3658,
+      "step": 2689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.906464464767567e-06,
+      "loss": 0.8333,
+      "step": 2690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900228962866631e-06,
+      "loss": 0.8385,
+      "step": 2691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.893993499761776e-06,
+      "loss": 0.8569,
+      "step": 2692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887758077877673e-06,
+      "loss": 0.8616,
+      "step": 2693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.881522699638976e-06,
+      "loss": 0.8293,
+      "step": 2694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87528736747031e-06,
+      "loss": 0.8318,
+      "step": 2695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869052083796302e-06,
+      "loss": 0.7585,
+      "step": 2696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862816851041541e-06,
+      "loss": 0.7778,
+      "step": 2697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856581671630609e-06,
+      "loss": 0.851,
+      "step": 2698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850346547988059e-06,
+      "loss": 0.8118,
+      "step": 2699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844111482538427e-06,
+      "loss": 0.8321,
+      "step": 2700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83787647770623e-06,
+      "loss": 0.8189,
+      "step": 2701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.831641535915953e-06,
+      "loss": 0.8761,
+      "step": 2702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82540665959206e-06,
+      "loss": 0.8186,
+      "step": 2703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81917185115899e-06,
+      "loss": 0.849,
+      "step": 2704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.812937113041155e-06,
+      "loss": 0.8192,
+      "step": 2705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806702447662946e-06,
+      "loss": 0.8514,
+      "step": 2706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800467857448713e-06,
+      "loss": 0.8379,
+      "step": 2707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79423334482279e-06,
+      "loss": 0.8185,
+      "step": 2708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.787998912209466e-06,
+      "loss": 0.8345,
+      "step": 2709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.781764562033011e-06,
+      "loss": 0.8248,
+      "step": 2710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775530296717666e-06,
+      "loss": 0.856,
+      "step": 2711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.769296118687623e-06,
+      "loss": 0.7937,
+      "step": 2712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763062030367054e-06,
+      "loss": 0.7515,
+      "step": 2713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.756828034180088e-06,
+      "loss": 0.7554,
+      "step": 2714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75059413255083e-06,
+      "loss": 0.8206,
+      "step": 2715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74436032790333e-06,
+      "loss": 0.8377,
+      "step": 2716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738126622661613e-06,
+      "loss": 0.8474,
+      "step": 2717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731893019249669e-06,
+      "loss": 0.8586,
+      "step": 2718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725659520091433e-06,
+      "loss": 0.7923,
+      "step": 2719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.719426127610818e-06,
+      "loss": 0.8482,
+      "step": 2720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.713192844231674e-06,
+      "loss": 0.8922,
+      "step": 2721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.706959672377826e-06,
+      "loss": 0.8009,
+      "step": 2722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700726614473055e-06,
+      "loss": 0.891,
+      "step": 2723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.694493672941086e-06,
+      "loss": 0.8244,
+      "step": 2724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.688260850205613e-06,
+      "loss": 0.8307,
+      "step": 2725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.682028148690267e-06,
+      "loss": 0.8437,
+      "step": 2726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.675795570818645e-06,
+      "loss": 0.8799,
+      "step": 2727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669563119014299e-06,
+      "loss": 0.8046,
+      "step": 2728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663330795700714e-06,
+      "loss": 0.8569,
+      "step": 2729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.657098603301347e-06,
+      "loss": 0.8335,
+      "step": 2730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.650866544239582e-06,
+      "loss": 0.8276,
+      "step": 2731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644634620938771e-06,
+      "loss": 0.8436,
+      "step": 2732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638402835822203e-06,
+      "loss": 0.9161,
+      "step": 2733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63217119131311e-06,
+      "loss": 0.8381,
+      "step": 2734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62593968983468e-06,
+      "loss": 0.7998,
+      "step": 2735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619708333810035e-06,
+      "loss": 0.8863,
+      "step": 2736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.613477125662255e-06,
+      "loss": 0.8769,
+      "step": 2737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.607246067814338e-06,
+      "loss": 0.8269,
+      "step": 2738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.601015162689245e-06,
+      "loss": 0.8431,
+      "step": 2739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594784412709876e-06,
+      "loss": 0.8215,
+      "step": 2740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588553820299057e-06,
+      "loss": 0.9094,
+      "step": 2741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.582323387879569e-06,
+      "loss": 0.9037,
+      "step": 2742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576093117874112e-06,
+      "loss": 0.7414,
+      "step": 2743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.569863012705345e-06,
+      "loss": 0.7868,
+      "step": 2744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563633074795843e-06,
+      "loss": 0.8672,
+      "step": 2745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557403306568128e-06,
+      "loss": 0.8688,
+      "step": 2746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551173710444658e-06,
+      "loss": 0.8836,
+      "step": 2747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.544944288847811e-06,
+      "loss": 0.7908,
+      "step": 2748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53871504419991e-06,
+      "loss": 0.8878,
+      "step": 2749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.532485978923199e-06,
+      "loss": 0.8115,
+      "step": 2750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.526257095439859e-06,
+      "loss": 0.8587,
+      "step": 2751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520028396172002e-06,
+      "loss": 0.8388,
+      "step": 2752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51379988354166e-06,
+      "loss": 0.7964,
+      "step": 2753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507571559970802e-06,
+      "loss": 0.8062,
+      "step": 2754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.501343427881317e-06,
+      "loss": 0.3467,
+      "step": 2755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49511548969502e-06,
+      "loss": 0.8466,
+      "step": 2756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.488887747833655e-06,
+      "loss": 0.7352,
+      "step": 2757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.482660204718882e-06,
+      "loss": 0.789,
+      "step": 2758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.476432862772297e-06,
+      "loss": 0.8105,
+      "step": 2759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.470205724415397e-06,
+      "loss": 0.8645,
+      "step": 2760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.463978792069624e-06,
+      "loss": 0.9099,
+      "step": 2761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.457752068156318e-06,
+      "loss": 0.8394,
+      "step": 2762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451525555096753e-06,
+      "loss": 0.9045,
+      "step": 2763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.445299255312117e-06,
+      "loss": 0.8899,
+      "step": 2764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.439073171223511e-06,
+      "loss": 0.8302,
+      "step": 2765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.432847305251957e-06,
+      "loss": 0.8528,
+      "step": 2766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.426621659818388e-06,
+      "loss": 0.8203,
+      "step": 2767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.420396237343654e-06,
+      "loss": 0.8702,
+      "step": 2768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.414171040248523e-06,
+      "loss": 0.8046,
+      "step": 2769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407946070953662e-06,
+      "loss": 0.7867,
+      "step": 2770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.401721331879665e-06,
+      "loss": 0.9136,
+      "step": 2771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395496825447022e-06,
+      "loss": 0.3475,
+      "step": 2772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.389272554076142e-06,
+      "loss": 0.7255,
+      "step": 2773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383048520187344e-06,
+      "loss": 0.8533,
+      "step": 2774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.376824726200847e-06,
+      "loss": 0.8061,
+      "step": 2775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37060117453678e-06,
+      "loss": 0.8612,
+      "step": 2776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.36437786761518e-06,
+      "loss": 0.8197,
+      "step": 2777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358154807855992e-06,
+      "loss": 0.8344,
+      "step": 2778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.351931997679049e-06,
+      "loss": 0.816,
+      "step": 2779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.345709439504104e-06,
+      "loss": 0.8389,
+      "step": 2780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.339487135750811e-06,
+      "loss": 0.9226,
+      "step": 2781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.333265088838713e-06,
+      "loss": 0.828,
+      "step": 2782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.327043301187264e-06,
+      "loss": 0.8643,
+      "step": 2783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.320821775215807e-06,
+      "loss": 0.3251,
+      "step": 2784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.314600513343595e-06,
+      "loss": 0.8665,
+      "step": 2785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308379517989775e-06,
+      "loss": 0.8381,
+      "step": 2786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.302158791573384e-06,
+      "loss": 0.8229,
+      "step": 2787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.295938336513362e-06,
+      "loss": 0.8564,
+      "step": 2788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.289718155228535e-06,
+      "loss": 0.7385,
+      "step": 2789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.283498250137636e-06,
+      "loss": 0.8296,
+      "step": 2790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277278623659273e-06,
+      "loss": 0.8928,
+      "step": 2791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.271059278211959e-06,
+      "loss": 0.8228,
+      "step": 2792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.264840216214099e-06,
+      "loss": 0.7886,
+      "step": 2793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.258621440083973e-06,
+      "loss": 0.781,
+      "step": 2794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252402952239767e-06,
+      "loss": 0.8257,
+      "step": 2795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246184755099545e-06,
+      "loss": 0.8191,
+      "step": 2796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.239966851081256e-06,
+      "loss": 0.8775,
+      "step": 2797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.233749242602745e-06,
+      "loss": 0.8257,
+      "step": 2798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.227531932081732e-06,
+      "loss": 0.7561,
+      "step": 2799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.221314921935833e-06,
+      "loss": 0.7845,
+      "step": 2800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21509821458253e-06,
+      "loss": 0.8392,
+      "step": 2801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.208881812439199e-06,
+      "loss": 0.7706,
+      "step": 2802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.202665717923103e-06,
+      "loss": 0.859,
+      "step": 2803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.196449933451366e-06,
+      "loss": 0.7874,
+      "step": 2804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.190234461441015e-06,
+      "loss": 0.3608,
+      "step": 2805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184019304308933e-06,
+      "loss": 0.8235,
+      "step": 2806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.177804464471897e-06,
+      "loss": 0.8425,
+      "step": 2807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171589944346552e-06,
+      "loss": 0.8323,
+      "step": 2808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.165375746349424e-06,
+      "loss": 0.8774,
+      "step": 2809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15916187289691e-06,
+      "loss": 0.7683,
+      "step": 2810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15294832640528e-06,
+      "loss": 0.8312,
+      "step": 2811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.146735109290685e-06,
+      "loss": 0.8316,
+      "step": 2812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.140522223969132e-06,
+      "loss": 0.8242,
+      "step": 2813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134309672856513e-06,
+      "loss": 0.8042,
+      "step": 2814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.128097458368593e-06,
+      "loss": 0.7828,
+      "step": 2815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121885582920987e-06,
+      "loss": 0.9129,
+      "step": 2816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115674048929199e-06,
+      "loss": 0.8093,
+      "step": 2817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.109462858808586e-06,
+      "loss": 0.8701,
+      "step": 2818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10325201497438e-06,
+      "loss": 0.8386,
+      "step": 2819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09704151984167e-06,
+      "loss": 0.846,
+      "step": 2820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090831375825416e-06,
+      "loss": 0.8805,
+      "step": 2821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.084621585340448e-06,
+      "loss": 0.8099,
+      "step": 2822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078412150801438e-06,
+      "loss": 0.8353,
+      "step": 2823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.072203074622941e-06,
+      "loss": 0.8887,
+      "step": 2824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.065994359219357e-06,
+      "loss": 0.8484,
+      "step": 2825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.059786007004953e-06,
+      "loss": 0.8215,
+      "step": 2826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.053578020393862e-06,
+      "loss": 0.7967,
+      "step": 2827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.047370401800055e-06,
+      "loss": 0.8233,
+      "step": 2828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041163153637382e-06,
+      "loss": 0.7914,
+      "step": 2829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03495627831953e-06,
+      "loss": 0.3165,
+      "step": 2830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.028749778260055e-06,
+      "loss": 0.8626,
+      "step": 2831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022543655872363e-06,
+      "loss": 0.7963,
+      "step": 2832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016337913569708e-06,
+      "loss": 0.8324,
+      "step": 2833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.010132553765204e-06,
+      "loss": 0.817,
+      "step": 2834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.003927578871807e-06,
+      "loss": 0.7857,
+      "step": 2835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.997722991302335e-06,
+      "loss": 0.8576,
+      "step": 2836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.991518793469444e-06,
+      "loss": 0.8662,
+      "step": 2837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.985314987785644e-06,
+      "loss": 0.8295,
+      "step": 2838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.979111576663295e-06,
+      "loss": 0.8195,
+      "step": 2839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9729085625146e-06,
+      "loss": 0.8377,
+      "step": 2840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.966705947751608e-06,
+      "loss": 0.7872,
+      "step": 2841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96050373478621e-06,
+      "loss": 0.8303,
+      "step": 2842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954301926030146e-06,
+      "loss": 0.8381,
+      "step": 2843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.948100523895002e-06,
+      "loss": 0.9378,
+      "step": 2844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94189953079219e-06,
+      "loss": 0.7629,
+      "step": 2845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.935698949132983e-06,
+      "loss": 0.8433,
+      "step": 2846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.929498781328477e-06,
+      "loss": 0.8931,
+      "step": 2847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923299029789617e-06,
+      "loss": 0.7925,
+      "step": 2848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917099696927188e-06,
+      "loss": 0.7517,
+      "step": 2849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9109007851518e-06,
+      "loss": 0.3428,
+      "step": 2850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.904702296873913e-06,
+      "loss": 0.8521,
+      "step": 2851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.898504234503811e-06,
+      "loss": 0.7893,
+      "step": 2852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.892306600451628e-06,
+      "loss": 0.8496,
+      "step": 2853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886109397127309e-06,
+      "loss": 0.8235,
+      "step": 2854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.879912626940647e-06,
+      "loss": 0.826,
+      "step": 2855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.873716292301273e-06,
+      "loss": 0.841,
+      "step": 2856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867520395618626e-06,
+      "loss": 0.8177,
+      "step": 2857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.861324939301998e-06,
+      "loss": 0.7627,
+      "step": 2858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.855129925760488e-06,
+      "loss": 0.8125,
+      "step": 2859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848935357403041e-06,
+      "loss": 0.8255,
+      "step": 2860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.842741236638427e-06,
+      "loss": 0.8312,
+      "step": 2861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836547565875227e-06,
+      "loss": 0.8403,
+      "step": 2862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.830354347521867e-06,
+      "loss": 0.8445,
+      "step": 2863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.824161583986578e-06,
+      "loss": 0.8136,
+      "step": 2864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817969277677434e-06,
+      "loss": 0.8312,
+      "step": 2865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811777431002307e-06,
+      "loss": 0.8131,
+      "step": 2866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80558604636891e-06,
+      "loss": 0.8194,
+      "step": 2867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.799395126184777e-06,
+      "loss": 0.8586,
+      "step": 2868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.793204672857244e-06,
+      "loss": 0.8552,
+      "step": 2869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.787014688793483e-06,
+      "loss": 0.8415,
+      "step": 2870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78082517640047e-06,
+      "loss": 0.8259,
+      "step": 2871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.774636138085013e-06,
+      "loss": 0.8843,
+      "step": 2872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76844757625372e-06,
+      "loss": 0.8758,
+      "step": 2873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76225949331302e-06,
+      "loss": 0.8115,
+      "step": 2874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.756071891669164e-06,
+      "loss": 0.8524,
+      "step": 2875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.749884773728197e-06,
+      "loss": 0.7616,
+      "step": 2876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.743698141895995e-06,
+      "loss": 0.7483,
+      "step": 2877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.737511998578237e-06,
+      "loss": 0.8374,
+      "step": 2878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.731326346180406e-06,
+      "loss": 0.8412,
+      "step": 2879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.725141187107809e-06,
+      "loss": 0.7939,
+      "step": 2880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718956523765542e-06,
+      "loss": 0.833,
+      "step": 2881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.712772358558526e-06,
+      "loss": 0.8229,
+      "step": 2882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.706588693891475e-06,
+      "loss": 0.7962,
+      "step": 2883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.700405532168921e-06,
+      "loss": 0.7755,
+      "step": 2884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.694222875795193e-06,
+      "loss": 0.82,
+      "step": 2885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.688040727174419e-06,
+      "loss": 0.8261,
+      "step": 2886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.681859088710538e-06,
+      "loss": 0.8628,
+      "step": 2887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675677962807285e-06,
+      "loss": 0.7599,
+      "step": 2888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669497351868199e-06,
+      "loss": 0.8454,
+      "step": 2889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663317258296624e-06,
+      "loss": 0.8831,
+      "step": 2890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.657137684495685e-06,
+      "loss": 0.8664,
+      "step": 2891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.650958632868325e-06,
+      "loss": 0.8991,
+      "step": 2892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.644780105817272e-06,
+      "loss": 0.8128,
+      "step": 2893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.638602105745058e-06,
+      "loss": 0.7515,
+      "step": 2894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.632424635053997e-06,
+      "loss": 0.8416,
+      "step": 2895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62624769614621e-06,
+      "loss": 0.8838,
+      "step": 2896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.620071291423613e-06,
+      "loss": 0.829,
+      "step": 2897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613895423287899e-06,
+      "loss": 0.849,
+      "step": 2898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607720094140567e-06,
+      "loss": 0.8508,
+      "step": 2899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.601545306382897e-06,
+      "loss": 0.8059,
+      "step": 2900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.595371062415962e-06,
+      "loss": 0.8338,
+      "step": 2901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58919736464063e-06,
+      "loss": 0.7592,
+      "step": 2902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.583024215457548e-06,
+      "loss": 0.7466,
+      "step": 2903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 0.8212,
+      "step": 2904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570679572469659e-06,
+      "loss": 0.8334,
+      "step": 2905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56450808346508e-06,
+      "loss": 0.8476,
+      "step": 2906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.558337152653211e-06,
+      "loss": 0.8403,
+      "step": 2907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.552166782433617e-06,
+      "loss": 0.806,
+      "step": 2908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54599697520566e-06,
+      "loss": 0.8261,
+      "step": 2909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53982773336847e-06,
+      "loss": 0.8445,
+      "step": 2910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53365905932097e-06,
+      "loss": 0.8148,
+      "step": 2911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.527490955461852e-06,
+      "loss": 0.3295,
+      "step": 2912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.521323424189593e-06,
+      "loss": 0.8674,
+      "step": 2913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.515156467902444e-06,
+      "loss": 0.8197,
+      "step": 2914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.508990088998431e-06,
+      "loss": 0.8192,
+      "step": 2915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502824289875366e-06,
+      "loss": 0.8313,
+      "step": 2916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496659072930815e-06,
+      "loss": 0.8063,
+      "step": 2917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.490494440562136e-06,
+      "loss": 0.8544,
+      "step": 2918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.484330395166455e-06,
+      "loss": 0.8329,
+      "step": 2919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.478166939140665e-06,
+      "loss": 0.8175,
+      "step": 2920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472004074881438e-06,
+      "loss": 0.8543,
+      "step": 2921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.465841804785202e-06,
+      "loss": 0.8026,
+      "step": 2922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.459680131248168e-06,
+      "loss": 0.8426,
+      "step": 2923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.453519056666314e-06,
+      "loss": 0.8294,
+      "step": 2924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.447358583435377e-06,
+      "loss": 0.7982,
+      "step": 2925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.441198713950865e-06,
+      "loss": 0.3814,
+      "step": 2926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.435039450608046e-06,
+      "loss": 0.8111,
+      "step": 2927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.428880795801965e-06,
+      "loss": 0.859,
+      "step": 2928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.422722751927415e-06,
+      "loss": 0.8077,
+      "step": 2929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.416565321378962e-06,
+      "loss": 0.8727,
+      "step": 2930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41040850655093e-06,
+      "loss": 0.8568,
+      "step": 2931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4042523098374e-06,
+      "loss": 0.7196,
+      "step": 2932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.398096733632223e-06,
+      "loss": 0.3266,
+      "step": 2933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.391941780328996e-06,
+      "loss": 0.8187,
+      "step": 2934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38578745232108e-06,
+      "loss": 0.7245,
+      "step": 2935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.379633752001597e-06,
+      "loss": 0.843,
+      "step": 2936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.373480681763418e-06,
+      "loss": 0.787,
+      "step": 2937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.367328243999174e-06,
+      "loss": 0.8483,
+      "step": 2938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36117644110124e-06,
+      "loss": 0.7993,
+      "step": 2939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355025275461764e-06,
+      "loss": 0.7983,
+      "step": 2940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.34887474947262e-06,
+      "loss": 0.3435,
+      "step": 2941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.342724865525455e-06,
+      "loss": 0.8608,
+      "step": 2942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33657562601166e-06,
+      "loss": 0.809,
+      "step": 2943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33042703332237e-06,
+      "loss": 0.8481,
+      "step": 2944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.324279089848475e-06,
+      "loss": 0.8476,
+      "step": 2945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.318131797980606e-06,
+      "loss": 0.7972,
+      "step": 2946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.311985160109148e-06,
+      "loss": 0.3796,
+      "step": 2947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.305839178624233e-06,
+      "loss": 0.7573,
+      "step": 2948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299693855915723e-06,
+      "loss": 0.8264,
+      "step": 2949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.293549194373243e-06,
+      "loss": 0.8099,
+      "step": 2950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.287405196386144e-06,
+      "loss": 0.825,
+      "step": 2951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28126186434353e-06,
+      "loss": 0.8647,
+      "step": 2952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.275119200634248e-06,
+      "loss": 0.8171,
+      "step": 2953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26897720764687e-06,
+      "loss": 0.8852,
+      "step": 2954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262835887769725e-06,
+      "loss": 0.8384,
+      "step": 2955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256695243390868e-06,
+      "loss": 0.314,
+      "step": 2956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2505552768981e-06,
+      "loss": 0.8704,
+      "step": 2957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.244415990678949e-06,
+      "loss": 0.8016,
+      "step": 2958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238277387120685e-06,
+      "loss": 0.3297,
+      "step": 2959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.232139468610316e-06,
+      "loss": 0.8242,
+      "step": 2960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226002237534574e-06,
+      "loss": 0.7336,
+      "step": 2961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21986569627993e-06,
+      "loss": 0.8144,
+      "step": 2962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213729847232583e-06,
+      "loss": 0.8746,
+      "step": 2963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.207594692778464e-06,
+      "loss": 0.787,
+      "step": 2964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.201460235303245e-06,
+      "loss": 0.8067,
+      "step": 2965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.195326477192305e-06,
+      "loss": 0.3419,
+      "step": 2966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.189193420830768e-06,
+      "loss": 0.7755,
+      "step": 2967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18306106860348e-06,
+      "loss": 0.3387,
+      "step": 2968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17692942289501e-06,
+      "loss": 0.7597,
+      "step": 2969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.170798486089666e-06,
+      "loss": 0.8457,
+      "step": 2970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.164668260571457e-06,
+      "loss": 0.8114,
+      "step": 2971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.158538748724139e-06,
+      "loss": 0.8597,
+      "step": 2972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.152409952931168e-06,
+      "loss": 0.7819,
+      "step": 2973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.146281875575748e-06,
+      "loss": 0.7921,
+      "step": 2974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.140154519040775e-06,
+      "loss": 0.3351,
+      "step": 2975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134027885708884e-06,
+      "loss": 0.7969,
+      "step": 2976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.127901977962428e-06,
+      "loss": 0.8237,
+      "step": 2977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121776798183468e-06,
+      "loss": 0.7997,
+      "step": 2978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115652348753791e-06,
+      "loss": 0.8255,
+      "step": 2979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.109528632054893e-06,
+      "loss": 0.7955,
+      "step": 2980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103405650467988e-06,
+      "loss": 0.8312,
+      "step": 2981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.097283406374012e-06,
+      "loss": 0.7733,
+      "step": 2982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.091161902153594e-06,
+      "loss": 0.833,
+      "step": 2983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.085041140187102e-06,
+      "loss": 0.8303,
+      "step": 2984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07892112285459e-06,
+      "loss": 0.7613,
+      "step": 2985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07280185253584e-06,
+      "loss": 0.7335,
+      "step": 2986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.066683331610333e-06,
+      "loss": 0.8216,
+      "step": 2987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.060565562457268e-06,
+      "loss": 0.7779,
+      "step": 2988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.054448547455546e-06,
+      "loss": 0.8496,
+      "step": 2989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04833228898377e-06,
+      "loss": 0.8127,
+      "step": 2990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042216789420263e-06,
+      "loss": 0.8285,
+      "step": 2991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.036102051143036e-06,
+      "loss": 0.8963,
+      "step": 2992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.029988076529813e-06,
+      "loss": 0.8423,
+      "step": 2993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.023874867958027e-06,
+      "loss": 0.7875,
+      "step": 2994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.017762427804799e-06,
+      "loss": 0.8538,
+      "step": 2995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.011650758446967e-06,
+      "loss": 0.7906,
+      "step": 2996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005539862261048e-06,
+      "loss": 0.8751,
+      "step": 2997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999429741623278e-06,
+      "loss": 0.8159,
+      "step": 2998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.993320398909587e-06,
+      "loss": 0.8447,
+      "step": 2999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987211836495597e-06,
+      "loss": 0.8176,
+      "step": 3000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981104056756633e-06,
+      "loss": 0.8323,
+      "step": 3001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.974997062067707e-06,
+      "loss": 0.7937,
+      "step": 3002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.968890854803536e-06,
+      "loss": 0.8209,
+      "step": 3003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96278543733852e-06,
+      "loss": 0.8732,
+      "step": 3004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95668081204676e-06,
+      "loss": 0.8493,
+      "step": 3005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.950576981302053e-06,
+      "loss": 0.89,
+      "step": 3006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.944473947477871e-06,
+      "loss": 0.719,
+      "step": 3007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.938371712947392e-06,
+      "loss": 0.7492,
+      "step": 3008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932270280083473e-06,
+      "loss": 0.8139,
+      "step": 3009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926169651258668e-06,
+      "loss": 0.8503,
+      "step": 3010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.920069828845211e-06,
+      "loss": 0.7922,
+      "step": 3011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.913970815215024e-06,
+      "loss": 0.8375,
+      "step": 3012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90787261273972e-06,
+      "loss": 0.8223,
+      "step": 3013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.901775223790585e-06,
+      "loss": 0.7798,
+      "step": 3014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.895678650738602e-06,
+      "loss": 0.8501,
+      "step": 3015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.889582895954427e-06,
+      "loss": 0.884,
+      "step": 3016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.883487961808401e-06,
+      "loss": 0.8511,
+      "step": 3017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87739385067055e-06,
+      "loss": 0.82,
+      "step": 3018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871300564910576e-06,
+      "loss": 0.8568,
+      "step": 3019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.865208106897856e-06,
+      "loss": 0.8016,
+      "step": 3020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85911647900145e-06,
+      "loss": 0.3327,
+      "step": 3021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.853025683590096e-06,
+      "loss": 0.8189,
+      "step": 3022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.846935723032213e-06,
+      "loss": 0.345,
+      "step": 3023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.840846599695878e-06,
+      "loss": 0.8354,
+      "step": 3024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.834758315948865e-06,
+      "loss": 0.864,
+      "step": 3025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.828670874158598e-06,
+      "loss": 0.7988,
+      "step": 3026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822584276692192e-06,
+      "loss": 0.823,
+      "step": 3027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.816498525916435e-06,
+      "loss": 0.7984,
+      "step": 3028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.810413624197764e-06,
+      "loss": 0.8652,
+      "step": 3029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80432957390231e-06,
+      "loss": 0.7748,
+      "step": 3030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.798246377395858e-06,
+      "loss": 0.8546,
+      "step": 3031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.792164037043874e-06,
+      "loss": 0.8608,
+      "step": 3032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.786082555211471e-06,
+      "loss": 0.8104,
+      "step": 3033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.780001934263447e-06,
+      "loss": 0.793,
+      "step": 3034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.773922176564264e-06,
+      "loss": 0.8205,
+      "step": 3035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.767843284478031e-06,
+      "loss": 0.7965,
+      "step": 3036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.761765260368546e-06,
+      "loss": 0.7989,
+      "step": 3037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75568810659924e-06,
+      "loss": 0.8807,
+      "step": 3038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.749611825533231e-06,
+      "loss": 0.8585,
+      "step": 3039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.743536419533289e-06,
+      "loss": 0.8472,
+      "step": 3040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.737461890961841e-06,
+      "loss": 0.7855,
+      "step": 3041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73138824218097e-06,
+      "loss": 0.8447,
+      "step": 3042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.725315475552428e-06,
+      "loss": 0.3258,
+      "step": 3043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71924359343761e-06,
+      "loss": 0.8049,
+      "step": 3044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713172598197587e-06,
+      "loss": 0.7574,
+      "step": 3045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707102492193058e-06,
+      "loss": 0.8027,
+      "step": 3046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.701033277784402e-06,
+      "loss": 0.86,
+      "step": 3047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69496495733163e-06,
+      "loss": 0.804,
+      "step": 3048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688897533194423e-06,
+      "loss": 0.769,
+      "step": 3049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.682831007732105e-06,
+      "loss": 0.8813,
+      "step": 3050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.676765383303649e-06,
+      "loss": 0.8551,
+      "step": 3051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670700662267683e-06,
+      "loss": 0.7673,
+      "step": 3052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664636846982476e-06,
+      "loss": 0.8279,
+      "step": 3053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.658573939805958e-06,
+      "loss": 0.3592,
+      "step": 3054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.652511943095687e-06,
+      "loss": 0.8307,
+      "step": 3055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.646450859208884e-06,
+      "loss": 0.8078,
+      "step": 3056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.640390690502412e-06,
+      "loss": 0.3326,
+      "step": 3057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.634331439332765e-06,
+      "loss": 0.8012,
+      "step": 3058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.628273108056099e-06,
+      "loss": 0.8644,
+      "step": 3059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.622215699028196e-06,
+      "loss": 0.8809,
+      "step": 3060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.616159214604491e-06,
+      "loss": 0.8356,
+      "step": 3061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.610103657140052e-06,
+      "loss": 0.8624,
+      "step": 3062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.604049028989591e-06,
+      "loss": 0.7458,
+      "step": 3063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597995332507462e-06,
+      "loss": 0.8258,
+      "step": 3064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591942570047643e-06,
+      "loss": 0.7914,
+      "step": 3065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585890743963767e-06,
+      "loss": 0.8304,
+      "step": 3066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.579839856609085e-06,
+      "loss": 0.8796,
+      "step": 3067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.573789910336494e-06,
+      "loss": 0.8697,
+      "step": 3068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567740907498529e-06,
+      "loss": 0.8121,
+      "step": 3069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.561692850447343e-06,
+      "loss": 0.752,
+      "step": 3070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.555645741534736e-06,
+      "loss": 0.8315,
+      "step": 3071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.549599583112131e-06,
+      "loss": 0.7561,
+      "step": 3072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.543554377530582e-06,
+      "loss": 0.8093,
+      "step": 3073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.537510127140778e-06,
+      "loss": 0.3373,
+      "step": 3074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.531466834293029e-06,
+      "loss": 0.8102,
+      "step": 3075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525424501337284e-06,
+      "loss": 0.9106,
+      "step": 3076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5193831306231e-06,
+      "loss": 0.7753,
+      "step": 3077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513342724499682e-06,
+      "loss": 0.817,
+      "step": 3078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.507303285315839e-06,
+      "loss": 0.8103,
+      "step": 3079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.501264815420019e-06,
+      "loss": 0.8585,
+      "step": 3080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.495227317160292e-06,
+      "loss": 0.8252,
+      "step": 3081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489190792884338e-06,
+      "loss": 0.764,
+      "step": 3082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.483155244939471e-06,
+      "loss": 0.8362,
+      "step": 3083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.47712067567262e-06,
+      "loss": 0.8693,
+      "step": 3084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.471087087430334e-06,
+      "loss": 0.8172,
+      "step": 3085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4650544825587865e-06,
+      "loss": 0.8396,
+      "step": 3086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.459022863403753e-06,
+      "loss": 0.8196,
+      "step": 3087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.452992232310645e-06,
+      "loss": 0.8004,
+      "step": 3088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446962591624471e-06,
+      "loss": 0.3377,
+      "step": 3089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.440933943689874e-06,
+      "loss": 0.7834,
+      "step": 3090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.434906290851091e-06,
+      "loss": 0.8023,
+      "step": 3091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428879635451986e-06,
+      "loss": 0.3407,
+      "step": 3092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.422853979836035e-06,
+      "loss": 0.8407,
+      "step": 3093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.416829326346313e-06,
+      "loss": 0.7552,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4108056773255245e-06,
+      "loss": 0.7963,
+      "step": 3095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.404783035115961e-06,
+      "loss": 0.8674,
+      "step": 3096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3987614020595385e-06,
+      "loss": 0.7725,
+      "step": 3097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3927407804977814e-06,
+      "loss": 0.8378,
+      "step": 3098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386721172771807e-06,
+      "loss": 0.8388,
+      "step": 3099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.380702581222353e-06,
+      "loss": 0.833,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.37468500818975e-06,
+      "loss": 0.3374,
+      "step": 3101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.368668456013941e-06,
+      "loss": 0.8594,
+      "step": 3102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362652927034474e-06,
+      "loss": 0.8508,
+      "step": 3103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3566384235904855e-06,
+      "loss": 0.7834,
+      "step": 3104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3506249480207284e-06,
+      "loss": 0.8554,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.344612502663548e-06,
+      "loss": 0.8253,
+      "step": 3106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.338601089856894e-06,
+      "loss": 0.8037,
+      "step": 3107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.332590711938303e-06,
+      "loss": 0.8129,
+      "step": 3108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.326581371244923e-06,
+      "loss": 0.852,
+      "step": 3109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.320573070113497e-06,
+      "loss": 0.7381,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.314565810880351e-06,
+      "loss": 0.7558,
+      "step": 3111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.308559595881424e-06,
+      "loss": 0.823,
+      "step": 3112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302554427452231e-06,
+      "loss": 0.7875,
+      "step": 3113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.296550307927892e-06,
+      "loss": 0.8195,
+      "step": 3114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.290547239643117e-06,
+      "loss": 0.8311,
+      "step": 3115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.284545224932207e-06,
+      "loss": 0.772,
+      "step": 3116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.278544266129054e-06,
+      "loss": 0.7893,
+      "step": 3117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.272544365567131e-06,
+      "loss": 0.861,
+      "step": 3118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.266545525579509e-06,
+      "loss": 0.8679,
+      "step": 3119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.260547748498851e-06,
+      "loss": 0.7771,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.254551036657389e-06,
+      "loss": 0.8496,
+      "step": 3121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24855539238696e-06,
+      "loss": 0.8093,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24256081801897e-06,
+      "loss": 0.8196,
+      "step": 3123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236567315884422e-06,
+      "loss": 0.832,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.23057488831389e-06,
+      "loss": 0.8724,
+      "step": 3125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.224583537637544e-06,
+      "loss": 0.796,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.218593266185125e-06,
+      "loss": 0.8281,
+      "step": 3127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.212604076285952e-06,
+      "loss": 0.8355,
+      "step": 3128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.206615970268938e-06,
+      "loss": 0.7668,
+      "step": 3129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.200628950462556e-06,
+      "loss": 0.3397,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.194643019194867e-06,
+      "loss": 0.8167,
+      "step": 3131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.188658178793516e-06,
+      "loss": 0.8111,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.182674431585703e-06,
+      "loss": 0.8438,
+      "step": 3133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1766917798982236e-06,
+      "loss": 0.7741,
+      "step": 3134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1707102260574376e-06,
+      "loss": 0.8115,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.164729772389276e-06,
+      "loss": 0.8099,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158750421219245e-06,
+      "loss": 0.8435,
+      "step": 3137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.152772174872424e-06,
+      "loss": 0.8575,
+      "step": 3138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.146795035673466e-06,
+      "loss": 0.8316,
+      "step": 3139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.14081900594658e-06,
+      "loss": 0.8726,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.13484408801556e-06,
+      "loss": 0.7516,
+      "step": 3141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.12887028420375e-06,
+      "loss": 0.7452,
+      "step": 3142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.122897596834078e-06,
+      "loss": 0.7986,
+      "step": 3143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.116926028229032e-06,
+      "loss": 0.8611,
+      "step": 3144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.110955580710656e-06,
+      "loss": 0.818,
+      "step": 3145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1049862566005725e-06,
+      "loss": 0.3083,
+      "step": 3146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.099018058219953e-06,
+      "loss": 0.799,
+      "step": 3147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.093050987889547e-06,
+      "loss": 0.8718,
+      "step": 3148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.087085047929652e-06,
+      "loss": 0.832,
+      "step": 3149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.081120240660128e-06,
+      "loss": 0.8076,
+      "step": 3150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.075156568400406e-06,
+      "loss": 0.8131,
+      "step": 3151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.069194033469455e-06,
+      "loss": 0.859,
+      "step": 3152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.063232638185825e-06,
+      "loss": 0.7733,
+      "step": 3153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.057272384867603e-06,
+      "loss": 0.8336,
+      "step": 3154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.051313275832443e-06,
+      "loss": 0.7619,
+      "step": 3155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.045355313397555e-06,
+      "loss": 0.7957,
+      "step": 3156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0393984998796976e-06,
+      "loss": 0.8525,
+      "step": 3157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.033442837595183e-06,
+      "loss": 0.8423,
+      "step": 3158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027488328859876e-06,
+      "loss": 0.7947,
+      "step": 3159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.021534975989197e-06,
+      "loss": 0.854,
+      "step": 3160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.015582781298118e-06,
+      "loss": 0.8536,
+      "step": 3161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00963174710115e-06,
+      "loss": 0.8862,
+      "step": 3162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.003681875712366e-06,
+      "loss": 0.7973,
+      "step": 3163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.997733169445372e-06,
+      "loss": 0.8504,
+      "step": 3164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.991785630613334e-06,
+      "loss": 0.7884,
+      "step": 3165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.985839261528966e-06,
+      "loss": 0.8562,
+      "step": 3166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.979894064504508e-06,
+      "loss": 0.85,
+      "step": 3167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9739500418517645e-06,
+      "loss": 0.8349,
+      "step": 3168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.968007195882074e-06,
+      "loss": 0.8246,
+      "step": 3169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.962065528906321e-06,
+      "loss": 0.8182,
+      "step": 3170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.956125043234925e-06,
+      "loss": 0.3188,
+      "step": 3171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.950185741177853e-06,
+      "loss": 0.854,
+      "step": 3172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9442476250446156e-06,
+      "loss": 0.8613,
+      "step": 3173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.938310697144245e-06,
+      "loss": 0.8718,
+      "step": 3174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.932374959785333e-06,
+      "loss": 0.832,
+      "step": 3175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.926440415275988e-06,
+      "loss": 0.8207,
+      "step": 3176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.920507065923871e-06,
+      "loss": 0.7873,
+      "step": 3177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.914574914036172e-06,
+      "loss": 0.8192,
+      "step": 3178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.908643961919614e-06,
+      "loss": 0.874,
+      "step": 3179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.902714211880454e-06,
+      "loss": 0.7681,
+      "step": 3180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.896785666224482e-06,
+      "loss": 0.8054,
+      "step": 3181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.890858327257022e-06,
+      "loss": 0.8059,
+      "step": 3182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.884932197282923e-06,
+      "loss": 0.8549,
+      "step": 3183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879007278606569e-06,
+      "loss": 0.7744,
+      "step": 3184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.873083573531876e-06,
+      "loss": 0.7828,
+      "step": 3185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.867161084362275e-06,
+      "loss": 0.8913,
+      "step": 3186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.861239813400742e-06,
+      "loss": 0.8361,
+      "step": 3187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.855319762949763e-06,
+      "loss": 0.7849,
+      "step": 3188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.849400935311359e-06,
+      "loss": 0.8092,
+      "step": 3189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.843483332787076e-06,
+      "loss": 0.7803,
+      "step": 3190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.837566957677974e-06,
+      "loss": 0.8806,
+      "step": 3191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.831651812284652e-06,
+      "loss": 0.8067,
+      "step": 3192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.82573789890721e-06,
+      "loss": 0.3381,
+      "step": 3193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.819825219845286e-06,
+      "loss": 0.8426,
+      "step": 3194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8139137773980364e-06,
+      "loss": 0.8023,
+      "step": 3195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8080035738641236e-06,
+      "loss": 0.788,
+      "step": 3196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.802094611541745e-06,
+      "loss": 0.7775,
+      "step": 3197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7961868927286005e-06,
+      "loss": 0.8673,
+      "step": 3198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.790280419721919e-06,
+      "loss": 0.8561,
+      "step": 3199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.784375194818435e-06,
+      "loss": 0.8124,
+      "step": 3200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.778471220314403e-06,
+      "loss": 0.8352,
+      "step": 3201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7725684985055965e-06,
+      "loss": 0.8851,
+      "step": 3202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.766667031687286e-06,
+      "loss": 0.7849,
+      "step": 3203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760766822154274e-06,
+      "loss": 0.7803,
+      "step": 3204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.754867872200854e-06,
+      "loss": 0.815,
+      "step": 3205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.748970184120844e-06,
+      "loss": 0.8329,
+      "step": 3206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.743073760207573e-06,
+      "loss": 0.8777,
+      "step": 3207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.737178602753859e-06,
+      "loss": 0.8185,
+      "step": 3208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.731284714052051e-06,
+      "loss": 0.8091,
+      "step": 3209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.72539209639399e-06,
+      "loss": 0.8652,
+      "step": 3210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719500752071033e-06,
+      "loss": 0.8562,
+      "step": 3211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.713610683374028e-06,
+      "loss": 0.7913,
+      "step": 3212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.707721892593339e-06,
+      "loss": 0.8067,
+      "step": 3213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7018343820188324e-06,
+      "loss": 0.8194,
+      "step": 3214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695948153939869e-06,
+      "loss": 0.7922,
+      "step": 3215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.690063210645318e-06,
+      "loss": 0.8256,
+      "step": 3216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.684179554423542e-06,
+      "loss": 0.835,
+      "step": 3217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6782971875624105e-06,
+      "loss": 0.3252,
+      "step": 3218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.672416112349293e-06,
+      "loss": 0.8093,
+      "step": 3219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.666536331071045e-06,
+      "loss": 0.7683,
+      "step": 3220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660657846014031e-06,
+      "loss": 0.8046,
+      "step": 3221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.654780659464105e-06,
+      "loss": 0.7685,
+      "step": 3222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.648904773706617e-06,
+      "loss": 0.7718,
+      "step": 3223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.643030191026419e-06,
+      "loss": 0.8616,
+      "step": 3224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.637156913707839e-06,
+      "loss": 0.8344,
+      "step": 3225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.631284944034716e-06,
+      "loss": 0.7918,
+      "step": 3226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6254142842903644e-06,
+      "loss": 0.7854,
+      "step": 3227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.619544936757607e-06,
+      "loss": 0.7871,
+      "step": 3228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.613676903718735e-06,
+      "loss": 0.3047,
+      "step": 3229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6078101874555465e-06,
+      "loss": 0.8263,
+      "step": 3230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601944790249322e-06,
+      "loss": 0.8049,
+      "step": 3231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596080714380825e-06,
+      "loss": 0.818,
+      "step": 3232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5902179621303115e-06,
+      "loss": 0.8798,
+      "step": 3233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5843565357775165e-06,
+      "loss": 0.7902,
+      "step": 3234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.578496437601661e-06,
+      "loss": 0.8291,
+      "step": 3235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.572637669881459e-06,
+      "loss": 0.8217,
+      "step": 3236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.566780234895089e-06,
+      "loss": 0.7745,
+      "step": 3237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56092413492023e-06,
+      "loss": 0.8013,
+      "step": 3238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5550693722340255e-06,
+      "loss": 0.8452,
+      "step": 3239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.54921594911311e-06,
+      "loss": 0.7991,
+      "step": 3240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.543363867833597e-06,
+      "loss": 0.7869,
+      "step": 3241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5375131306710725e-06,
+      "loss": 0.3059,
+      "step": 3242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.531663739900603e-06,
+      "loss": 0.7514,
+      "step": 3243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.525815697796729e-06,
+      "loss": 0.822,
+      "step": 3244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519969006633473e-06,
+      "loss": 0.8251,
+      "step": 3245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.514123668684321e-06,
+      "loss": 0.8179,
+      "step": 3246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.508279686222243e-06,
+      "loss": 0.3287,
+      "step": 3247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.502437061519684e-06,
+      "loss": 0.7788,
+      "step": 3248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.496595796848545e-06,
+      "loss": 0.7925,
+      "step": 3249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.490755894480218e-06,
+      "loss": 0.7517,
+      "step": 3250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484917356685546e-06,
+      "loss": 0.8389,
+      "step": 3251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479080185734859e-06,
+      "loss": 0.8369,
+      "step": 3252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4732443838979454e-06,
+      "loss": 0.8376,
+      "step": 3253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467409953444061e-06,
+      "loss": 0.8367,
+      "step": 3254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.461576896641939e-06,
+      "loss": 0.8343,
+      "step": 3255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.455745215759761e-06,
+      "loss": 0.8241,
+      "step": 3256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.449914913065189e-06,
+      "loss": 0.7962,
+      "step": 3257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.444085990825338e-06,
+      "loss": 0.8169,
+      "step": 3258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.438258451306794e-06,
+      "loss": 0.8199,
+      "step": 3259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.432432296775606e-06,
+      "loss": 0.8054,
+      "step": 3260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.426607529497273e-06,
+      "loss": 0.8247,
+      "step": 3261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.42078415173677e-06,
+      "loss": 0.8221,
+      "step": 3262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4149621657585174e-06,
+      "loss": 0.8117,
+      "step": 3263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.409141573826407e-06,
+      "loss": 0.7937,
+      "step": 3264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.403322378203782e-06,
+      "loss": 0.8023,
+      "step": 3265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.39750458115344e-06,
+      "loss": 0.9092,
+      "step": 3266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.391688184937643e-06,
+      "loss": 0.8143,
+      "step": 3267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.385873191818099e-06,
+      "loss": 0.7926,
+      "step": 3268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.380059604055975e-06,
+      "loss": 0.8141,
+      "step": 3269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.374247423911897e-06,
+      "loss": 0.8871,
+      "step": 3270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.368436653645931e-06,
+      "loss": 0.7445,
+      "step": 3271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.362627295517606e-06,
+      "loss": 0.8742,
+      "step": 3272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.356819351785897e-06,
+      "loss": 0.7993,
+      "step": 3273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.351012824709228e-06,
+      "loss": 0.7636,
+      "step": 3274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.345207716545474e-06,
+      "loss": 0.8231,
+      "step": 3275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.339404029551959e-06,
+      "loss": 0.8695,
+      "step": 3276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333601765985458e-06,
+      "loss": 0.8262,
+      "step": 3277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.327800928102178e-06,
+      "loss": 0.8447,
+      "step": 3278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.322001518157792e-06,
+      "loss": 0.8143,
+      "step": 3279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.316203538407397e-06,
+      "loss": 0.8207,
+      "step": 3280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.310406991105551e-06,
+      "loss": 0.8648,
+      "step": 3281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30461187850625e-06,
+      "loss": 0.8765,
+      "step": 3282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2988182028629245e-06,
+      "loss": 0.8041,
+      "step": 3283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.293025966428456e-06,
+      "loss": 0.8423,
+      "step": 3284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287235171455157e-06,
+      "loss": 0.8578,
+      "step": 3285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.281445820194795e-06,
+      "loss": 0.3385,
+      "step": 3286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.275657914898554e-06,
+      "loss": 0.8586,
+      "step": 3287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.269871457817074e-06,
+      "loss": 0.7795,
+      "step": 3288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.264086451200431e-06,
+      "loss": 0.8336,
+      "step": 3289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.258302897298119e-06,
+      "loss": 0.8253,
+      "step": 3290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252520798359092e-06,
+      "loss": 0.852,
+      "step": 3291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.246740156631714e-06,
+      "loss": 0.8179,
+      "step": 3292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240960974363803e-06,
+      "loss": 0.7529,
+      "step": 3293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2351832538025995e-06,
+      "loss": 0.8051,
+      "step": 3294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.229406997194774e-06,
+      "loss": 0.8457,
+      "step": 3295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.223632206786434e-06,
+      "loss": 0.8202,
+      "step": 3296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.217858884823109e-06,
+      "loss": 0.7978,
+      "step": 3297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.212087033549765e-06,
+      "loss": 0.3407,
+      "step": 3298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206316655210798e-06,
+      "loss": 0.8161,
+      "step": 3299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2005477520500145e-06,
+      "loss": 0.8466,
+      "step": 3300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.194780326310672e-06,
+      "loss": 0.8755,
+      "step": 3301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18901438023543e-06,
+      "loss": 0.8046,
+      "step": 3302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.183249916066389e-06,
+      "loss": 0.8164,
+      "step": 3303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177486936045067e-06,
+      "loss": 0.8237,
+      "step": 3304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.171725442412404e-06,
+      "loss": 0.8263,
+      "step": 3305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.165965437408766e-06,
+      "loss": 0.8221,
+      "step": 3306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.160206923273934e-06,
+      "loss": 0.816,
+      "step": 3307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.154449902247119e-06,
+      "loss": 0.8644,
+      "step": 3308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.148694376566937e-06,
+      "loss": 0.776,
+      "step": 3309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.142940348471437e-06,
+      "loss": 0.336,
+      "step": 3310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.137187820198081e-06,
+      "loss": 0.8474,
+      "step": 3311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.131436793983738e-06,
+      "loss": 0.7614,
+      "step": 3312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.125687272064713e-06,
+      "loss": 0.8115,
+      "step": 3313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.119939256676701e-06,
+      "loss": 0.8326,
+      "step": 3314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.114192750054833e-06,
+      "loss": 0.7997,
+      "step": 3315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.108447754433644e-06,
+      "loss": 0.8411,
+      "step": 3316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.10270427204708e-06,
+      "loss": 0.8148,
+      "step": 3317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.096962305128506e-06,
+      "loss": 0.7893,
+      "step": 3318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.091221855910685e-06,
+      "loss": 0.8173,
+      "step": 3319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.085482926625803e-06,
+      "loss": 0.7479,
+      "step": 3320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.079745519505446e-06,
+      "loss": 0.8376,
+      "step": 3321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.074009636780612e-06,
+      "loss": 0.8375,
+      "step": 3322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.068275280681709e-06,
+      "loss": 0.7953,
+      "step": 3323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0625424534385425e-06,
+      "loss": 0.8075,
+      "step": 3324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056811157280333e-06,
+      "loss": 0.8038,
+      "step": 3325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.051081394435702e-06,
+      "loss": 0.8046,
+      "step": 3326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.045353167132671e-06,
+      "loss": 0.789,
+      "step": 3327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.039626477598673e-06,
+      "loss": 0.7894,
+      "step": 3328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.033901328060532e-06,
+      "loss": 0.7477,
+      "step": 3329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.028177720744486e-06,
+      "loss": 0.7863,
+      "step": 3330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.022455657876158e-06,
+      "loss": 0.7515,
+      "step": 3331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.016735141680587e-06,
+      "loss": 0.8347,
+      "step": 3332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.011016174382196e-06,
+      "loss": 0.8049,
+      "step": 3333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.005298758204814e-06,
+      "loss": 0.819,
+      "step": 3334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.999582895371669e-06,
+      "loss": 0.8432,
+      "step": 3335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.993868588105375e-06,
+      "loss": 0.7985,
+      "step": 3336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.988155838627951e-06,
+      "loss": 0.8445,
+      "step": 3337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9824446491608035e-06,
+      "loss": 0.8394,
+      "step": 3338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.976735021924735e-06,
+      "loss": 0.8148,
+      "step": 3339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.971026959139947e-06,
+      "loss": 0.8268,
+      "step": 3340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.965320463026019e-06,
+      "loss": 0.7988,
+      "step": 3341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.959615535801933e-06,
+      "loss": 0.8127,
+      "step": 3342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.953912179686052e-06,
+      "loss": 0.8006,
+      "step": 3343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.948210396896137e-06,
+      "loss": 0.8535,
+      "step": 3344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.942510189649333e-06,
+      "loss": 0.3633,
+      "step": 3345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.936811560162169e-06,
+      "loss": 0.8483,
+      "step": 3346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.931114510650566e-06,
+      "loss": 0.7755,
+      "step": 3347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.925419043329827e-06,
+      "loss": 0.8178,
+      "step": 3348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.919725160414645e-06,
+      "loss": 0.8582,
+      "step": 3349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9140328641190855e-06,
+      "loss": 0.8107,
+      "step": 3350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9083421566566105e-06,
+      "loss": 0.8567,
+      "step": 3351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.902653040240059e-06,
+      "loss": 0.7942,
+      "step": 3352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.896965517081645e-06,
+      "loss": 0.8051,
+      "step": 3353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.891279589392975e-06,
+      "loss": 0.7594,
+      "step": 3354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.885595259385022e-06,
+      "loss": 0.8162,
+      "step": 3355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.879912529268148e-06,
+      "loss": 0.8816,
+      "step": 3356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.874231401252091e-06,
+      "loss": 0.7455,
+      "step": 3357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868551877545962e-06,
+      "loss": 0.806,
+      "step": 3358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.862873960358251e-06,
+      "loss": 0.8185,
+      "step": 3359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.857197651896819e-06,
+      "loss": 0.8021,
+      "step": 3360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.851522954368909e-06,
+      "loss": 0.8123,
+      "step": 3361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 0.7809,
+      "step": 3362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8401784009394815e-06,
+      "loss": 0.3431,
+      "step": 3363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.834508549449305e-06,
+      "loss": 0.8494,
+      "step": 3364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828840317715331e-06,
+      "loss": 0.8432,
+      "step": 3365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.823173707941665e-06,
+      "loss": 0.8319,
+      "step": 3366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8175087223317674e-06,
+      "loss": 0.7939,
+      "step": 3367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.811845363088477e-06,
+      "loss": 0.7915,
+      "step": 3368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.806183632414003e-06,
+      "loss": 0.3186,
+      "step": 3369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.800523532509907e-06,
+      "loss": 0.888,
+      "step": 3370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.794865065577134e-06,
+      "loss": 0.8243,
+      "step": 3371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7892082338159774e-06,
+      "loss": 0.7748,
+      "step": 3372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.783553039426107e-06,
+      "loss": 0.783,
+      "step": 3373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.777899484606555e-06,
+      "loss": 0.876,
+      "step": 3374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7722475715557056e-06,
+      "loss": 0.797,
+      "step": 3375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.766597302471319e-06,
+      "loss": 0.8185,
+      "step": 3376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.760948679550502e-06,
+      "loss": 0.8085,
+      "step": 3377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755301704989735e-06,
+      "loss": 0.3746,
+      "step": 3378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7496563809848434e-06,
+      "loss": 0.8276,
+      "step": 3379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7440127097310214e-06,
+      "loss": 0.7903,
+      "step": 3380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.738370693422822e-06,
+      "loss": 0.8463,
+      "step": 3381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7327303342541415e-06,
+      "loss": 0.7753,
+      "step": 3382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.727091634418243e-06,
+      "loss": 0.8576,
+      "step": 3383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.721454596107747e-06,
+      "loss": 0.8346,
+      "step": 3384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7158192215146154e-06,
+      "loss": 0.36,
+      "step": 3385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.710185512830171e-06,
+      "loss": 0.768,
+      "step": 3386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.70455347224509e-06,
+      "loss": 0.8478,
+      "step": 3387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6989231019494016e-06,
+      "loss": 0.8063,
+      "step": 3388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.693294404132474e-06,
+      "loss": 0.8229,
+      "step": 3389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.687667380983037e-06,
+      "loss": 0.325,
+      "step": 3390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.682042034689168e-06,
+      "loss": 0.7907,
+      "step": 3391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6764183674382834e-06,
+      "loss": 0.7728,
+      "step": 3392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.67079638141716e-06,
+      "loss": 0.8037,
+      "step": 3393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.665176078811905e-06,
+      "loss": 0.8286,
+      "step": 3394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.659557461807987e-06,
+      "loss": 0.8532,
+      "step": 3395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653940532590205e-06,
+      "loss": 0.834,
+      "step": 3396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.648325293342714e-06,
+      "loss": 0.8442,
+      "step": 3397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.642711746249006e-06,
+      "loss": 0.8689,
+      "step": 3398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.637099893491911e-06,
+      "loss": 0.8027,
+      "step": 3399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6314897372536115e-06,
+      "loss": 0.8023,
+      "step": 3400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.625881279715615e-06,
+      "loss": 0.7676,
+      "step": 3401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.620274523058781e-06,
+      "loss": 0.3429,
+      "step": 3402
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.614669469463307e-06,
+      "loss": 0.836,
+      "step": 3403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.609066121108717e-06,
+      "loss": 0.8378,
+      "step": 3404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603464480173884e-06,
+      "loss": 0.8398,
+      "step": 3405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.597864548837011e-06,
+      "loss": 0.7729,
+      "step": 3406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.592266329275641e-06,
+      "loss": 0.791,
+      "step": 3407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.586669823666646e-06,
+      "loss": 0.8601,
+      "step": 3408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.581075034186232e-06,
+      "loss": 0.8106,
+      "step": 3409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.575481963009945e-06,
+      "loss": 0.8373,
+      "step": 3410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.569890612312649e-06,
+      "loss": 0.7904,
+      "step": 3411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.564300984268556e-06,
+      "loss": 0.8621,
+      "step": 3412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.558713081051192e-06,
+      "loss": 0.775,
+      "step": 3413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5531269048334216e-06,
+      "loss": 0.8631,
+      "step": 3414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.547542457787442e-06,
+      "loss": 0.7556,
+      "step": 3415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.541959742084761e-06,
+      "loss": 0.8193,
+      "step": 3416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.536378759896237e-06,
+      "loss": 0.7598,
+      "step": 3417
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.530799513392029e-06,
+      "loss": 0.8603,
+      "step": 3418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.525222004741639e-06,
+      "loss": 0.7829,
+      "step": 3419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.519646236113891e-06,
+      "loss": 0.7555,
+      "step": 3420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.514072209676924e-06,
+      "loss": 0.7855,
+      "step": 3421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.508499927598209e-06,
+      "loss": 0.8088,
+      "step": 3422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.502929392044528e-06,
+      "loss": 0.7716,
+      "step": 3423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497360605181998e-06,
+      "loss": 0.8149,
+      "step": 3424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.491793569176042e-06,
+      "loss": 0.7839,
+      "step": 3425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4862282861914095e-06,
+      "loss": 0.8311,
+      "step": 3426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.480664758392169e-06,
+      "loss": 0.3351,
+      "step": 3427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.475102987941705e-06,
+      "loss": 0.8018,
+      "step": 3428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.469542977002722e-06,
+      "loss": 0.7522,
+      "step": 3429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.463984727737227e-06,
+      "loss": 0.8185,
+      "step": 3430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.45842824230656e-06,
+      "loss": 0.36,
+      "step": 3431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.452873522871365e-06,
+      "loss": 0.8279,
+      "step": 3432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.447320571591598e-06,
+      "loss": 0.3727,
+      "step": 3433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.441769390626537e-06,
+      "loss": 0.7684,
+      "step": 3434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.436219982134756e-06,
+      "loss": 0.7967,
+      "step": 3435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.430672348274155e-06,
+      "loss": 0.8265,
+      "step": 3436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.42512649120194e-06,
+      "loss": 0.7969,
+      "step": 3437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.419582413074618e-06,
+      "loss": 0.3407,
+      "step": 3438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.414040116048016e-06,
+      "loss": 0.7957,
+      "step": 3439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.408499602277256e-06,
+      "loss": 0.8321,
+      "step": 3440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4029608739167815e-06,
+      "loss": 0.8694,
+      "step": 3441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.397423933120325e-06,
+      "loss": 0.7776,
+      "step": 3442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3918887820409364e-06,
+      "loss": 0.8128,
+      "step": 3443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.386355422830969e-06,
+      "loss": 0.8177,
+      "step": 3444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3808238576420694e-06,
+      "loss": 0.823,
+      "step": 3445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.375294088625196e-06,
+      "loss": 0.7876,
+      "step": 3446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36976611793061e-06,
+      "loss": 0.8342,
+      "step": 3447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.364239947707861e-06,
+      "loss": 0.8229,
+      "step": 3448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.358715580105813e-06,
+      "loss": 0.7775,
+      "step": 3449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3531930172726195e-06,
+      "loss": 0.8253,
+      "step": 3450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.347672261355742e-06,
+      "loss": 0.8012,
+      "step": 3451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.342153314501923e-06,
+      "loss": 0.7721,
+      "step": 3452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3366361788572215e-06,
+      "loss": 0.7714,
+      "step": 3453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.331120856566972e-06,
+      "loss": 0.825,
+      "step": 3454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.325607349775818e-06,
+      "loss": 0.8805,
+      "step": 3455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3200956606277006e-06,
+      "loss": 0.7797,
+      "step": 3456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.314585791265835e-06,
+      "loss": 0.7142,
+      "step": 3457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.309077743832749e-06,
+      "loss": 0.8986,
+      "step": 3458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.303571520470245e-06,
+      "loss": 0.7744,
+      "step": 3459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29806712331943e-06,
+      "loss": 0.8868,
+      "step": 3460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.292564554520698e-06,
+      "loss": 0.7903,
+      "step": 3461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28706381621372e-06,
+      "loss": 0.8415,
+      "step": 3462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.281564910537476e-06,
+      "loss": 0.8639,
+      "step": 3463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.27606783963021e-06,
+      "loss": 0.8027,
+      "step": 3464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.270572605629471e-06,
+      "loss": 0.8363,
+      "step": 3465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2650792106720905e-06,
+      "loss": 0.7996,
+      "step": 3466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.259587656894174e-06,
+      "loss": 0.3087,
+      "step": 3467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.254097946431121e-06,
+      "loss": 0.7982,
+      "step": 3468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.248610081417617e-06,
+      "loss": 0.8026,
+      "step": 3469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.243124063987618e-06,
+      "loss": 0.8203,
+      "step": 3470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.237639896274369e-06,
+      "loss": 0.7842,
+      "step": 3471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.232157580410397e-06,
+      "loss": 0.8305,
+      "step": 3472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226677118527512e-06,
+      "loss": 0.8039,
+      "step": 3473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221198512756789e-06,
+      "loss": 0.8494,
+      "step": 3474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.215721765228596e-06,
+      "loss": 0.8452,
+      "step": 3475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2102468780725675e-06,
+      "loss": 0.8237,
+      "step": 3476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.204773853417624e-06,
+      "loss": 0.7959,
+      "step": 3477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.199302693391958e-06,
+      "loss": 0.8314,
+      "step": 3478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.193833400123034e-06,
+      "loss": 0.7735,
+      "step": 3479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.188365975737595e-06,
+      "loss": 0.7857,
+      "step": 3480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.182900422361652e-06,
+      "loss": 0.8177,
+      "step": 3481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1774367421204975e-06,
+      "loss": 0.857,
+      "step": 3482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171974937138682e-06,
+      "loss": 0.7677,
+      "step": 3483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1665150095400395e-06,
+      "loss": 0.7969,
+      "step": 3484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1610569614476735e-06,
+      "loss": 0.7737,
+      "step": 3485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1556007949839435e-06,
+      "loss": 0.8488,
+      "step": 3486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.150146512270494e-06,
+      "loss": 0.8179,
+      "step": 3487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.144694115428222e-06,
+      "loss": 0.7664,
+      "step": 3488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.139243606577302e-06,
+      "loss": 0.8202,
+      "step": 3489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1337949878371705e-06,
+      "loss": 0.7998,
+      "step": 3490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1283482613265305e-06,
+      "loss": 0.837,
+      "step": 3491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1229034291633505e-06,
+      "loss": 0.8342,
+      "step": 3492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.117460493464852e-06,
+      "loss": 0.892,
+      "step": 3493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1120194563475315e-06,
+      "loss": 0.7833,
+      "step": 3494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.106580319927146e-06,
+      "loss": 0.7476,
+      "step": 3495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.101143086318703e-06,
+      "loss": 0.8085,
+      "step": 3496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.095707757636484e-06,
+      "loss": 0.8024,
+      "step": 3497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090274335994014e-06,
+      "loss": 0.7636,
+      "step": 3498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.084842823504096e-06,
+      "loss": 0.3438,
+      "step": 3499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.07941322227877e-06,
+      "loss": 0.8404,
+      "step": 3500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.073985534429349e-06,
+      "loss": 0.7947,
+      "step": 3501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.068559762066396e-06,
+      "loss": 0.8065,
+      "step": 3502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.063135907299726e-06,
+      "loss": 0.7654,
+      "step": 3503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.057713972238414e-06,
+      "loss": 0.8005,
+      "step": 3504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.05229395899078e-06,
+      "loss": 0.8328,
+      "step": 3505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.046875869664407e-06,
+      "loss": 0.7928,
+      "step": 3506
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.04145970636613e-06,
+      "loss": 0.8462,
+      "step": 3507
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.03604547120202e-06,
+      "loss": 0.7942,
+      "step": 3508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0306331662774185e-06,
+      "loss": 0.8563,
+      "step": 3509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.025222793696898e-06,
+      "loss": 0.8102,
+      "step": 3510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.019814355564292e-06,
+      "loss": 0.8144,
+      "step": 3511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.014407853982679e-06,
+      "loss": 0.8461,
+      "step": 3512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.009003291054382e-06,
+      "loss": 0.8046,
+      "step": 3513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.003600668880975e-06,
+      "loss": 0.8074,
+      "step": 3514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.998199989563269e-06,
+      "loss": 0.7843,
+      "step": 3515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.99280125520133e-06,
+      "loss": 0.8508,
+      "step": 3516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.987404467894456e-06,
+      "loss": 0.7839,
+      "step": 3517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9820096297411945e-06,
+      "loss": 0.8161,
+      "step": 3518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.976616742839342e-06,
+      "loss": 0.7964,
+      "step": 3519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.97122580928592e-06,
+      "loss": 0.7729,
+      "step": 3520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965836831177206e-06,
+      "loss": 0.7945,
+      "step": 3521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.960449810608705e-06,
+      "loss": 0.8724,
+      "step": 3522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.955064749675168e-06,
+      "loss": 0.7856,
+      "step": 3523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.949681650470587e-06,
+      "loss": 0.3385,
+      "step": 3524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.944300515088178e-06,
+      "loss": 0.2956,
+      "step": 3525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.938921345620411e-06,
+      "loss": 0.7859,
+      "step": 3526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.933544144158972e-06,
+      "loss": 0.8163,
+      "step": 3527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.928168912794803e-06,
+      "loss": 0.8314,
+      "step": 3528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.922795653618059e-06,
+      "loss": 0.7964,
+      "step": 3529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917424368718141e-06,
+      "loss": 0.8418,
+      "step": 3530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.912055060183687e-06,
+      "loss": 0.8145,
+      "step": 3531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.906687730102547e-06,
+      "loss": 0.7613,
+      "step": 3532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.90132238056182e-06,
+      "loss": 0.365,
+      "step": 3533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.895959013647828e-06,
+      "loss": 0.8005,
+      "step": 3534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890597631446124e-06,
+      "loss": 0.7901,
+      "step": 3535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.885238236041487e-06,
+      "loss": 0.8115,
+      "step": 3536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879880829517921e-06,
+      "loss": 0.8385,
+      "step": 3537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.874525413958665e-06,
+      "loss": 0.8414,
+      "step": 3538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8691719914461725e-06,
+      "loss": 0.3372,
+      "step": 3539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.86382056406213e-06,
+      "loss": 0.8306,
+      "step": 3540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.858471133887451e-06,
+      "loss": 0.8302,
+      "step": 3541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853123703002262e-06,
+      "loss": 0.8242,
+      "step": 3542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8477782734859215e-06,
+      "loss": 0.7929,
+      "step": 3543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.842434847417001e-06,
+      "loss": 0.8163,
+      "step": 3544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.837093426873306e-06,
+      "loss": 0.7963,
+      "step": 3545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.831754013931844e-06,
+      "loss": 0.8374,
+      "step": 3546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826416610668858e-06,
+      "loss": 0.8068,
+      "step": 3547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.821081219159806e-06,
+      "loss": 0.7899,
+      "step": 3548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.815747841479356e-06,
+      "loss": 0.7539,
+      "step": 3549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.810416479701402e-06,
+      "loss": 0.7654,
+      "step": 3550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8050871358990446e-06,
+      "loss": 0.7382,
+      "step": 3551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.79975981214461e-06,
+      "loss": 0.9072,
+      "step": 3552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.794434510509632e-06,
+      "loss": 0.8784,
+      "step": 3553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.789111233064867e-06,
+      "loss": 0.8007,
+      "step": 3554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.783789981880267e-06,
+      "loss": 0.7839,
+      "step": 3555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7784707590250124e-06,
+      "loss": 0.8827,
+      "step": 3556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773153566567489e-06,
+      "loss": 0.7884,
+      "step": 3557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.767838406575297e-06,
+      "loss": 0.841,
+      "step": 3558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.762525281115235e-06,
+      "loss": 0.3137,
+      "step": 3559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7572141922533245e-06,
+      "loss": 0.785,
+      "step": 3560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.751905142054783e-06,
+      "loss": 0.8186,
+      "step": 3561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.746598132584047e-06,
+      "loss": 0.819,
+      "step": 3562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.741293165904747e-06,
+      "loss": 0.7991,
+      "step": 3563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.735990244079727e-06,
+      "loss": 0.8296,
+      "step": 3564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.73068936917104e-06,
+      "loss": 0.8496,
+      "step": 3565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7253905432399295e-06,
+      "loss": 0.8157,
+      "step": 3566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.720093768346858e-06,
+      "loss": 0.7786,
+      "step": 3567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.714799046551474e-06,
+      "loss": 0.7779,
+      "step": 3568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.709506379912641e-06,
+      "loss": 0.3363,
+      "step": 3569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.704215770488422e-06,
+      "loss": 0.8065,
+      "step": 3570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.698927220336068e-06,
+      "loss": 0.8314,
+      "step": 3571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6936407315120465e-06,
+      "loss": 0.8236,
+      "step": 3572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.688356306072009e-06,
+      "loss": 0.8589,
+      "step": 3573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.683073946070812e-06,
+      "loss": 0.8198,
+      "step": 3574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677793653562507e-06,
+      "loss": 0.8007,
+      "step": 3575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.672515430600347e-06,
+      "loss": 0.7845,
+      "step": 3576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667239279236768e-06,
+      "loss": 0.862,
+      "step": 3577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6619652015234095e-06,
+      "loss": 0.7653,
+      "step": 3578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.656693199511108e-06,
+      "loss": 0.848,
+      "step": 3579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65142327524988e-06,
+      "loss": 0.8368,
+      "step": 3580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.646155430788944e-06,
+      "loss": 0.7522,
+      "step": 3581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.640889668176713e-06,
+      "loss": 0.7865,
+      "step": 3582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.635625989460778e-06,
+      "loss": 0.3479,
+      "step": 3583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.630364396687932e-06,
+      "loss": 0.7278,
+      "step": 3584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.625104891904144e-06,
+      "loss": 0.7964,
+      "step": 3585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619847477154584e-06,
+      "loss": 0.8127,
+      "step": 3586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.614592154483607e-06,
+      "loss": 0.7879,
+      "step": 3587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.609338925934743e-06,
+      "loss": 0.8596,
+      "step": 3588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6040877935507245e-06,
+      "loss": 0.8046,
+      "step": 3589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.598838759373453e-06,
+      "loss": 0.8403,
+      "step": 3590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.593591825444028e-06,
+      "loss": 0.7715,
+      "step": 3591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.588346993802719e-06,
+      "loss": 0.8018,
+      "step": 3592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.583104266488989e-06,
+      "loss": 0.7617,
+      "step": 3593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.577863645541483e-06,
+      "loss": 0.8629,
+      "step": 3594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.572625132998013e-06,
+      "loss": 0.8055,
+      "step": 3595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567388730895586e-06,
+      "loss": 0.8247,
+      "step": 3596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.562154441270383e-06,
+      "loss": 0.341,
+      "step": 3597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.556922266157762e-06,
+      "loss": 0.8183,
+      "step": 3598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551692207592265e-06,
+      "loss": 0.9032,
+      "step": 3599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.546464267607601e-06,
+      "loss": 0.8285,
+      "step": 3600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5412384482366665e-06,
+      "loss": 0.8631,
+      "step": 3601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536014751511518e-06,
+      "loss": 0.7492,
+      "step": 3602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530793179463407e-06,
+      "loss": 0.7597,
+      "step": 3603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.525573734122738e-06,
+      "loss": 0.7623,
+      "step": 3604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.520356417519104e-06,
+      "loss": 0.8297,
+      "step": 3605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5151412316812656e-06,
+      "loss": 0.8305,
+      "step": 3606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.509928178637149e-06,
+      "loss": 0.8481,
+      "step": 3607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504717260413863e-06,
+      "loss": 0.7652,
+      "step": 3608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.49950847903767e-06,
+      "loss": 0.786,
+      "step": 3609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.494301836534016e-06,
+      "loss": 0.8405,
+      "step": 3610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.489097334927514e-06,
+      "loss": 0.7755,
+      "step": 3611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.483894976241933e-06,
+      "loss": 0.3253,
+      "step": 3612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.478694762500222e-06,
+      "loss": 0.7888,
+      "step": 3613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.473496695724484e-06,
+      "loss": 0.7866,
+      "step": 3614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4683007779359975e-06,
+      "loss": 0.8189,
+      "step": 3615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463107011155204e-06,
+      "loss": 0.7843,
+      "step": 3616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4579153974016986e-06,
+      "loss": 0.8137,
+      "step": 3617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4527259386942515e-06,
+      "loss": 0.8085,
+      "step": 3618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.447538637050787e-06,
+      "loss": 0.316,
+      "step": 3619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4423534944883984e-06,
+      "loss": 0.7728,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.437170513023328e-06,
+      "loss": 0.8089,
+      "step": 3621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4319896946709875e-06,
+      "loss": 0.7577,
+      "step": 3622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.426811041445946e-06,
+      "loss": 0.8019,
+      "step": 3623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.421634555361923e-06,
+      "loss": 0.8511,
+      "step": 3624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416460238431809e-06,
+      "loss": 0.8109,
+      "step": 3625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.411288092667636e-06,
+      "loss": 0.3357,
+      "step": 3626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4061181200806e-06,
+      "loss": 0.7616,
+      "step": 3627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4009503226810576e-06,
+      "loss": 0.7746,
+      "step": 3628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.395784702478504e-06,
+      "loss": 0.7879,
+      "step": 3629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.390621261481603e-06,
+      "loss": 0.8232,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38546000169816e-06,
+      "loss": 0.8142,
+      "step": 3631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.380300925135138e-06,
+      "loss": 0.8192,
+      "step": 3632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.375144033798655e-06,
+      "loss": 0.7783,
+      "step": 3633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.369989329693967e-06,
+      "loss": 0.8209,
+      "step": 3634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.364836814825493e-06,
+      "loss": 0.8373,
+      "step": 3635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359686491196789e-06,
+      "loss": 0.833,
+      "step": 3636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.354538360810564e-06,
+      "loss": 0.8071,
+      "step": 3637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349392425668683e-06,
+      "loss": 0.7849,
+      "step": 3638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.344248687772138e-06,
+      "loss": 0.8032,
+      "step": 3639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3391071491210834e-06,
+      "loss": 0.8688,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333967811714811e-06,
+      "loss": 0.3225,
+      "step": 3641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.328830677551763e-06,
+      "loss": 0.8723,
+      "step": 3642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3236957486295115e-06,
+      "loss": 0.7346,
+      "step": 3643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318563026944784e-06,
+      "loss": 0.7976,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3134325144934484e-06,
+      "loss": 0.8176,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.308304213270504e-06,
+      "loss": 0.8432,
+      "step": 3646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.303178125270104e-06,
+      "loss": 0.7886,
+      "step": 3647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.298054252485525e-06,
+      "loss": 0.8108,
+      "step": 3648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292932596909199e-06,
+      "loss": 0.7909,
+      "step": 3649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.287813160532682e-06,
+      "loss": 0.8376,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282695945346674e-06,
+      "loss": 0.783,
+      "step": 3651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277580953341015e-06,
+      "loss": 0.7791,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272468186504668e-06,
+      "loss": 0.735,
+      "step": 3653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.267357646825746e-06,
+      "loss": 0.8142,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262249336291481e-06,
+      "loss": 0.8115,
+      "step": 3655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257143256888248e-06,
+      "loss": 0.824,
+      "step": 3656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2520394106015574e-06,
+      "loss": 0.7747,
+      "step": 3657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.246937799416037e-06,
+      "loss": 0.325,
+      "step": 3658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2418384253154585e-06,
+      "loss": 0.8025,
+      "step": 3659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2367412902827185e-06,
+      "loss": 0.8491,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.23164639629985e-06,
+      "loss": 0.7903,
+      "step": 3661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.226553745347999e-06,
+      "loss": 0.8067,
+      "step": 3662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.221463339407454e-06,
+      "loss": 0.8769,
+      "step": 3663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.216375180457626e-06,
+      "loss": 0.7875,
+      "step": 3664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.211289270477047e-06,
+      "loss": 0.7789,
+      "step": 3665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.206205611443386e-06,
+      "loss": 0.819,
+      "step": 3666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.201124205333421e-06,
+      "loss": 0.8059,
+      "step": 3667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.196045054123068e-06,
+      "loss": 0.8012,
+      "step": 3668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.190968159787364e-06,
+      "loss": 0.8219,
+      "step": 3669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.185893524300458e-06,
+      "loss": 0.8079,
+      "step": 3670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.180821149635634e-06,
+      "loss": 0.7736,
+      "step": 3671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.175751037765286e-06,
+      "loss": 0.8165,
+      "step": 3672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.170683190660934e-06,
+      "loss": 0.8784,
+      "step": 3673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.16561761029322e-06,
+      "loss": 0.814,
+      "step": 3674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1605542986318955e-06,
+      "loss": 0.7589,
+      "step": 3675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.155493257645842e-06,
+      "loss": 0.8367,
+      "step": 3676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.150434489303043e-06,
+      "loss": 0.8065,
+      "step": 3677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.145377995570611e-06,
+      "loss": 0.8464,
+      "step": 3678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.140323778414772e-06,
+      "loss": 0.7783,
+      "step": 3679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.135271839800857e-06,
+      "loss": 0.8133,
+      "step": 3680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130222181693323e-06,
+      "loss": 0.8337,
+      "step": 3681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.125174806055735e-06,
+      "loss": 0.8501,
+      "step": 3682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.120129714850773e-06,
+      "loss": 0.8287,
+      "step": 3683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.115086910040222e-06,
+      "loss": 0.7364,
+      "step": 3684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.110046393584985e-06,
+      "loss": 0.8077,
+      "step": 3685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.105008167445076e-06,
+      "loss": 0.7542,
+      "step": 3686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.099972233579608e-06,
+      "loss": 0.776,
+      "step": 3687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.094938593946817e-06,
+      "loss": 0.9079,
+      "step": 3688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089907250504032e-06,
+      "loss": 0.3329,
+      "step": 3689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.084878205207702e-06,
+      "loss": 0.7542,
+      "step": 3690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.079851460013377e-06,
+      "loss": 0.3311,
+      "step": 3691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074827016875711e-06,
+      "loss": 0.8325,
+      "step": 3692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069804877748468e-06,
+      "loss": 0.7843,
+      "step": 3693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064785044584505e-06,
+      "loss": 0.7917,
+      "step": 3694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0597675193357995e-06,
+      "loss": 0.8475,
+      "step": 3695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0547523039534135e-06,
+      "loss": 0.8075,
+      "step": 3696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.049739400387524e-06,
+      "loss": 0.3517,
+      "step": 3697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.044728810587406e-06,
+      "loss": 0.7912,
+      "step": 3698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039720536501428e-06,
+      "loss": 0.8669,
+      "step": 3699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034714580077066e-06,
+      "loss": 0.3674,
+      "step": 3700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.029710943260895e-06,
+      "loss": 0.7881,
+      "step": 3701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.024709627998579e-06,
+      "loss": 0.7767,
+      "step": 3702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019710636234888e-06,
+      "loss": 0.7839,
+      "step": 3703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.014713969913686e-06,
+      "loss": 0.8071,
+      "step": 3704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009719630977937e-06,
+      "loss": 0.8215,
+      "step": 3705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.004727621369688e-06,
+      "loss": 0.8099,
+      "step": 3706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999737943030088e-06,
+      "loss": 0.8367,
+      "step": 3707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994750597899389e-06,
+      "loss": 0.8525,
+      "step": 3708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989765587916914e-06,
+      "loss": 0.8491,
+      "step": 3709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984782915021098e-06,
+      "loss": 0.8321,
+      "step": 3710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979802581149453e-06,
+      "loss": 0.7927,
+      "step": 3711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.974824588238595e-06,
+      "loss": 0.7802,
+      "step": 3712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.969848938224215e-06,
+      "loss": 0.3307,
+      "step": 3713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.964875633041102e-06,
+      "loss": 0.8128,
+      "step": 3714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.959904674623137e-06,
+      "loss": 0.8476,
+      "step": 3715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954936064903276e-06,
+      "loss": 0.7599,
+      "step": 3716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949969805813575e-06,
+      "loss": 0.8551,
+      "step": 3717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945005899285163e-06,
+      "loss": 0.7817,
+      "step": 3718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940044347248263e-06,
+      "loss": 0.8558,
+      "step": 3719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935085151632185e-06,
+      "loss": 0.7767,
+      "step": 3720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9301283143653125e-06,
+      "loss": 0.8309,
+      "step": 3721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925173837375118e-06,
+      "loss": 0.7496,
+      "step": 3722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920221722588161e-06,
+      "loss": 0.8213,
+      "step": 3723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.915271971930069e-06,
+      "loss": 0.763,
+      "step": 3724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.910324587325564e-06,
+      "loss": 0.7755,
+      "step": 3725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.905379570698439e-06,
+      "loss": 0.8252,
+      "step": 3726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.900436923971576e-06,
+      "loss": 0.7933,
+      "step": 3727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895496649066919e-06,
+      "loss": 0.8169,
+      "step": 3728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.89055874790551e-06,
+      "loss": 0.8083,
+      "step": 3729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.885623222407448e-06,
+      "loss": 0.3456,
+      "step": 3730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8806900744919205e-06,
+      "loss": 0.7933,
+      "step": 3731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875759306077196e-06,
+      "loss": 0.8672,
+      "step": 3732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.870830919080598e-06,
+      "loss": 0.7846,
+      "step": 3733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.865904915418544e-06,
+      "loss": 0.8192,
+      "step": 3734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.86098129700651e-06,
+      "loss": 0.7631,
+      "step": 3735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.856060065759053e-06,
+      "loss": 0.8732,
+      "step": 3736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851141223589805e-06,
+      "loss": 0.7899,
+      "step": 3737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846224772411454e-06,
+      "loss": 0.7701,
+      "step": 3738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8413107141357765e-06,
+      "loss": 0.8709,
+      "step": 3739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836399050673601e-06,
+      "loss": 0.7819,
+      "step": 3740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.831489783934843e-06,
+      "loss": 0.7967,
+      "step": 3741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.826582915828468e-06,
+      "loss": 0.8626,
+      "step": 3742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.821678448262522e-06,
+      "loss": 0.8278,
+      "step": 3743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816776383144111e-06,
+      "loss": 0.7442,
+      "step": 3744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8118767223794105e-06,
+      "loss": 0.3164,
+      "step": 3745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8069794678736614e-06,
+      "loss": 0.8535,
+      "step": 3746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8020846215311612e-06,
+      "loss": 0.7599,
+      "step": 3747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7971921852552807e-06,
+      "loss": 0.7989,
+      "step": 3748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79230216094845e-06,
+      "loss": 0.7902,
+      "step": 3749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7874145505121563e-06,
+      "loss": 0.7865,
+      "step": 3750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.782529355846959e-06,
+      "loss": 0.7943,
+      "step": 3751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.777646578852464e-06,
+      "loss": 0.8155,
+      "step": 3752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7727662214273496e-06,
+      "loss": 0.8851,
+      "step": 3753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.767888285469351e-06,
+      "loss": 0.3282,
+      "step": 3754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.763012772875253e-06,
+      "loss": 0.3123,
+      "step": 3755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.75813968554091e-06,
+      "loss": 0.7969,
+      "step": 3756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7532690253612226e-06,
+      "loss": 0.8007,
+      "step": 3757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.748400794230158e-06,
+      "loss": 0.786,
+      "step": 3758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7435349940407274e-06,
+      "loss": 0.8514,
+      "step": 3759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7386716266850066e-06,
+      "loss": 0.7648,
+      "step": 3760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.733810694054124e-06,
+      "loss": 0.7924,
+      "step": 3761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7289521980382535e-06,
+      "loss": 0.855,
+      "step": 3762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.724096140526633e-06,
+      "loss": 0.8461,
+      "step": 3763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.719242523407539e-06,
+      "loss": 0.7369,
+      "step": 3764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7143913485683103e-06,
+      "loss": 0.8646,
+      "step": 3765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.709542617895332e-06,
+      "loss": 0.3168,
+      "step": 3766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.704696333274037e-06,
+      "loss": 0.7968,
+      "step": 3767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6998524965889136e-06,
+      "loss": 0.8031,
+      "step": 3768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6950111097234866e-06,
+      "loss": 0.769,
+      "step": 3769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6901721745603414e-06,
+      "loss": 0.8094,
+      "step": 3770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6853356929810966e-06,
+      "loss": 0.8793,
+      "step": 3771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6805016668664284e-06,
+      "loss": 0.7859,
+      "step": 3772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6756700980960548e-06,
+      "loss": 0.7571,
+      "step": 3773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6708409885487305e-06,
+      "loss": 0.8101,
+      "step": 3774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6660143401022687e-06,
+      "loss": 0.7958,
+      "step": 3775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6611901546335093e-06,
+      "loss": 0.8346,
+      "step": 3776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6563684340183458e-06,
+      "loss": 0.824,
+      "step": 3777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.651549180131714e-06,
+      "loss": 0.7886,
+      "step": 3778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6467323948475786e-06,
+      "loss": 0.8744,
+      "step": 3779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6419180800389607e-06,
+      "loss": 0.7195,
+      "step": 3780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.637106237577903e-06,
+      "loss": 0.8033,
+      "step": 3781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6322968693355008e-06,
+      "loss": 0.834,
+      "step": 3782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.627489977181887e-06,
+      "loss": 0.8139,
+      "step": 3783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.622685562986219e-06,
+      "loss": 0.7919,
+      "step": 3784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6178836286167052e-06,
+      "loss": 0.8246,
+      "step": 3785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6130841759405776e-06,
+      "loss": 0.7715,
+      "step": 3786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6082872068241124e-06,
+      "loss": 0.8132,
+      "step": 3787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6034927231326156e-06,
+      "loss": 0.8307,
+      "step": 3788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.59870072673043e-06,
+      "loss": 0.7897,
+      "step": 3789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.593911219480929e-06,
+      "loss": 0.7839,
+      "step": 3790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5891242032465144e-06,
+      "loss": 0.8607,
+      "step": 3791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5843396798886275e-06,
+      "loss": 0.7947,
+      "step": 3792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.579557651267731e-06,
+      "loss": 0.8288,
+      "step": 3793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.574778119243325e-06,
+      "loss": 0.8359,
+      "step": 3794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5700010856739385e-06,
+      "loss": 0.7854,
+      "step": 3795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5652265524171205e-06,
+      "loss": 0.795,
+      "step": 3796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5604545213294616e-06,
+      "loss": 0.7512,
+      "step": 3797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5556849942665626e-06,
+      "loss": 0.8035,
+      "step": 3798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.550917973083069e-06,
+      "loss": 0.7758,
+      "step": 3799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546153459632634e-06,
+      "loss": 0.8343,
+      "step": 3800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5413914557679486e-06,
+      "loss": 0.8456,
+      "step": 3801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5366319633407255e-06,
+      "loss": 0.8373,
+      "step": 3802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.531874984201694e-06,
+      "loss": 0.8685,
+      "step": 3803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5271205202006163e-06,
+      "loss": 0.8329,
+      "step": 3804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5223685731862657e-06,
+      "loss": 0.8535,
+      "step": 3805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5176191450064445e-06,
+      "loss": 0.7845,
+      "step": 3806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.512872237507974e-06,
+      "loss": 0.8238,
+      "step": 3807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.508127852536698e-06,
+      "loss": 0.7688,
+      "step": 3808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5033859919374693e-06,
+      "loss": 0.8442,
+      "step": 3809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4986466575541688e-06,
+      "loss": 0.7691,
+      "step": 3810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4939098512296933e-06,
+      "loss": 0.8035,
+      "step": 3811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4891755748059587e-06,
+      "loss": 0.8587,
+      "step": 3812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.484443830123887e-06,
+      "loss": 0.79,
+      "step": 3813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4797146190234286e-06,
+      "loss": 0.7943,
+      "step": 3814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4749879433435364e-06,
+      "loss": 0.7791,
+      "step": 3815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4702638049221902e-06,
+      "loss": 0.7425,
+      "step": 3816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4655422055963715e-06,
+      "loss": 0.8138,
+      "step": 3817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4608231472020815e-06,
+      "loss": 0.7661,
+      "step": 3818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4561066315743365e-06,
+      "loss": 0.8271,
+      "step": 3819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 0.8236,
+      "step": 3820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.446681235953564e-06,
+      "loss": 0.8296,
+      "step": 3821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4419723596256126e-06,
+      "loss": 0.7751,
+      "step": 3822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.437266033394354e-06,
+      "loss": 0.7854,
+      "step": 3823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.43256225908985e-06,
+      "loss": 0.8273,
+      "step": 3824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4278610385411615e-06,
+      "loss": 0.7718,
+      "step": 3825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.423162373576372e-06,
+      "loss": 0.8165,
+      "step": 3826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4184662660225567e-06,
+      "loss": 0.815,
+      "step": 3827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4137727177058023e-06,
+      "loss": 0.8418,
+      "step": 3828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.409081730451205e-06,
+      "loss": 0.8036,
+      "step": 3829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4043933060828606e-06,
+      "loss": 0.7344,
+      "step": 3830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.399707446423862e-06,
+      "loss": 0.7379,
+      "step": 3831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.395024153296317e-06,
+      "loss": 0.7878,
+      "step": 3832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.390343428521331e-06,
+      "loss": 0.8073,
+      "step": 3833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3856652739190034e-06,
+      "loss": 0.8168,
+      "step": 3834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.380989691308442e-06,
+      "loss": 0.8122,
+      "step": 3835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3763166825077576e-06,
+      "loss": 0.3436,
+      "step": 3836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3716462493340473e-06,
+      "loss": 0.8302,
+      "step": 3837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3669783936034216e-06,
+      "loss": 0.7967,
+      "step": 3838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3623131171309755e-06,
+      "loss": 0.8453,
+      "step": 3839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.357650421730809e-06,
+      "loss": 0.8293,
+      "step": 3840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.352990309216022e-06,
+      "loss": 0.7215,
+      "step": 3841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3483327813986967e-06,
+      "loss": 0.8651,
+      "step": 3842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.343677840089924e-06,
+      "loss": 0.8759,
+      "step": 3843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3390254870997796e-06,
+      "loss": 0.8319,
+      "step": 3844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3343757242373397e-06,
+      "loss": 0.8328,
+      "step": 3845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3297285533106648e-06,
+      "loss": 0.8335,
+      "step": 3846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3250839761268174e-06,
+      "loss": 0.7926,
+      "step": 3847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3204419944918476e-06,
+      "loss": 0.7385,
+      "step": 3848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.315802610210791e-06,
+      "loss": 0.7596,
+      "step": 3849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3111658250876787e-06,
+      "loss": 0.8249,
+      "step": 3850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.306531640925532e-06,
+      "loss": 0.7502,
+      "step": 3851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3019000595263573e-06,
+      "loss": 0.7213,
+      "step": 3852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.297271082691155e-06,
+      "loss": 0.8424,
+      "step": 3853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2926447122199e-06,
+      "loss": 0.7879,
+      "step": 3854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.288020949911569e-06,
+      "loss": 0.8074,
+      "step": 3855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.283399797564112e-06,
+      "loss": 0.3263,
+      "step": 3856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2787812569744716e-06,
+      "loss": 0.8233,
+      "step": 3857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2741653299385767e-06,
+      "loss": 0.8274,
+      "step": 3858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2695520182513285e-06,
+      "loss": 0.8432,
+      "step": 3859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2649413237066253e-06,
+      "loss": 0.807,
+      "step": 3860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.260333248097335e-06,
+      "loss": 0.9058,
+      "step": 3861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.25572779321532e-06,
+      "loss": 0.8095,
+      "step": 3862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2511249608514083e-06,
+      "loss": 0.7982,
+      "step": 3863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2465247527954226e-06,
+      "loss": 0.8187,
+      "step": 3864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.241927170836161e-06,
+      "loss": 0.8435,
+      "step": 3865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.237332216761392e-06,
+      "loss": 0.7802,
+      "step": 3866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2327398923578755e-06,
+      "loss": 0.84,
+      "step": 3867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2281501994113363e-06,
+      "loss": 0.7538,
+      "step": 3868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2235631397064837e-06,
+      "loss": 0.7945,
+      "step": 3869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.218978715027006e-06,
+      "loss": 0.8517,
+      "step": 3870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.214396927155555e-06,
+      "loss": 0.805,
+      "step": 3871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.209817777873767e-06,
+      "loss": 0.7813,
+      "step": 3872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.20524126896225e-06,
+      "loss": 0.7962,
+      "step": 3873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2006674022005857e-06,
+      "loss": 0.8489,
+      "step": 3874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1960961793673285e-06,
+      "loss": 0.8039,
+      "step": 3875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.19152760224e-06,
+      "loss": 0.7606,
+      "step": 3876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1869616725951012e-06,
+      "loss": 0.7468,
+      "step": 3877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1823983922080926e-06,
+      "loss": 0.8185,
+      "step": 3878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1778377628534194e-06,
+      "loss": 0.8513,
+      "step": 3879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1732797863044795e-06,
+      "loss": 0.7771,
+      "step": 3880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1687244643336514e-06,
+      "loss": 0.7919,
+      "step": 3881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.16417179871228e-06,
+      "loss": 0.8013,
+      "step": 3882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1596217912106687e-06,
+      "loss": 0.8907,
+      "step": 3883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1550744435980995e-06,
+      "loss": 0.8315,
+      "step": 3884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1505297576428074e-06,
+      "loss": 0.7873,
+      "step": 3885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.145987735112003e-06,
+      "loss": 0.7975,
+      "step": 3886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1414483777718585e-06,
+      "loss": 0.7841,
+      "step": 3887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.136911687387503e-06,
+      "loss": 0.8225,
+      "step": 3888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.132377665723039e-06,
+      "loss": 0.7314,
+      "step": 3889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1278463145415205e-06,
+      "loss": 0.7796,
+      "step": 3890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1233176356049733e-06,
+      "loss": 0.3347,
+      "step": 3891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.118791630674378e-06,
+      "loss": 0.8257,
+      "step": 3892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1142683015096797e-06,
+      "loss": 0.7645,
+      "step": 3893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.109747649869774e-06,
+      "loss": 0.7717,
+      "step": 3894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105229677512527e-06,
+      "loss": 0.8227,
+      "step": 3895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1007143861947575e-06,
+      "loss": 0.7749,
+      "step": 3896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0962017776722364e-06,
+      "loss": 0.3204,
+      "step": 3897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091691853699702e-06,
+      "loss": 0.8195,
+      "step": 3898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.087184616030844e-06,
+      "loss": 0.848,
+      "step": 3899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.082680066418301e-06,
+      "loss": 0.8388,
+      "step": 3900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0781782066136812e-06,
+      "loss": 0.851,
+      "step": 3901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0736790383675287e-06,
+      "loss": 0.7123,
+      "step": 3902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.069182563429355e-06,
+      "loss": 0.8297,
+      "step": 3903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0646887835476226e-06,
+      "loss": 0.8367,
+      "step": 3904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0601977004697357e-06,
+      "loss": 0.7951,
+      "step": 3905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055709315942066e-06,
+      "loss": 0.8717,
+      "step": 3906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0512236317099173e-06,
+      "loss": 0.8212,
+      "step": 3907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0467406495175633e-06,
+      "loss": 0.7947,
+      "step": 3908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0422603711082076e-06,
+      "loss": 0.7818,
+      "step": 3909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.037782798224015e-06,
+      "loss": 0.7782,
+      "step": 3910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0333079326060997e-06,
+      "loss": 0.8436,
+      "step": 3911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.028835775994512e-06,
+      "loss": 0.858,
+      "step": 3912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0243663301282567e-06,
+      "loss": 0.8041,
+      "step": 3913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.019899596745284e-06,
+      "loss": 0.7956,
+      "step": 3914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0154355775824916e-06,
+      "loss": 0.7817,
+      "step": 3915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.010974274375712e-06,
+      "loss": 0.7848,
+      "step": 3916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0065156888597313e-06,
+      "loss": 0.7906,
+      "step": 3917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0020598227682794e-06,
+      "loss": 0.8322,
+      "step": 3918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9976066778340186e-06,
+      "loss": 0.8179,
+      "step": 3919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.993156255788565e-06,
+      "loss": 0.7999,
+      "step": 3920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9887085583624655e-06,
+      "loss": 0.8455,
+      "step": 3921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9842635872852135e-06,
+      "loss": 0.7625,
+      "step": 3922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9798213442852466e-06,
+      "loss": 0.3353,
+      "step": 3923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9753818310899276e-06,
+      "loss": 0.8251,
+      "step": 3924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9709450494255744e-06,
+      "loss": 0.7867,
+      "step": 3925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.966511001017428e-06,
+      "loss": 0.7671,
+      "step": 3926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.962079687589676e-06,
+      "loss": 0.7978,
+      "step": 3927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9576511108654436e-06,
+      "loss": 0.8281,
+      "step": 3928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.953225272566782e-06,
+      "loss": 0.791,
+      "step": 3929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.948802174414688e-06,
+      "loss": 0.8114,
+      "step": 3930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.944381818129084e-06,
+      "loss": 0.7481,
+      "step": 3931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9399642054288336e-06,
+      "loss": 0.3312,
+      "step": 3932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9355493380317334e-06,
+      "loss": 0.8119,
+      "step": 3933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9311372176545028e-06,
+      "loss": 0.8093,
+      "step": 3934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9267278460128045e-06,
+      "loss": 0.7941,
+      "step": 3935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9223212248212275e-06,
+      "loss": 0.7724,
+      "step": 3936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.917917355793294e-06,
+      "loss": 0.879,
+      "step": 3937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9135162406414476e-06,
+      "loss": 0.3016,
+      "step": 3938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.909117881077069e-06,
+      "loss": 0.8159,
+      "step": 3939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9047222788104712e-06,
+      "loss": 0.7927,
+      "step": 3940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9003294355508816e-06,
+      "loss": 0.8085,
+      "step": 3941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8959393530064696e-06,
+      "loss": 0.7091,
+      "step": 3942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.891552032884315e-06,
+      "loss": 0.7811,
+      "step": 3943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8871674768904377e-06,
+      "loss": 0.7897,
+      "step": 3944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.882785686729781e-06,
+      "loss": 0.7855,
+      "step": 3945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8784066641062013e-06,
+      "loss": 0.8502,
+      "step": 3946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8740304107224936e-06,
+      "loss": 0.3001,
+      "step": 3947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8696569282803622e-06,
+      "loss": 0.7698,
+      "step": 3948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8652862184804452e-06,
+      "loss": 0.8297,
+      "step": 3949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8609182830223015e-06,
+      "loss": 0.8359,
+      "step": 3950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8565531236043997e-06,
+      "loss": 0.7377,
+      "step": 3951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8521907419241445e-06,
+      "loss": 0.7738,
+      "step": 3952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.847831139677848e-06,
+      "loss": 0.7892,
+      "step": 3953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8434743185607483e-06,
+      "loss": 0.7841,
+      "step": 3954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8391202802670025e-06,
+      "loss": 0.8401,
+      "step": 3955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.83476902648968e-06,
+      "loss": 0.3096,
+      "step": 3956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.830420558920772e-06,
+      "loss": 0.7852,
+      "step": 3957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.826074879251186e-06,
+      "loss": 0.7629,
+      "step": 3958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8217319891707474e-06,
+      "loss": 0.7739,
+      "step": 3959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8173918903681874e-06,
+      "loss": 0.8304,
+      "step": 3960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.813054584531163e-06,
+      "loss": 0.7683,
+      "step": 3961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8087200733462427e-06,
+      "loss": 0.791,
+      "step": 3962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8043883584988996e-06,
+      "loss": 0.8076,
+      "step": 3963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.800059441673533e-06,
+      "loss": 0.8118,
+      "step": 3964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.795733324553441e-06,
+      "loss": 0.8017,
+      "step": 3965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.791410008820845e-06,
+      "loss": 0.8251,
+      "step": 3966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7870894961568662e-06,
+      "loss": 0.8229,
+      "step": 3967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.782771788241543e-06,
+      "loss": 0.8314,
+      "step": 3968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.778456886753824e-06,
+      "loss": 0.3021,
+      "step": 3969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7741447933715595e-06,
+      "loss": 0.8161,
+      "step": 3970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7698355097715167e-06,
+      "loss": 0.7678,
+      "step": 3971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.765529037629359e-06,
+      "loss": 0.8112,
+      "step": 3972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7612253786196663e-06,
+      "loss": 0.7758,
+      "step": 3973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.756924534415926e-06,
+      "loss": 0.8349,
+      "step": 3974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7526265066905177e-06,
+      "loss": 0.811,
+      "step": 3975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7483312971147403e-06,
+      "loss": 0.7745,
+      "step": 3976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7440389073587927e-06,
+      "loss": 0.8425,
+      "step": 3977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7397493390917694e-06,
+      "loss": 0.7547,
+      "step": 3978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.735462593981677e-06,
+      "loss": 0.8156,
+      "step": 3979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7311786736954228e-06,
+      "loss": 0.7751,
+      "step": 3980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.726897579898815e-06,
+      "loss": 0.835,
+      "step": 3981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.722619314256557e-06,
+      "loss": 0.8337,
+      "step": 3982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7183438784322646e-06,
+      "loss": 0.7885,
+      "step": 3983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.714071274088438e-06,
+      "loss": 0.8122,
+      "step": 3984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7098015028864878e-06,
+      "loss": 0.8058,
+      "step": 3985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.705534566486723e-06,
+      "loss": 0.776,
+      "step": 3986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7012704665483413e-06,
+      "loss": 0.797,
+      "step": 3987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.697009204729447e-06,
+      "loss": 0.7544,
+      "step": 3988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6927507826870338e-06,
+      "loss": 0.8887,
+      "step": 3989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.688495202076994e-06,
+      "loss": 0.7721,
+      "step": 3990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.68424246455412e-06,
+      "loss": 0.7982,
+      "step": 3991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6799925717720866e-06,
+      "loss": 0.8349,
+      "step": 3992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6757455253834763e-06,
+      "loss": 0.7839,
+      "step": 3993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6715013270397517e-06,
+      "loss": 0.8367,
+      "step": 3994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.667259978391281e-06,
+      "loss": 0.7909,
+      "step": 3995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.66302148108731e-06,
+      "loss": 0.7957,
+      "step": 3996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6587858367759878e-06,
+      "loss": 0.8119,
+      "step": 3997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.654553047104349e-06,
+      "loss": 0.7952,
+      "step": 3998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6503231137183192e-06,
+      "loss": 0.75,
+      "step": 3999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6460960382627155e-06,
+      "loss": 0.7887,
+      "step": 4000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6418718223812357e-06,
+      "loss": 0.7888,
+      "step": 4001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6376504677164727e-06,
+      "loss": 0.8197,
+      "step": 4002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.633431975909909e-06,
+      "loss": 0.8214,
+      "step": 4003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.629216348601905e-06,
+      "loss": 0.7295,
+      "step": 4004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.625003587431717e-06,
+      "loss": 0.3229,
+      "step": 4005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6207936940374767e-06,
+      "loss": 0.7805,
+      "step": 4006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.616586670056208e-06,
+      "loss": 0.8912,
+      "step": 4007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6123825171238216e-06,
+      "loss": 0.7502,
+      "step": 4008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6081812368751015e-06,
+      "loss": 0.8146,
+      "step": 4009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6039828309437245e-06,
+      "loss": 0.7854,
+      "step": 4010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.599787300962241e-06,
+      "loss": 0.8356,
+      "step": 4011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.595594648562093e-06,
+      "loss": 0.7697,
+      "step": 4012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.591404875373593e-06,
+      "loss": 0.8524,
+      "step": 4013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5872179830259413e-06,
+      "loss": 0.8316,
+      "step": 4014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.583033973147221e-06,
+      "loss": 0.8597,
+      "step": 4015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.57885284736438e-06,
+      "loss": 0.7659,
+      "step": 4016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5746746073032624e-06,
+      "loss": 0.8057,
+      "step": 4017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5704992545885755e-06,
+      "loss": 0.7802,
+      "step": 4018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5663267908439115e-06,
+      "loss": 0.803,
+      "step": 4019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5621572176917387e-06,
+      "loss": 0.8272,
+      "step": 4020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.557990536753401e-06,
+      "loss": 0.8403,
+      "step": 4021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55382674964912e-06,
+      "loss": 0.8117,
+      "step": 4022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5496658579979815e-06,
+      "loss": 0.7656,
+      "step": 4023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5455078634179585e-06,
+      "loss": 0.7817,
+      "step": 4024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.541352767525894e-06,
+      "loss": 0.8542,
+      "step": 4025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5372005719374957e-06,
+      "loss": 0.3534,
+      "step": 4026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5330512782673554e-06,
+      "loss": 0.8415,
+      "step": 4027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5289048881289256e-06,
+      "loss": 0.785,
+      "step": 4028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5247614031345412e-06,
+      "loss": 0.8155,
+      "step": 4029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5206208248953932e-06,
+      "loss": 0.3378,
+      "step": 4030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5164831550215552e-06,
+      "loss": 0.8462,
+      "step": 4031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.512348395121967e-06,
+      "loss": 0.8405,
+      "step": 4032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5082165468044306e-06,
+      "loss": 0.755,
+      "step": 4033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.504087611675623e-06,
+      "loss": 0.8306,
+      "step": 4034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4999615913410823e-06,
+      "loss": 0.7307,
+      "step": 4035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4958384874052166e-06,
+      "loss": 0.3276,
+      "step": 4036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4917183014713052e-06,
+      "loss": 0.8516,
+      "step": 4037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4876010351414792e-06,
+      "loss": 0.8526,
+      "step": 4038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4834866900167478e-06,
+      "loss": 0.7727,
+      "step": 4039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4793752676969795e-06,
+      "loss": 0.7835,
+      "step": 4040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4752667697809007e-06,
+      "loss": 0.77,
+      "step": 4041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4711611978661076e-06,
+      "loss": 0.8612,
+      "step": 4042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4670585535490586e-06,
+      "loss": 0.8752,
+      "step": 4043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4629588384250737e-06,
+      "loss": 0.7681,
+      "step": 4044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4588620540883257e-06,
+      "loss": 0.8408,
+      "step": 4045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4547682021318596e-06,
+      "loss": 0.8261,
+      "step": 4046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.450677284147569e-06,
+      "loss": 0.7532,
+      "step": 4047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4465893017262156e-06,
+      "loss": 0.829,
+      "step": 4048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4425042564574186e-06,
+      "loss": 0.7568,
+      "step": 4049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4384221499296466e-06,
+      "loss": 0.8044,
+      "step": 4050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4343429837302367e-06,
+      "loss": 0.78,
+      "step": 4051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4302667594453734e-06,
+      "loss": 0.7621,
+      "step": 4052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4261934786601025e-06,
+      "loss": 0.7872,
+      "step": 4053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.422123142958328e-06,
+      "loss": 0.7577,
+      "step": 4054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.418055753922799e-06,
+      "loss": 0.8189,
+      "step": 4055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.413991313135129e-06,
+      "loss": 0.7804,
+      "step": 4056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.409929822175775e-06,
+      "loss": 0.8039,
+      "step": 4057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4058712826240595e-06,
+      "loss": 0.8253,
+      "step": 4058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.401815696058143e-06,
+      "loss": 0.8487,
+      "step": 4059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.39776306405505e-06,
+      "loss": 0.827,
+      "step": 4060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3937133881906484e-06,
+      "loss": 0.8545,
+      "step": 4061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.389666670039664e-06,
+      "loss": 0.7621,
+      "step": 4062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3856229111756622e-06,
+      "loss": 0.8021,
+      "step": 4063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.381582113171066e-06,
+      "loss": 0.8447,
+      "step": 4064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3775442775971448e-06,
+      "loss": 0.7853,
+      "step": 4065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.373509406024017e-06,
+      "loss": 0.8527,
+      "step": 4066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3694775000206427e-06,
+      "loss": 0.8013,
+      "step": 4067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3654485611548394e-06,
+      "loss": 0.3008,
+      "step": 4068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3614225909932575e-06,
+      "loss": 0.7805,
+      "step": 4069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.357399591101404e-06,
+      "loss": 0.3504,
+      "step": 4070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3533795630436297e-06,
+      "loss": 0.7908,
+      "step": 4071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3493625083831217e-06,
+      "loss": 0.8434,
+      "step": 4072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.345348428681922e-06,
+      "loss": 0.7942,
+      "step": 4073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.341337325500904e-06,
+      "loss": 0.7513,
+      "step": 4074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3373292003997963e-06,
+      "loss": 0.2983,
+      "step": 4075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3333240549371572e-06,
+      "loss": 0.7848,
+      "step": 4076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.329321890670394e-06,
+      "loss": 0.809,
+      "step": 4077
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.325322709155757e-06,
+      "loss": 0.8052,
+      "step": 4078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3213265119483265e-06,
+      "loss": 0.8467,
+      "step": 4079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3173333006020337e-06,
+      "loss": 0.3213,
+      "step": 4080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3133430766696383e-06,
+      "loss": 0.7916,
+      "step": 4081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3093558417027473e-06,
+      "loss": 0.7884,
+      "step": 4082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3053715972517998e-06,
+      "loss": 0.8881,
+      "step": 4083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3013903448660745e-06,
+      "loss": 0.6912,
+      "step": 4084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.297412086093689e-06,
+      "loss": 0.7713,
+      "step": 4085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2934368224815894e-06,
+      "loss": 0.8084,
+      "step": 4086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2894645555755645e-06,
+      "loss": 0.8426,
+      "step": 4087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.285495286920232e-06,
+      "loss": 0.8561,
+      "step": 4088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.281529018059049e-06,
+      "loss": 0.8034,
+      "step": 4089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.277565750534305e-06,
+      "loss": 0.8247,
+      "step": 4090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.273605485887116e-06,
+      "loss": 0.7764,
+      "step": 4091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2696482256574426e-06,
+      "loss": 0.7288,
+      "step": 4092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.265693971384063e-06,
+      "loss": 0.8295,
+      "step": 4093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2617427246045976e-06,
+      "loss": 0.7401,
+      "step": 4094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2577944868554947e-06,
+      "loss": 0.7925,
+      "step": 4095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2538492596720275e-06,
+      "loss": 0.8014,
+      "step": 4096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2499070445883054e-06,
+      "loss": 0.8044,
+      "step": 4097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.245967843137259e-06,
+      "loss": 0.8198,
+      "step": 4098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.242031656850654e-06,
+      "loss": 0.7891,
+      "step": 4099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.238098487259084e-06,
+      "loss": 0.8331,
+      "step": 4100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.234168335891961e-06,
+      "loss": 0.8078,
+      "step": 4101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2302412042775344e-06,
+      "loss": 0.7903,
+      "step": 4102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.226317093942868e-06,
+      "loss": 0.7319,
+      "step": 4103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2223960064138605e-06,
+      "loss": 0.7766,
+      "step": 4104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.218477943215229e-06,
+      "loss": 0.7381,
+      "step": 4105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.214562905870521e-06,
+      "loss": 0.8513,
+      "step": 4106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.210650895902102e-06,
+      "loss": 0.7962,
+      "step": 4107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2067419148311585e-06,
+      "loss": 0.7848,
+      "step": 4108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2028359641777066e-06,
+      "loss": 0.7974,
+      "step": 4109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.198933045460575e-06,
+      "loss": 0.8454,
+      "step": 4110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1950331601974207e-06,
+      "loss": 0.7793,
+      "step": 4111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1911363099047222e-06,
+      "loss": 0.8256,
+      "step": 4112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1872424960977667e-06,
+      "loss": 0.8412,
+      "step": 4113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.183351720290676e-06,
+      "loss": 0.789,
+      "step": 4114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1794639839963762e-06,
+      "loss": 0.7141,
+      "step": 4115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1755792887266236e-06,
+      "loss": 0.8201,
+      "step": 4116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1716976359919818e-06,
+      "loss": 0.8513,
+      "step": 4117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1678190273018376e-06,
+      "loss": 0.8006,
+      "step": 4118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1639434641643954e-06,
+      "loss": 0.7997,
+      "step": 4119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1600709480866667e-06,
+      "loss": 0.7731,
+      "step": 4120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1562014805744913e-06,
+      "loss": 0.7713,
+      "step": 4121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.152335063132508e-06,
+      "loss": 0.8452,
+      "step": 4122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1484716972641828e-06,
+      "loss": 0.7367,
+      "step": 4123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.144611384471792e-06,
+      "loss": 0.8035,
+      "step": 4124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1407541262564157e-06,
+      "loss": 0.7265,
+      "step": 4125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1368999241179578e-06,
+      "loss": 0.7573,
+      "step": 4126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.133048779555129e-06,
+      "loss": 0.7801,
+      "step": 4127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.129200694065452e-06,
+      "loss": 0.791,
+      "step": 4128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1253556691452603e-06,
+      "loss": 0.8462,
+      "step": 4129
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.121513706289692e-06,
+      "loss": 0.8015,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1176748069927035e-06,
+      "loss": 0.8305,
+      "step": 4131
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1138389727470508e-06,
+      "loss": 0.3211,
+      "step": 4132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.110006205044307e-06,
+      "loss": 0.7635,
+      "step": 4133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.106176505374844e-06,
+      "loss": 0.7929,
+      "step": 4134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.102349875227847e-06,
+      "loss": 0.7909,
+      "step": 4135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.098526316091308e-06,
+      "loss": 0.7857,
+      "step": 4136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.094705829452018e-06,
+      "loss": 0.7899,
+      "step": 4137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.090888416795582e-06,
+      "loss": 0.7701,
+      "step": 4138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0870740796064015e-06,
+      "loss": 0.8197,
+      "step": 4139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0832628193676864e-06,
+      "loss": 0.8455,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0794546375614543e-06,
+      "loss": 0.3241,
+      "step": 4141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0756495356685147e-06,
+      "loss": 0.8051,
+      "step": 4142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0718475151684913e-06,
+      "loss": 0.791,
+      "step": 4143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0680485775397983e-06,
+      "loss": 0.8469,
+      "step": 4144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.064252724259662e-06,
+      "loss": 0.8061,
+      "step": 4145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.060459956804103e-06,
+      "loss": 0.7618,
+      "step": 4146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0566702766479454e-06,
+      "loss": 0.7905,
+      "step": 4147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.052883685264806e-06,
+      "loss": 0.6819,
+      "step": 4148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0491001841271073e-06,
+      "loss": 0.8686,
+      "step": 4149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0453197747060728e-06,
+      "loss": 0.7993,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0415424584717113e-06,
+      "loss": 0.7589,
+      "step": 4151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0377682368928407e-06,
+      "loss": 0.7551,
+      "step": 4152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.033997111437074e-06,
+      "loss": 0.7927,
+      "step": 4153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0302290835708115e-06,
+      "loss": 0.8311,
+      "step": 4154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.026464154759261e-06,
+      "loss": 0.7784,
+      "step": 4155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.022702326466416e-06,
+      "loss": 0.8718,
+      "step": 4156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0189436001550668e-06,
+      "loss": 0.8514,
+      "step": 4157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.015187977286804e-06,
+      "loss": 0.7936,
+      "step": 4158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0114354593220007e-06,
+      "loss": 0.776,
+      "step": 4159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.007686047719831e-06,
+      "loss": 0.8561,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0039397439382544e-06,
+      "loss": 0.8271,
+      "step": 4161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0001965494340303e-06,
+      "loss": 0.8064,
+      "step": 4162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9964564656626993e-06,
+      "loss": 0.8502,
+      "step": 4163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9927194940785998e-06,
+      "loss": 0.8024,
+      "step": 4164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9889856361348603e-06,
+      "loss": 0.8256,
+      "step": 4165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9852548932833893e-06,
+      "loss": 0.81,
+      "step": 4166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.981527266974895e-06,
+      "loss": 0.784,
+      "step": 4167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977802758658869e-06,
+      "loss": 0.7437,
+      "step": 4168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9740813697835914e-06,
+      "loss": 0.3387,
+      "step": 4169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9703631017961243e-06,
+      "loss": 0.8583,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9666479561423247e-06,
+      "loss": 0.7904,
+      "step": 4171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962935934266833e-06,
+      "loss": 0.8002,
+      "step": 4172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.959227037613067e-06,
+      "loss": 0.8153,
+      "step": 4173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9555212676232396e-06,
+      "loss": 0.8427,
+      "step": 4174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9518186257383453e-06,
+      "loss": 0.823,
+      "step": 4175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9481191133981557e-06,
+      "loss": 0.8002,
+      "step": 4176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9444227320412355e-06,
+      "loss": 0.8155,
+      "step": 4177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9407294831049226e-06,
+      "loss": 0.8262,
+      "step": 4178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.937039368025346e-06,
+      "loss": 0.8086,
+      "step": 4179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.933352388237406e-06,
+      "loss": 0.7417,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.929668545174791e-06,
+      "loss": 0.3522,
+      "step": 4181
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9259878402699704e-06,
+      "loss": 0.8438,
+      "step": 4182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9223102749541856e-06,
+      "loss": 0.7557,
+      "step": 4183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9186358506574686e-06,
+      "loss": 0.7725,
+      "step": 4184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914964568808616e-06,
+      "loss": 0.7671,
+      "step": 4185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9112964308352145e-06,
+      "loss": 0.7557,
+      "step": 4186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.907631438163626e-06,
+      "loss": 0.7818,
+      "step": 4187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9039695922189817e-06,
+      "loss": 0.7864,
+      "step": 4188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.900310894425197e-06,
+      "loss": 0.7701,
+      "step": 4189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8966553462049608e-06,
+      "loss": 0.7631,
+      "step": 4190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8930029489797408e-06,
+      "loss": 0.8074,
+      "step": 4191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8893537041697707e-06,
+      "loss": 0.8308,
+      "step": 4192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8857076131940643e-06,
+      "loss": 0.8361,
+      "step": 4193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8820646774704131e-06,
+      "loss": 0.7818,
+      "step": 4194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8784248984153708e-06,
+      "loss": 0.8139,
+      "step": 4195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8747882774442749e-06,
+      "loss": 0.8247,
+      "step": 4196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8711548159712246e-06,
+      "loss": 0.8107,
+      "step": 4197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.867524515409097e-06,
+      "loss": 0.7818,
+      "step": 4198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8638973771695446e-06,
+      "loss": 0.8694,
+      "step": 4199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8602734026629753e-06,
+      "loss": 0.7563,
+      "step": 4200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8566525932985836e-06,
+      "loss": 0.8192,
+      "step": 4201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8530349504843193e-06,
+      "loss": 0.7942,
+      "step": 4202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8494204756269097e-06,
+      "loss": 0.7109,
+      "step": 4203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8458091701318504e-06,
+      "loss": 0.758,
+      "step": 4204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8422010354033958e-06,
+      "loss": 0.8347,
+      "step": 4205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8385960728445795e-06,
+      "loss": 0.8453,
+      "step": 4206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8349942838571898e-06,
+      "loss": 0.8181,
+      "step": 4207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.831395669841789e-06,
+      "loss": 0.8024,
+      "step": 4208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.827800232197705e-06,
+      "loss": 0.7867,
+      "step": 4209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8242079723230232e-06,
+      "loss": 0.8017,
+      "step": 4210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8206188916145995e-06,
+      "loss": 0.7505,
+      "step": 4211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8170329914680529e-06,
+      "loss": 0.7895,
+      "step": 4212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.813450273277767e-06,
+      "loss": 0.8111,
+      "step": 4213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8098707384368818e-06,
+      "loss": 0.8528,
+      "step": 4214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8062943883373052e-06,
+      "loss": 0.7822,
+      "step": 4215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8027212243697079e-06,
+      "loss": 0.8159,
+      "step": 4216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.799151247923514e-06,
+      "loss": 0.7883,
+      "step": 4217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7955844603869176e-06,
+      "loss": 0.8194,
+      "step": 4218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7920208631468638e-06,
+      "loss": 0.7375,
+      "step": 4219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7884604575890641e-06,
+      "loss": 0.7216,
+      "step": 4220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7849032450979876e-06,
+      "loss": 0.3215,
+      "step": 4221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7813492270568566e-06,
+      "loss": 0.7544,
+      "step": 4222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7777984048476605e-06,
+      "loss": 0.7944,
+      "step": 4223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7742507798511345e-06,
+      "loss": 0.3355,
+      "step": 4224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7707063534467816e-06,
+      "loss": 0.7835,
+      "step": 4225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7671651270128531e-06,
+      "loss": 0.8118,
+      "step": 4226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7636271019263596e-06,
+      "loss": 0.749,
+      "step": 4227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7600922795630693e-06,
+      "loss": 0.7927,
+      "step": 4228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7565606612974983e-06,
+      "loss": 0.8122,
+      "step": 4229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7530322485029206e-06,
+      "loss": 0.7667,
+      "step": 4230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.749507042551365e-06,
+      "loss": 0.8361,
+      "step": 4231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.745985044813615e-06,
+      "loss": 0.8002,
+      "step": 4232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7424662566591967e-06,
+      "loss": 0.796,
+      "step": 4233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7389506794563993e-06,
+      "loss": 0.7744,
+      "step": 4234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7354383145722608e-06,
+      "loss": 0.7883,
+      "step": 4235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7319291633725633e-06,
+      "loss": 0.8446,
+      "step": 4236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7284232272218503e-06,
+      "loss": 0.7535,
+      "step": 4237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7249205074834019e-06,
+      "loss": 0.3542,
+      "step": 4238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7214210055192604e-06,
+      "loss": 0.7969,
+      "step": 4239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7179247226902107e-06,
+      "loss": 0.7707,
+      "step": 4240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7144316603557832e-06,
+      "loss": 0.7234,
+      "step": 4241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7109418198742644e-06,
+      "loss": 0.8131,
+      "step": 4242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7074552026026759e-06,
+      "loss": 0.739,
+      "step": 4243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7039718098967983e-06,
+      "loss": 0.8692,
+      "step": 4244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7004916431111529e-06,
+      "loss": 0.8001,
+      "step": 4245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6970147035990025e-06,
+      "loss": 0.7842,
+      "step": 4246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6935409927123647e-06,
+      "loss": 0.7842,
+      "step": 4247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.69007051180199e-06,
+      "loss": 0.7729,
+      "step": 4248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6866032622173822e-06,
+      "loss": 0.8023,
+      "step": 4249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6831392453067885e-06,
+      "loss": 0.7777,
+      "step": 4250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6796784624171891e-06,
+      "loss": 0.3149,
+      "step": 4251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6762209148943164e-06,
+      "loss": 0.8683,
+      "step": 4252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.672766604082643e-06,
+      "loss": 0.8388,
+      "step": 4253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6693155313253828e-06,
+      "loss": 0.8053,
+      "step": 4254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6658676979644862e-06,
+      "loss": 0.7351,
+      "step": 4255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.662423105340647e-06,
+      "loss": 0.8661,
+      "step": 4256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.658981754793304e-06,
+      "loss": 0.7622,
+      "step": 4257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6555436476606246e-06,
+      "loss": 0.3201,
+      "step": 4258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.652108785279526e-06,
+      "loss": 0.3171,
+      "step": 4259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6486771689856528e-06,
+      "loss": 0.8189,
+      "step": 4260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.645248800113396e-06,
+      "loss": 0.8337,
+      "step": 4261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.641823679995883e-06,
+      "loss": 0.7914,
+      "step": 4262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6384018099649712e-06,
+      "loss": 0.849,
+      "step": 4263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6349831913512626e-06,
+      "loss": 0.7564,
+      "step": 4264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6315678254840883e-06,
+      "loss": 0.7907,
+      "step": 4265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6281557136915172e-06,
+      "loss": 0.7817,
+      "step": 4266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6247468573003567e-06,
+      "loss": 0.8315,
+      "step": 4267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6213412576361398e-06,
+      "loss": 0.8278,
+      "step": 4268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.617938916023143e-06,
+      "loss": 0.7652,
+      "step": 4269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6145398337843654e-06,
+      "loss": 0.7879,
+      "step": 4270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.611144012241549e-06,
+      "loss": 0.7923,
+      "step": 4271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6077514527151573e-06,
+      "loss": 0.843,
+      "step": 4272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.604362156524395e-06,
+      "loss": 0.7928,
+      "step": 4273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6009761249871936e-06,
+      "loss": 0.7972,
+      "step": 4274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5975933594202142e-06,
+      "loss": 0.8253,
+      "step": 4275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5942138611388536e-06,
+      "loss": 0.7789,
+      "step": 4276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.590837631457226e-06,
+      "loss": 0.7742,
+      "step": 4277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 0.796,
+      "step": 4278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5840949831433172e-06,
+      "loss": 0.8125,
+      "step": 4279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5807285671329197e-06,
+      "loss": 0.773,
+      "step": 4280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5773654249660342e-06,
+      "loss": 0.8401,
+      "step": 4281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5740055579504176e-06,
+      "loss": 0.8137,
+      "step": 4282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5706489673925618e-06,
+      "loss": 0.7942,
+      "step": 4283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.567295654597677e-06,
+      "loss": 0.8356,
+      "step": 4284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5639456208697058e-06,
+      "loss": 0.8419,
+      "step": 4285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5605988675113137e-06,
+      "loss": 0.7597,
+      "step": 4286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5572553958238867e-06,
+      "loss": 0.7692,
+      "step": 4287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5539152071075402e-06,
+      "loss": 0.8555,
+      "step": 4288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5505783026611076e-06,
+      "loss": 0.7561,
+      "step": 4289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5472446837821487e-06,
+      "loss": 0.8264,
+      "step": 4290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5439143517669496e-06,
+      "loss": 0.753,
+      "step": 4291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5405873079105083e-06,
+      "loss": 0.8916,
+      "step": 4292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5372635535065506e-06,
+      "loss": 0.7917,
+      "step": 4293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.533943089847526e-06,
+      "loss": 0.8524,
+      "step": 4294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5306259182245963e-06,
+      "loss": 0.7625,
+      "step": 4295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5273120399276486e-06,
+      "loss": 0.7972,
+      "step": 4296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5240014562452888e-06,
+      "loss": 0.803,
+      "step": 4297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5206941684648446e-06,
+      "loss": 0.7322,
+      "step": 4298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5173901778723532e-06,
+      "loss": 0.8193,
+      "step": 4299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.514089485752579e-06,
+      "loss": 0.7934,
+      "step": 4300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5107920933889976e-06,
+      "loss": 0.7986,
+      "step": 4301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.507498002063803e-06,
+      "loss": 0.7739,
+      "step": 4302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5042072130579122e-06,
+      "loss": 0.8044,
+      "step": 4303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5009197276509446e-06,
+      "loss": 0.7625,
+      "step": 4304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.49763554712125e-06,
+      "loss": 0.8002,
+      "step": 4305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4943546727458802e-06,
+      "loss": 0.8607,
+      "step": 4306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.491077105800608e-06,
+      "loss": 0.8339,
+      "step": 4307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4878028475599237e-06,
+      "loss": 0.7859,
+      "step": 4308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4845318992970203e-06,
+      "loss": 0.8037,
+      "step": 4309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4812642622838148e-06,
+      "loss": 0.8185,
+      "step": 4310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4779999377909271e-06,
+      "loss": 0.821,
+      "step": 4311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4747389270876977e-06,
+      "loss": 0.827,
+      "step": 4312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4714812314421711e-06,
+      "loss": 0.7847,
+      "step": 4313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4682268521211075e-06,
+      "loss": 0.7511,
+      "step": 4314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4649757903899752e-06,
+      "loss": 0.6948,
+      "step": 4315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.461728047512958e-06,
+      "loss": 0.7338,
+      "step": 4316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4584836247529377e-06,
+      "loss": 0.8925,
+      "step": 4317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.455242523371515e-06,
+      "loss": 0.7977,
+      "step": 4318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4520047446289964e-06,
+      "loss": 0.7615,
+      "step": 4319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4487702897843969e-06,
+      "loss": 0.7716,
+      "step": 4320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.445539160095436e-06,
+      "loss": 0.8237,
+      "step": 4321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4423113568185433e-06,
+      "loss": 0.799,
+      "step": 4322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4390868812088521e-06,
+      "loss": 0.8041,
+      "step": 4323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4358657345202043e-06,
+      "loss": 0.817,
+      "step": 4324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.43264791800515e-06,
+      "loss": 0.8299,
+      "step": 4325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.429433432914935e-06,
+      "loss": 0.7988,
+      "step": 4326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4262222804995217e-06,
+      "loss": 0.7706,
+      "step": 4327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4230144620075648e-06,
+      "loss": 0.7573,
+      "step": 4328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4198099786864327e-06,
+      "loss": 0.8087,
+      "step": 4329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.416608831782188e-06,
+      "loss": 0.7675,
+      "step": 4330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4134110225396024e-06,
+      "loss": 0.8198,
+      "step": 4331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4102165522021516e-06,
+      "loss": 0.8249,
+      "step": 4332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4070254220120039e-06,
+      "loss": 0.8038,
+      "step": 4333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.403837633210038e-06,
+      "loss": 0.7771,
+      "step": 4334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.400653187035824e-06,
+      "loss": 0.7779,
+      "step": 4335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3974720847276412e-06,
+      "loss": 0.7982,
+      "step": 4336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3942943275224641e-06,
+      "loss": 0.779,
+      "step": 4337
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3911199166559675e-06,
+      "loss": 0.7927,
+      "step": 4338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3879488533625286e-06,
+      "loss": 0.8046,
+      "step": 4339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3847811388752119e-06,
+      "loss": 0.8405,
+      "step": 4340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3816167744257902e-06,
+      "loss": 0.8455,
+      "step": 4341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3784557612447325e-06,
+      "loss": 0.8138,
+      "step": 4342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3752981005611988e-06,
+      "loss": 0.7989,
+      "step": 4343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3721437936030524e-06,
+      "loss": 0.7285,
+      "step": 4344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3689928415968456e-06,
+      "loss": 0.7763,
+      "step": 4345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.365845245767835e-06,
+      "loss": 0.7725,
+      "step": 4346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3627010073399605e-06,
+      "loss": 0.7657,
+      "step": 4347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.359560127535865e-06,
+      "loss": 0.7879,
+      "step": 4348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3564226075768883e-06,
+      "loss": 0.8035,
+      "step": 4349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.353288448683051e-06,
+      "loss": 0.826,
+      "step": 4350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3501576520730809e-06,
+      "loss": 0.798,
+      "step": 4351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.347030218964387e-06,
+      "loss": 0.7867,
+      "step": 4352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3439061505730777e-06,
+      "loss": 0.8321,
+      "step": 4353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3407854481139514e-06,
+      "loss": 0.8143,
+      "step": 4354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3376681128004943e-06,
+      "loss": 0.689,
+      "step": 4355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3345541458448896e-06,
+      "loss": 0.783,
+      "step": 4356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3314435484580035e-06,
+      "loss": 0.8143,
+      "step": 4357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3283363218493962e-06,
+      "loss": 0.7997,
+      "step": 4358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3252324672273165e-06,
+      "loss": 0.7597,
+      "step": 4359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3221319857987024e-06,
+      "loss": 0.7984,
+      "step": 4360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.319034878769181e-06,
+      "loss": 0.8355,
+      "step": 4361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3159411473430638e-06,
+      "loss": 0.3293,
+      "step": 4362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3128507927233536e-06,
+      "loss": 0.7577,
+      "step": 4363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3097638161117365e-06,
+      "loss": 0.7459,
+      "step": 4364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.306680218708587e-06,
+      "loss": 0.8106,
+      "step": 4365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.303600001712969e-06,
+      "loss": 0.7631,
+      "step": 4366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3005231663226237e-06,
+      "loss": 0.796,
+      "step": 4367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.297449713733987e-06,
+      "loss": 0.8184,
+      "step": 4368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2943796451421686e-06,
+      "loss": 0.8194,
+      "step": 4369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2913129617409725e-06,
+      "loss": 0.8293,
+      "step": 4370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2882496647228827e-06,
+      "loss": 0.7853,
+      "step": 4371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2851897552790626e-06,
+      "loss": 0.7667,
+      "step": 4372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2821332345993654e-06,
+      "loss": 0.7808,
+      "step": 4373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2790801038723178e-06,
+      "loss": 0.843,
+      "step": 4374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.276030364285138e-06,
+      "loss": 0.8252,
+      "step": 4375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2729840170237174e-06,
+      "loss": 0.8119,
+      "step": 4376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2699410632726327e-06,
+      "loss": 0.77,
+      "step": 4377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2669015042151422e-06,
+      "loss": 0.7395,
+      "step": 4378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2638653410331781e-06,
+      "loss": 0.79,
+      "step": 4379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2608325749073591e-06,
+      "loss": 0.8546,
+      "step": 4380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2578032070169777e-06,
+      "loss": 0.8137,
+      "step": 4381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2547772385400081e-06,
+      "loss": 0.818,
+      "step": 4382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2517546706531047e-06,
+      "loss": 0.7637,
+      "step": 4383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2487355045315907e-06,
+      "loss": 0.8159,
+      "step": 4384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.245719741349477e-06,
+      "loss": 0.7357,
+      "step": 4385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2427073822794445e-06,
+      "loss": 0.7722,
+      "step": 4386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2396984284928514e-06,
+      "loss": 0.8011,
+      "step": 4387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2366928811597378e-06,
+      "loss": 0.7634,
+      "step": 4388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2336907414488075e-06,
+      "loss": 0.8314,
+      "step": 4389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2306920105274522e-06,
+      "loss": 0.8413,
+      "step": 4390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2276966895617271e-06,
+      "loss": 0.8412,
+      "step": 4391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2247047797163703e-06,
+      "loss": 0.825,
+      "step": 4392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2217162821547856e-06,
+      "loss": 0.7971,
+      "step": 4393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.218731198039056e-06,
+      "loss": 0.753,
+      "step": 4394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.215749528529936e-06,
+      "loss": 0.7633,
+      "step": 4395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.212771274786847e-06,
+      "loss": 0.8051,
+      "step": 4396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2097964379678928e-06,
+      "loss": 0.8531,
+      "step": 4397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2068250192298359e-06,
+      "loss": 0.8224,
+      "step": 4398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2038570197281185e-06,
+      "loss": 0.8069,
+      "step": 4399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.200892440616851e-06,
+      "loss": 0.7342,
+      "step": 4400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1979312830488144e-06,
+      "loss": 0.2925,
+      "step": 4401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1949735481754565e-06,
+      "loss": 0.8759,
+      "step": 4402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1920192371468952e-06,
+      "loss": 0.7519,
+      "step": 4403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1890683511119227e-06,
+      "loss": 0.8077,
+      "step": 4404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1861208912179878e-06,
+      "loss": 0.8633,
+      "step": 4405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.183176858611218e-06,
+      "loss": 0.7906,
+      "step": 4406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1802362544364032e-06,
+      "loss": 0.7786,
+      "step": 4407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1772990798369987e-06,
+      "loss": 0.8349,
+      "step": 4408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1743653359551322e-06,
+      "loss": 0.8049,
+      "step": 4409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1714350239315897e-06,
+      "loss": 0.7647,
+      "step": 4410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1685081449058266e-06,
+      "loss": 0.787,
+      "step": 4411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.165584700015967e-06,
+      "loss": 0.8367,
+      "step": 4412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1626646903987904e-06,
+      "loss": 0.7994,
+      "step": 4413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1597481171897517e-06,
+      "loss": 0.7903,
+      "step": 4414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1568349815229584e-06,
+      "loss": 0.8303,
+      "step": 4415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1539252845311889e-06,
+      "loss": 0.7551,
+      "step": 4416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1510190273458854e-06,
+      "loss": 0.8016,
+      "step": 4417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1481162110971444e-06,
+      "loss": 0.8135,
+      "step": 4418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.145216836913733e-06,
+      "loss": 0.8626,
+      "step": 4419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1423209059230721e-06,
+      "loss": 0.7531,
+      "step": 4420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1394284192512517e-06,
+      "loss": 0.8546,
+      "step": 4421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1365393780230172e-06,
+      "loss": 0.7723,
+      "step": 4422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1336537833617777e-06,
+      "loss": 0.7978,
+      "step": 4423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.130771636389596e-06,
+      "loss": 0.7515,
+      "step": 4424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1278929382272007e-06,
+      "loss": 0.8122,
+      "step": 4425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1250176899939791e-06,
+      "loss": 0.7745,
+      "step": 4426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1221458928079697e-06,
+      "loss": 0.8508,
+      "step": 4427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1192775477858776e-06,
+      "loss": 0.749,
+      "step": 4428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1164126560430633e-06,
+      "loss": 0.7896,
+      "step": 4429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1135512186935392e-06,
+      "loss": 0.821,
+      "step": 4430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1106932368499824e-06,
+      "loss": 0.7789,
+      "step": 4431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1078387116237198e-06,
+      "loss": 0.8253,
+      "step": 4432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1049876441247399e-06,
+      "loss": 0.8046,
+      "step": 4433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1021400354616786e-06,
+      "loss": 0.7587,
+      "step": 4434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0992958867418358e-06,
+      "loss": 0.8375,
+      "step": 4435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0964551990711624e-06,
+      "loss": 0.8415,
+      "step": 4436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0936179735542607e-06,
+      "loss": 0.7214,
+      "step": 4437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0907842112943922e-06,
+      "loss": 0.7943,
+      "step": 4438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0879539133934646e-06,
+      "loss": 0.828,
+      "step": 4439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0851270809520443e-06,
+      "loss": 0.8191,
+      "step": 4440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0823037150693517e-06,
+      "loss": 0.7894,
+      "step": 4441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0794838168432498e-06,
+      "loss": 0.8102,
+      "step": 4442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0766673873702637e-06,
+      "loss": 0.7789,
+      "step": 4443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0738544277455632e-06,
+      "loss": 0.7655,
+      "step": 4444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0710449390629719e-06,
+      "loss": 0.806,
+      "step": 4445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0682389224149648e-06,
+      "loss": 0.7974,
+      "step": 4446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0654363788926614e-06,
+      "loss": 0.746,
+      "step": 4447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0626373095858367e-06,
+      "loss": 0.7709,
+      "step": 4448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0598417155829078e-06,
+      "loss": 0.8575,
+      "step": 4449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0570495979709505e-06,
+      "loss": 0.8396,
+      "step": 4450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0542609578356778e-06,
+      "loss": 0.7625,
+      "step": 4451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0514757962614576e-06,
+      "loss": 0.7598,
+      "step": 4452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0486941143313045e-06,
+      "loss": 0.7889,
+      "step": 4453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.045915913126877e-06,
+      "loss": 0.8541,
+      "step": 4454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0431411937284842e-06,
+      "loss": 0.8193,
+      "step": 4455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.040369957215076e-06,
+      "loss": 0.7746,
+      "step": 4456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.037602204664252e-06,
+      "loss": 0.7941,
+      "step": 4457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0348379371522577e-06,
+      "loss": 0.7928,
+      "step": 4458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0320771557539788e-06,
+      "loss": 0.8779,
+      "step": 4459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0293198615429523e-06,
+      "loss": 0.8027,
+      "step": 4460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0265660555913504e-06,
+      "loss": 0.3211,
+      "step": 4461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0238157389699954e-06,
+      "loss": 0.838,
+      "step": 4462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0210689127483552e-06,
+      "loss": 0.8235,
+      "step": 4463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0183255779945312e-06,
+      "loss": 0.3273,
+      "step": 4464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0155857357752741e-06,
+      "loss": 0.8085,
+      "step": 4465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.012849387155974e-06,
+      "loss": 0.8134,
+      "step": 4466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0101165332006668e-06,
+      "loss": 0.7137,
+      "step": 4467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0073871749720221e-06,
+      "loss": 0.8041,
+      "step": 4468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0046613135313544e-06,
+      "loss": 0.8072,
+      "step": 4469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0019389499386212e-06,
+      "loss": 0.8177,
+      "step": 4470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.992200852524124e-07,
+      "loss": 0.8345,
+      "step": 4471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.965047205299649e-07,
+      "loss": 0.7188,
+      "step": 4472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.937928568271493e-07,
+      "loss": 0.7653,
+      "step": 4473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.910844951984788e-07,
+      "loss": 0.7943,
+      "step": 4474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.883796366971021e-07,
+      "loss": 0.7869,
+      "step": 4475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.856782823748067e-07,
+      "loss": 0.8486,
+      "step": 4476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.829804332820182e-07,
+      "loss": 0.8014,
+      "step": 4477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.802860904677959e-07,
+      "loss": 0.8182,
+      "step": 4478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.775952549798406e-07,
+      "loss": 0.3562,
+      "step": 4479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.749079278644846e-07,
+      "loss": 0.8167,
+      "step": 4480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.722241101667007e-07,
+      "loss": 0.8001,
+      "step": 4481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.695438029300941e-07,
+      "loss": 0.7553,
+      "step": 4482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.668670071969033e-07,
+      "loss": 0.8175,
+      "step": 4483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.641937240080046e-07,
+      "loss": 0.8015,
+      "step": 4484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.615239544029066e-07,
+      "loss": 0.801,
+      "step": 4485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.588576994197563e-07,
+      "loss": 0.8032,
+      "step": 4486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.561949600953246e-07,
+      "loss": 0.7931,
+      "step": 4487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.535357374650234e-07,
+      "loss": 0.8137,
+      "step": 4488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.508800325628953e-07,
+      "loss": 0.8288,
+      "step": 4489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.482278464216121e-07,
+      "loss": 0.7688,
+      "step": 4490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.4557918007248e-07,
+      "loss": 0.738,
+      "step": 4491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.429340345454397e-07,
+      "loss": 0.7622,
+      "step": 4492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.402924108690536e-07,
+      "loss": 0.8007,
+      "step": 4493
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.376543100705238e-07,
+      "loss": 0.8315,
+      "step": 4494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.350197331756761e-07,
+      "loss": 0.813,
+      "step": 4495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.323886812089722e-07,
+      "loss": 0.7867,
+      "step": 4496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.29761155193496e-07,
+      "loss": 0.7647,
+      "step": 4497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.271371561509657e-07,
+      "loss": 0.8337,
+      "step": 4498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.245166851017284e-07,
+      "loss": 0.7482,
+      "step": 4499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.218997430647548e-07,
+      "loss": 0.794,
+      "step": 4500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.192863310576472e-07,
+      "loss": 0.8395,
+      "step": 4501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.166764500966319e-07,
+      "loss": 0.7727,
+      "step": 4502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.140701011965647e-07,
+      "loss": 0.7915,
+      "step": 4503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.114672853709305e-07,
+      "loss": 0.7802,
+      "step": 4504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.088680036318342e-07,
+      "loss": 0.7674,
+      "step": 4505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.062722569900085e-07,
+      "loss": 0.7756,
+      "step": 4506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.036800464548157e-07,
+      "loss": 0.8067,
+      "step": 4507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.010913730342397e-07,
+      "loss": 0.812,
+      "step": 4508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.985062377348863e-07,
+      "loss": 0.7702,
+      "step": 4509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.959246415619905e-07,
+      "loss": 0.7463,
+      "step": 4510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.933465855194101e-07,
+      "loss": 0.8048,
+      "step": 4511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.907720706096223e-07,
+      "loss": 0.814,
+      "step": 4512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.88201097833733e-07,
+      "loss": 0.8107,
+      "step": 4513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.856336681914646e-07,
+      "loss": 0.8268,
+      "step": 4514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.830697826811674e-07,
+      "loss": 0.8064,
+      "step": 4515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.805094422998117e-07,
+      "loss": 0.8088,
+      "step": 4516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.779526480429868e-07,
+      "loss": 0.7571,
+      "step": 4517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.753994009049082e-07,
+      "loss": 0.8147,
+      "step": 4518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.728497018784065e-07,
+      "loss": 0.7879,
+      "step": 4519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.703035519549341e-07,
+      "loss": 0.7813,
+      "step": 4520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.677609521245689e-07,
+      "loss": 0.8042,
+      "step": 4521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.65221903375999e-07,
+      "loss": 0.7694,
+      "step": 4522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.626864066965401e-07,
+      "loss": 0.8308,
+      "step": 4523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.601544630721203e-07,
+      "loss": 0.784,
+      "step": 4524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.576260734872909e-07,
+      "loss": 0.8173,
+      "step": 4525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.551012389252178e-07,
+      "loss": 0.7826,
+      "step": 4526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.525799603676843e-07,
+      "loss": 0.7836,
+      "step": 4527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.500622387950963e-07,
+      "loss": 0.7987,
+      "step": 4528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.475480751864695e-07,
+      "loss": 0.8142,
+      "step": 4529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.450374705194431e-07,
+      "loss": 0.8407,
+      "step": 4530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.425304257702649e-07,
+      "loss": 0.755,
+      "step": 4531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.40026941913803e-07,
+      "loss": 0.7316,
+      "step": 4532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.37527019923543e-07,
+      "loss": 0.8001,
+      "step": 4533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.350306607715774e-07,
+      "loss": 0.8065,
+      "step": 4534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.325378654286231e-07,
+      "loss": 0.8322,
+      "step": 4535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.300486348640036e-07,
+      "loss": 0.7704,
+      "step": 4536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.275629700456588e-07,
+      "loss": 0.8051,
+      "step": 4537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.25080871940146e-07,
+      "loss": 0.8351,
+      "step": 4538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.226023415126272e-07,
+      "loss": 0.7727,
+      "step": 4539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.201273797268861e-07,
+      "loss": 0.8265,
+      "step": 4540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.176559875453105e-07,
+      "loss": 0.7931,
+      "step": 4541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.151881659289085e-07,
+      "loss": 0.7526,
+      "step": 4542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.127239158372902e-07,
+      "loss": 0.785,
+      "step": 4543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.102632382286857e-07,
+      "loss": 0.7417,
+      "step": 4544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.07806134059933e-07,
+      "loss": 0.3292,
+      "step": 4545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.053526042864768e-07,
+      "loss": 0.7783,
+      "step": 4546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.029026498623771e-07,
+      "loss": 0.8215,
+      "step": 4547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.004562717403019e-07,
+      "loss": 0.7897,
+      "step": 4548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.98013470871527e-07,
+      "loss": 0.833,
+      "step": 4549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.955742482059381e-07,
+      "loss": 0.8026,
+      "step": 4550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.931386046920308e-07,
+      "loss": 0.8193,
+      "step": 4551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.907065412769089e-07,
+      "loss": 0.8388,
+      "step": 4552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.8827805890628e-07,
+      "loss": 0.7293,
+      "step": 4553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.858531585244655e-07,
+      "loss": 0.7935,
+      "step": 4554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.834318410743891e-07,
+      "loss": 0.8207,
+      "step": 4555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.810141074975819e-07,
+      "loss": 0.7762,
+      "step": 4556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.785999587341853e-07,
+      "loss": 0.7858,
+      "step": 4557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.761893957229405e-07,
+      "loss": 0.8159,
+      "step": 4558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.737824194012022e-07,
+      "loss": 0.7953,
+      "step": 4559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.713790307049207e-07,
+      "loss": 0.8258,
+      "step": 4560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.689792305686583e-07,
+      "loss": 0.7502,
+      "step": 4561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.665830199255842e-07,
+      "loss": 0.771,
+      "step": 4562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.641903997074607e-07,
+      "loss": 0.7581,
+      "step": 4563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.61801370844667e-07,
+      "loss": 0.7881,
+      "step": 4564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.594159342661767e-07,
+      "loss": 0.8009,
+      "step": 4565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.570340908995688e-07,
+      "loss": 0.7736,
+      "step": 4566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.546558416710292e-07,
+      "loss": 0.8177,
+      "step": 4567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.522811875053392e-07,
+      "loss": 0.7773,
+      "step": 4568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.499101293258881e-07,
+      "loss": 0.8509,
+      "step": 4569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.47542668054666e-07,
+      "loss": 0.7945,
+      "step": 4570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.451788046122588e-07,
+      "loss": 0.7837,
+      "step": 4571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.428185399178601e-07,
+      "loss": 0.806,
+      "step": 4572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.404618748892612e-07,
+      "loss": 0.3393,
+      "step": 4573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.381088104428558e-07,
+      "loss": 0.7715,
+      "step": 4574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.35759347493632e-07,
+      "loss": 0.8207,
+      "step": 4575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.334134869551857e-07,
+      "loss": 0.7644,
+      "step": 4576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.310712297397038e-07,
+      "loss": 0.7863,
+      "step": 4577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.287325767579756e-07,
+      "loss": 0.7633,
+      "step": 4578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.263975289193937e-07,
+      "loss": 0.8298,
+      "step": 4579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.240660871319383e-07,
+      "loss": 0.8634,
+      "step": 4580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.217382523021988e-07,
+      "loss": 0.3302,
+      "step": 4581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.19414025335351e-07,
+      "loss": 0.8126,
+      "step": 4582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.170934071351754e-07,
+      "loss": 0.8818,
+      "step": 4583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.147763986040502e-07,
+      "loss": 0.3539,
+      "step": 4584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.124630006429423e-07,
+      "loss": 0.75,
+      "step": 4585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.101532141514233e-07,
+      "loss": 0.8079,
+      "step": 4586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.078470400276516e-07,
+      "loss": 0.3201,
+      "step": 4587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.055444791683907e-07,
+      "loss": 0.8313,
+      "step": 4588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.032455324689902e-07,
+      "loss": 0.7648,
+      "step": 4589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.009502008233993e-07,
+      "loss": 0.8061,
+      "step": 4590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.986584851241607e-07,
+      "loss": 0.8609,
+      "step": 4591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.963703862624104e-07,
+      "loss": 0.7994,
+      "step": 4592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.940859051278814e-07,
+      "loss": 0.8101,
+      "step": 4593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.918050426088918e-07,
+      "loss": 0.8232,
+      "step": 4594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.8952779959236e-07,
+      "loss": 0.8558,
+      "step": 4595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.872541769637964e-07,
+      "loss": 0.7715,
+      "step": 4596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.849841756072984e-07,
+      "loss": 0.8484,
+      "step": 4597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.82717796405562e-07,
+      "loss": 0.7641,
+      "step": 4598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.804550402398679e-07,
+      "loss": 0.7861,
+      "step": 4599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.781959079900958e-07,
+      "loss": 0.8731,
+      "step": 4600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.75940400534707e-07,
+      "loss": 0.744,
+      "step": 4601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.736885187507602e-07,
+      "loss": 0.3004,
+      "step": 4602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.71440263513905e-07,
+      "loss": 0.8217,
+      "step": 4603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.691956356983731e-07,
+      "loss": 0.8125,
+      "step": 4604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.669546361769952e-07,
+      "loss": 0.8509,
+      "step": 4605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.647172658211831e-07,
+      "loss": 0.7922,
+      "step": 4606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.624835255009432e-07,
+      "loss": 0.7846,
+      "step": 4607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.602534160848673e-07,
+      "loss": 0.7621,
+      "step": 4608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.58026938440135e-07,
+      "loss": 0.7708,
+      "step": 4609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.558040934325183e-07,
+      "loss": 0.7797,
+      "step": 4610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.535848819263679e-07,
+      "loss": 0.7539,
+      "step": 4611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.513693047846292e-07,
+      "loss": 0.8209,
+      "step": 4612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.49157362868832e-07,
+      "loss": 0.8533,
+      "step": 4613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.469490570390935e-07,
+      "loss": 0.7935,
+      "step": 4614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.44744388154116e-07,
+      "loss": 0.7832,
+      "step": 4615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.425433570711848e-07,
+      "loss": 0.8069,
+      "step": 4616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.403459646461774e-07,
+      "loss": 0.8485,
+      "step": 4617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.381522117335482e-07,
+      "loss": 0.8363,
+      "step": 4618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.359620991863425e-07,
+      "loss": 0.8155,
+      "step": 4619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.337756278561902e-07,
+      "loss": 0.7935,
+      "step": 4620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.315927985933001e-07,
+      "loss": 0.8465,
+      "step": 4621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.294136122464701e-07,
+      "loss": 0.7478,
+      "step": 4622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.272380696630765e-07,
+      "loss": 0.7835,
+      "step": 4623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.250661716890838e-07,
+      "loss": 0.7733,
+      "step": 4624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.228979191690387e-07,
+      "loss": 0.8297,
+      "step": 4625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.207333129460646e-07,
+      "loss": 0.3201,
+      "step": 4626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.185723538618738e-07,
+      "loss": 0.7568,
+      "step": 4627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.164150427567572e-07,
+      "loss": 0.8027,
+      "step": 4628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.142613804695885e-07,
+      "loss": 0.3151,
+      "step": 4629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.121113678378199e-07,
+      "loss": 0.8883,
+      "step": 4630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.09965005697487e-07,
+      "loss": 0.8263,
+      "step": 4631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.078222948832058e-07,
+      "loss": 0.7965,
+      "step": 4632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.056832362281728e-07,
+      "loss": 0.8412,
+      "step": 4633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.035478305641617e-07,
+      "loss": 0.7935,
+      "step": 4634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.014160787215273e-07,
+      "loss": 0.8044,
+      "step": 4635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.992879815292052e-07,
+      "loss": 0.8752,
+      "step": 4636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.971635398147091e-07,
+      "loss": 0.7997,
+      "step": 4637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.950427544041293e-07,
+      "loss": 0.806,
+      "step": 4638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.929256261221361e-07,
+      "loss": 0.7964,
+      "step": 4639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.908121557919755e-07,
+      "loss": 0.7715,
+      "step": 4640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.887023442354744e-07,
+      "loss": 0.83,
+      "step": 4641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.865961922730379e-07,
+      "loss": 0.7722,
+      "step": 4642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.844937007236407e-07,
+      "loss": 0.838,
+      "step": 4643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.823948704048443e-07,
+      "loss": 0.7517,
+      "step": 4644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.802997021327761e-07,
+      "loss": 0.7495,
+      "step": 4645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.782081967221498e-07,
+      "loss": 0.8442,
+      "step": 4646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.761203549862448e-07,
+      "loss": 0.8473,
+      "step": 4647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.740361777369241e-07,
+      "loss": 0.6776,
+      "step": 4648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.719556657846226e-07,
+      "loss": 0.791,
+      "step": 4649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.698788199383476e-07,
+      "loss": 0.7984,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.678056410056854e-07,
+      "loss": 0.8552,
+      "step": 4651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.65736129792791e-07,
+      "loss": 0.8799,
+      "step": 4652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.636702871043986e-07,
+      "loss": 0.3205,
+      "step": 4653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.616081137438123e-07,
+      "loss": 0.8017,
+      "step": 4654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.595496105129139e-07,
+      "loss": 0.8024,
+      "step": 4655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.574947782121498e-07,
+      "loss": 0.8196,
+      "step": 4656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.554436176405475e-07,
+      "loss": 0.7453,
+      "step": 4657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.533961295957013e-07,
+      "loss": 0.8201,
+      "step": 4658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.513523148737809e-07,
+      "loss": 0.8312,
+      "step": 4659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.493121742695251e-07,
+      "loss": 0.7785,
+      "step": 4660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.472757085762459e-07,
+      "loss": 0.7929,
+      "step": 4661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.452429185858221e-07,
+      "loss": 0.7618,
+      "step": 4662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.432138050887115e-07,
+      "loss": 0.8051,
+      "step": 4663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.411883688739328e-07,
+      "loss": 0.8063,
+      "step": 4664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.391666107290805e-07,
+      "loss": 0.7725,
+      "step": 4665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.371485314403202e-07,
+      "loss": 0.7713,
+      "step": 4666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.351341317923808e-07,
+      "loss": 0.8131,
+      "step": 4667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.331234125685669e-07,
+      "loss": 0.7973,
+      "step": 4668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.311163745507464e-07,
+      "loss": 0.7483,
+      "step": 4669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.291130185193594e-07,
+      "loss": 0.8238,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.271133452534149e-07,
+      "loss": 0.7355,
+      "step": 4671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.251173555304845e-07,
+      "loss": 0.8609,
+      "step": 4672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.23125050126716e-07,
+      "loss": 0.7913,
+      "step": 4673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.211364298168142e-07,
+      "loss": 0.8629,
+      "step": 4674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.191514953740606e-07,
+      "loss": 0.813,
+      "step": 4675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.171702475702967e-07,
+      "loss": 0.7692,
+      "step": 4676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.15192687175935e-07,
+      "loss": 0.732,
+      "step": 4677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.132188149599526e-07,
+      "loss": 0.7758,
+      "step": 4678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.11248631689889e-07,
+      "loss": 0.8531,
+      "step": 4679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.092821381318558e-07,
+      "loss": 0.307,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.073193350505224e-07,
+      "loss": 0.8088,
+      "step": 4681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.053602232091304e-07,
+      "loss": 0.813,
+      "step": 4682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.034048033694816e-07,
+      "loss": 0.8505,
+      "step": 4683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.014530762919423e-07,
+      "loss": 0.7694,
+      "step": 4684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.995050427354464e-07,
+      "loss": 0.6567,
+      "step": 4685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.975607034574859e-07,
+      "loss": 0.7808,
+      "step": 4686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.956200592141213e-07,
+      "loss": 0.8521,
+      "step": 4687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.936831107599749e-07,
+      "loss": 0.7918,
+      "step": 4688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.917498588482295e-07,
+      "loss": 0.7814,
+      "step": 4689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.898203042306337e-07,
+      "loss": 0.8681,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.878944476574965e-07,
+      "loss": 0.8108,
+      "step": 4691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.859722898776898e-07,
+      "loss": 0.8157,
+      "step": 4692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.840538316386457e-07,
+      "loss": 0.8287,
+      "step": 4693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.821390736863607e-07,
+      "loss": 0.7341,
+      "step": 4694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.802280167653906e-07,
+      "loss": 0.7864,
+      "step": 4695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.783206616188496e-07,
+      "loss": 0.8628,
+      "step": 4696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.764170089884179e-07,
+      "loss": 0.8238,
+      "step": 4697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7451705961433115e-07,
+      "loss": 0.8013,
+      "step": 4698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7262081423538717e-07,
+      "loss": 0.8505,
+      "step": 4699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7072827358894626e-07,
+      "loss": 0.8149,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.688394384109207e-07,
+      "loss": 0.7931,
+      "step": 4701
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6695430943578865e-07,
+      "loss": 0.8063,
+      "step": 4702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6507288739658264e-07,
+      "loss": 0.7723,
+      "step": 4703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6319517302489957e-07,
+      "loss": 0.8315,
+      "step": 4704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6132116705088657e-07,
+      "loss": 0.8311,
+      "step": 4705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5945087020325517e-07,
+      "loss": 0.3366,
+      "step": 4706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.575842832092736e-07,
+      "loss": 0.8366,
+      "step": 4707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5572140679476463e-07,
+      "loss": 0.8109,
+      "step": 4708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5386224168411097e-07,
+      "loss": 0.8263,
+      "step": 4709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.520067886002488e-07,
+      "loss": 0.8304,
+      "step": 4710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.501550482646755e-07,
+      "loss": 0.7749,
+      "step": 4711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4830702139744386e-07,
+      "loss": 0.8201,
+      "step": 4712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4646270871715693e-07,
+      "loss": 0.8037,
+      "step": 4713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.446221109409821e-07,
+      "loss": 0.8535,
+      "step": 4714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.427852287846346e-07,
+      "loss": 0.819,
+      "step": 4715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.409520629623898e-07,
+      "loss": 0.7566,
+      "step": 4716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.391226141870786e-07,
+      "loss": 0.7885,
+      "step": 4717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3729688317007966e-07,
+      "loss": 0.8147,
+      "step": 4718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.35474870621333e-07,
+      "loss": 0.8073,
+      "step": 4719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.336565772493306e-07,
+      "loss": 0.7688,
+      "step": 4720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.318420037611182e-07,
+      "loss": 0.7714,
+      "step": 4721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.300311508622945e-07,
+      "loss": 0.7587,
+      "step": 4722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2822401925700973e-07,
+      "loss": 0.2925,
+      "step": 4723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.264206096479728e-07,
+      "loss": 0.7829,
+      "step": 4724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.246209227364395e-07,
+      "loss": 0.7872,
+      "step": 4725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.228249592222211e-07,
+      "loss": 0.8313,
+      "step": 4726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.210327198036779e-07,
+      "loss": 0.8076,
+      "step": 4727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1924420517772567e-07,
+      "loss": 0.8456,
+      "step": 4728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1745941603983155e-07,
+      "loss": 0.8566,
+      "step": 4729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.156783530840114e-07,
+      "loss": 0.7232,
+      "step": 4730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.139010170028346e-07,
+      "loss": 0.3395,
+      "step": 4731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.121274084874194e-07,
+      "loss": 0.82,
+      "step": 4732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1035752822743524e-07,
+      "loss": 0.7784,
+      "step": 4733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0859137691110495e-07,
+      "loss": 0.8073,
+      "step": 4734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0682895522519474e-07,
+      "loss": 0.8441,
+      "step": 4735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 0.8486,
+      "step": 4736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0331530348446944e-07,
+      "loss": 0.7447,
+      "step": 4737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.015640747959404e-07,
+      "loss": 0.8358,
+      "step": 4738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9981657847040776e-07,
+      "loss": 0.8191,
+      "step": 4739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9807281518738916e-07,
+      "loss": 0.7601,
+      "step": 4740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9633278562494657e-07,
+      "loss": 0.8592,
+      "step": 4741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9459649045969216e-07,
+      "loss": 0.8176,
+      "step": 4742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9286393036678914e-07,
+      "loss": 0.7842,
+      "step": 4743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.911351060199431e-07,
+      "loss": 0.7708,
+      "step": 4744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.894100180914095e-07,
+      "loss": 0.8076,
+      "step": 4745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.876886672519931e-07,
+      "loss": 0.7552,
+      "step": 4746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.859710541710393e-07,
+      "loss": 0.8337,
+      "step": 4747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.842571795164485e-07,
+      "loss": 0.8076,
+      "step": 4748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8254704395465856e-07,
+      "loss": 0.8095,
+      "step": 4749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8084064815066104e-07,
+      "loss": 0.826,
+      "step": 4750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.791379927679884e-07,
+      "loss": 0.8849,
+      "step": 4751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.774390784687187e-07,
+      "loss": 0.8187,
+      "step": 4752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7574390591348044e-07,
+      "loss": 0.7819,
+      "step": 4753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7405247576144055e-07,
+      "loss": 0.7791,
+      "step": 4754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.723647886703141e-07,
+      "loss": 0.7667,
+      "step": 4755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.706808452963606e-07,
+      "loss": 0.8221,
+      "step": 4756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.690006462943829e-07,
+      "loss": 0.7895,
+      "step": 4757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6732419231773063e-07,
+      "loss": 0.7921,
+      "step": 4758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.656514840182912e-07,
+      "loss": 0.8296,
+      "step": 4759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.63982522046501e-07,
+      "loss": 0.8291,
+      "step": 4760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.623173070513375e-07,
+      "loss": 0.8145,
+      "step": 4761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6065583968032283e-07,
+      "loss": 0.7504,
+      "step": 4762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.58998120579519e-07,
+      "loss": 0.7781,
+      "step": 4763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5734415039353263e-07,
+      "loss": 0.8635,
+      "step": 4764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.556939297655115e-07,
+      "loss": 0.8,
+      "step": 4765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.540474593371468e-07,
+      "loss": 0.8527,
+      "step": 4766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.524047397486696e-07,
+      "loss": 0.3522,
+      "step": 4767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.507657716388535e-07,
+      "loss": 0.8373,
+      "step": 4768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.491305556450131e-07,
+      "loss": 0.8619,
+      "step": 4769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474990924030042e-07,
+      "loss": 0.8548,
+      "step": 4770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.458713825472226e-07,
+      "loss": 0.79,
+      "step": 4771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.442474267106077e-07,
+      "loss": 0.8514,
+      "step": 4772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4262722552463213e-07,
+      "loss": 0.7969,
+      "step": 4773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4101077961931627e-07,
+      "loss": 0.7506,
+      "step": 4774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.393980896232174e-07,
+      "loss": 0.8186,
+      "step": 4775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3778915616342947e-07,
+      "loss": 0.8018,
+      "step": 4776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.361839798655897e-07,
+      "loss": 0.7854,
+      "step": 4777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3458256135387225e-07,
+      "loss": 0.75,
+      "step": 4778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3298490125099116e-07,
+      "loss": 0.8245,
+      "step": 4779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3139100017819835e-07,
+      "loss": 0.7783,
+      "step": 4780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2980085875528254e-07,
+      "loss": 0.764,
+      "step": 4781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2821447760057467e-07,
+      "loss": 0.7591,
+      "step": 4782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2663185733093796e-07,
+      "loss": 0.8021,
+      "step": 4783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.250529985617801e-07,
+      "loss": 0.7684,
+      "step": 4784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2347790190703776e-07,
+      "loss": 0.7891,
+      "step": 4785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2190656797919215e-07,
+      "loss": 0.7592,
+      "step": 4786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.203389973892579e-07,
+      "loss": 0.7795,
+      "step": 4787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.187751907467851e-07,
+      "loss": 0.7855,
+      "step": 4788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.172151486598629e-07,
+      "loss": 0.8094,
+      "step": 4789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.15658871735115e-07,
+      "loss": 0.8634,
+      "step": 4790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.141063605777028e-07,
+      "loss": 0.8093,
+      "step": 4791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.125576157913213e-07,
+      "loss": 0.8273,
+      "step": 4792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.110126379781997e-07,
+      "loss": 0.8105,
+      "step": 4793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0947142773910863e-07,
+      "loss": 0.839,
+      "step": 4794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0793398567334544e-07,
+      "loss": 0.7886,
+      "step": 4795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.064003123787496e-07,
+      "loss": 0.8403,
+      "step": 4796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0487040845168847e-07,
+      "loss": 0.8223,
+      "step": 4797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.033442744870685e-07,
+      "loss": 0.7754,
+      "step": 4798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0182191107833047e-07,
+      "loss": 0.7989,
+      "step": 4799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0030331881744315e-07,
+      "loss": 0.7656,
+      "step": 4800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.987884982949163e-07,
+      "loss": 0.8326,
+      "step": 4801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9727745009978994e-07,
+      "loss": 0.2786,
+      "step": 4802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.95770174819634e-07,
+      "loss": 0.8123,
+      "step": 4803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9426667304055525e-07,
+      "loss": 0.7715,
+      "step": 4804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9276694534719375e-07,
+      "loss": 0.7401,
+      "step": 4805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.912709923227197e-07,
+      "loss": 0.8543,
+      "step": 4806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.897788145488345e-07,
+      "loss": 0.777,
+      "step": 4807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8829041260577394e-07,
+      "loss": 0.7758,
+      "step": 4808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.868057870723073e-07,
+      "loss": 0.8234,
+      "step": 4809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.853249385257295e-07,
+      "loss": 0.3104,
+      "step": 4810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.838478675418732e-07,
+      "loss": 0.7479,
+      "step": 4811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8237457469509676e-07,
+      "loss": 0.8302,
+      "step": 4812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8090506055829304e-07,
+      "loss": 0.8096,
+      "step": 4813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.794393257028838e-07,
+      "loss": 0.3065,
+      "step": 4814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7797737069882203e-07,
+      "loss": 0.8369,
+      "step": 4815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.765191961145908e-07,
+      "loss": 0.7504,
+      "step": 4816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.750648025172031e-07,
+      "loss": 0.7923,
+      "step": 4817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7361419047220115e-07,
+      "loss": 0.7844,
+      "step": 4818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.72167360543657e-07,
+      "loss": 0.8536,
+      "step": 4819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.707243132941717e-07,
+      "loss": 0.7807,
+      "step": 4820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6928504928487755e-07,
+      "loss": 0.3101,
+      "step": 4821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6784956907543234e-07,
+      "loss": 0.7761,
+      "step": 4822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.664178732240252e-07,
+      "loss": 0.762,
+      "step": 4823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6498996228737193e-07,
+      "loss": 0.8398,
+      "step": 4824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6356583682071945e-07,
+      "loss": 0.7679,
+      "step": 4825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6214549737783813e-07,
+      "loss": 0.3074,
+      "step": 4826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.607289445110284e-07,
+      "loss": 0.7984,
+      "step": 4827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5931617877112183e-07,
+      "loss": 0.8205,
+      "step": 4828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.579072007074701e-07,
+      "loss": 0.7909,
+      "step": 4829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5650201086795946e-07,
+      "loss": 0.7965,
+      "step": 4830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.551006097989961e-07,
+      "loss": 0.7775,
+      "step": 4831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.537029980455197e-07,
+      "loss": 0.7701,
+      "step": 4832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.52309176150991e-07,
+      "loss": 0.8104,
+      "step": 4833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.509191446574e-07,
+      "loss": 0.3273,
+      "step": 4834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.49532904105263e-07,
+      "loss": 0.8364,
+      "step": 4835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.481504550336189e-07,
+      "loss": 0.7839,
+      "step": 4836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.467717979800355e-07,
+      "loss": 0.8218,
+      "step": 4837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4539693348060835e-07,
+      "loss": 0.8099,
+      "step": 4838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.440258620699498e-07,
+      "loss": 0.729,
+      "step": 4839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.426585842812079e-07,
+      "loss": 0.7805,
+      "step": 4840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.412951006460451e-07,
+      "loss": 0.7659,
+      "step": 4841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.399354116946584e-07,
+      "loss": 0.7884,
+      "step": 4842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.385795179557604e-07,
+      "loss": 0.8398,
+      "step": 4843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3722741995659382e-07,
+      "loss": 0.8102,
+      "step": 4844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3587911822292364e-07,
+      "loss": 0.7724,
+      "step": 4845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3453461327903825e-07,
+      "loss": 0.7219,
+      "step": 4846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3319390564775168e-07,
+      "loss": 0.3371,
+      "step": 4847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.318569958503958e-07,
+      "loss": 0.8665,
+      "step": 4848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3052388440683248e-07,
+      "loss": 0.7704,
+      "step": 4849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.291945718354438e-07,
+      "loss": 0.7814,
+      "step": 4850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.278690586531329e-07,
+      "loss": 0.7897,
+      "step": 4851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2654734537532864e-07,
+      "loss": 0.807,
+      "step": 4852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2522943251597874e-07,
+      "loss": 0.8205,
+      "step": 4853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2391532058755772e-07,
+      "loss": 0.787,
+      "step": 4854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2260501010105795e-07,
+      "loss": 0.8765,
+      "step": 4855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2129850156599408e-07,
+      "loss": 0.6871,
+      "step": 4856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1999579549040752e-07,
+      "loss": 0.8219,
+      "step": 4857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.186968923808519e-07,
+      "loss": 0.7118,
+      "step": 4858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1740179274241103e-07,
+      "loss": 0.8473,
+      "step": 4859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1611049707868425e-07,
+      "loss": 0.3312,
+      "step": 4860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1482300589179327e-07,
+      "loss": 0.8097,
+      "step": 4861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1353931968238206e-07,
+      "loss": 0.3344,
+      "step": 4862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.122594389496113e-07,
+      "loss": 0.8172,
+      "step": 4863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1098336419116628e-07,
+      "loss": 0.7755,
+      "step": 4864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.097110959032489e-07,
+      "loss": 0.7554,
+      "step": 4865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0844263458058122e-07,
+      "loss": 0.7488,
+      "step": 4866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0717798071640761e-07,
+      "loss": 0.7302,
+      "step": 4867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0591713480248909e-07,
+      "loss": 0.8288,
+      "step": 4868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0466009732910908e-07,
+      "loss": 0.8087,
+      "step": 4869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0340686878506655e-07,
+      "loss": 0.8515,
+      "step": 4870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0215744965768057e-07,
+      "loss": 0.8002,
+      "step": 4871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0091184043278923e-07,
+      "loss": 0.8376,
+      "step": 4872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9967004159474946e-07,
+      "loss": 0.7508,
+      "step": 4873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9843205362643724e-07,
+      "loss": 0.8164,
+      "step": 4874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9719787700924308e-07,
+      "loss": 0.8063,
+      "step": 4875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9596751222308085e-07,
+      "loss": 0.7846,
+      "step": 4876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9474095974637673e-07,
+      "loss": 0.825,
+      "step": 4877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9351822005607922e-07,
+      "loss": 0.8228,
+      "step": 4878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.922992936276513e-07,
+      "loss": 0.7282,
+      "step": 4879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9108418093507385e-07,
+      "loss": 0.8138,
+      "step": 4880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8987288245084556e-07,
+      "loss": 0.7589,
+      "step": 4881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8866539864597855e-07,
+      "loss": 0.8211,
+      "step": 4882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8746172999000722e-07,
+      "loss": 0.3457,
+      "step": 4883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.862618769509794e-07,
+      "loss": 0.8336,
+      "step": 4884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8506583999545747e-07,
+      "loss": 0.7612,
+      "step": 4885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8387361958852378e-07,
+      "loss": 0.7505,
+      "step": 4886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8268521619377312e-07,
+      "loss": 0.8041,
+      "step": 4887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8150063027331798e-07,
+      "loss": 0.7663,
+      "step": 4888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8031986228778664e-07,
+      "loss": 0.7647,
+      "step": 4889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7914291269632067e-07,
+      "loss": 0.8327,
+      "step": 4890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7796978195658067e-07,
+      "loss": 0.6958,
+      "step": 4891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7680047052473837e-07,
+      "loss": 0.8365,
+      "step": 4892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.756349788554812e-07,
+      "loss": 0.3503,
+      "step": 4893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.744733074020133e-07,
+      "loss": 0.777,
+      "step": 4894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7331545661604997e-07,
+      "loss": 0.7381,
+      "step": 4895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7216142694782668e-07,
+      "loss": 0.3357,
+      "step": 4896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.710112188460844e-07,
+      "loss": 0.78,
+      "step": 4897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6986483275808762e-07,
+      "loss": 0.7899,
+      "step": 4898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6872226912960532e-07,
+      "loss": 0.7494,
+      "step": 4899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6758352840492874e-07,
+      "loss": 0.811,
+      "step": 4900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6644861102685595e-07,
+      "loss": 0.7981,
+      "step": 4901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6531751743670053e-07,
+      "loss": 0.7532,
+      "step": 4902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6419024807429296e-07,
+      "loss": 0.7804,
+      "step": 4903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.630668033779692e-07,
+      "loss": 0.7748,
+      "step": 4904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.619471837845854e-07,
+      "loss": 0.321,
+      "step": 4905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608313897295033e-07,
+      "loss": 0.8281,
+      "step": 4906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5971942164660359e-07,
+      "loss": 0.7881,
+      "step": 4907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5861127996827597e-07,
+      "loss": 0.8596,
+      "step": 4908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.575069651254224e-07,
+      "loss": 0.797,
+      "step": 4909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5640647754745607e-07,
+      "loss": 0.7917,
+      "step": 4910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5530981766230356e-07,
+      "loss": 0.7952,
+      "step": 4911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5421698589640267e-07,
+      "loss": 0.826,
+      "step": 4912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.531279826747023e-07,
+      "loss": 0.8162,
+      "step": 4913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5204280842066266e-07,
+      "loss": 0.7708,
+      "step": 4914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5096146355625396e-07,
+      "loss": 0.7967,
+      "step": 4915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4988394850195876e-07,
+      "loss": 0.7522,
+      "step": 4916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4881026367677188e-07,
+      "loss": 0.8011,
+      "step": 4917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4774040949819378e-07,
+      "loss": 0.8646,
+      "step": 4918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4667438638224064e-07,
+      "loss": 0.7737,
+      "step": 4919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4561219474343747e-07,
+      "loss": 0.8727,
+      "step": 4920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4455383499481613e-07,
+      "loss": 0.8655,
+      "step": 4921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4349930754792408e-07,
+      "loss": 0.8634,
+      "step": 4922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4244861281281218e-07,
+      "loss": 0.7992,
+      "step": 4923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.41401751198047e-07,
+      "loss": 0.7828,
+      "step": 4924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4035872311070175e-07,
+      "loss": 0.7431,
+      "step": 4925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3931952895635649e-07,
+      "loss": 0.8038,
+      "step": 4926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3828416913910682e-07,
+      "loss": 0.7729,
+      "step": 4927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.372526440615507e-07,
+      "loss": 0.3338,
+      "step": 4928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.362249541247984e-07,
+      "loss": 0.7664,
+      "step": 4929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3520109972846918e-07,
+      "loss": 0.7536,
+      "step": 4930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3418108127069008e-07,
+      "loss": 0.8315,
+      "step": 4931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3316489914809605e-07,
+      "loss": 0.7711,
+      "step": 4932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3215255375582992e-07,
+      "loss": 0.8205,
+      "step": 4933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3114404548754566e-07,
+      "loss": 0.8302,
+      "step": 4934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3013937473540073e-07,
+      "loss": 0.8439,
+      "step": 4935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.291385418900626e-07,
+      "loss": 0.8044,
+      "step": 4936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.281415473407077e-07,
+      "loss": 0.811,
+      "step": 4937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2714839147501712e-07,
+      "loss": 0.8045,
+      "step": 4938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2615907467918187e-07,
+      "loss": 0.8551,
+      "step": 4939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.251735973378976e-07,
+      "loss": 0.8038,
+      "step": 4940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2419195983436881e-07,
+      "loss": 0.7833,
+      "step": 4941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2321416255030694e-07,
+      "loss": 0.7375,
+      "step": 4942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2224020586593e-07,
+      "loss": 0.7904,
+      "step": 4943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2127009015996062e-07,
+      "loss": 0.8216,
+      "step": 4944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2030381580963036e-07,
+      "loss": 0.7629,
+      "step": 4945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1934138319067646e-07,
+      "loss": 0.7461,
+      "step": 4946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1838279267734176e-07,
+      "loss": 0.7715,
+      "step": 4947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1742804464237368e-07,
+      "loss": 0.7578,
+      "step": 4948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1647713945702966e-07,
+      "loss": 0.7986,
+      "step": 4949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1553007749106837e-07,
+      "loss": 0.8033,
+      "step": 4950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1458685911275746e-07,
+      "loss": 0.7631,
+      "step": 4951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1364748468886688e-07,
+      "loss": 0.7452,
+      "step": 4952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1271195458467555e-07,
+      "loss": 0.7998,
+      "step": 4953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1178026916396356e-07,
+      "loss": 0.7813,
+      "step": 4954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1085242878901892e-07,
+      "loss": 0.7476,
+      "step": 4955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0992843382063301e-07,
+      "loss": 0.8126,
+      "step": 4956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0900828461810176e-07,
+      "loss": 0.8111,
+      "step": 4957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0809198153922673e-07,
+      "loss": 0.8279,
+      "step": 4958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0717952494031514e-07,
+      "loss": 0.857,
+      "step": 4959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0627091517617427e-07,
+      "loss": 0.849,
+      "step": 4960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0536615260011928e-07,
+      "loss": 0.7855,
+      "step": 4961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0446523756396764e-07,
+      "loss": 0.8022,
+      "step": 4962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0356817041804246e-07,
+      "loss": 0.7908,
+      "step": 4963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.026749515111669e-07,
+      "loss": 0.7919,
+      "step": 4964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0178558119067316e-07,
+      "loss": 0.7895,
+      "step": 4965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0090005980239236e-07,
+      "loss": 0.7789,
+      "step": 4966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0001838769065908e-07,
+      "loss": 0.8468,
+      "step": 4967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.914056519831572e-08,
+      "loss": 0.8051,
+      "step": 4968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.826659266670258e-08,
+      "loss": 0.8494,
+      "step": 4969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.739647043566447e-08,
+      "loss": 0.808,
+      "step": 4970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.653019884355075e-08,
+      "loss": 0.7884,
+      "step": 4971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.566777822721196e-08,
+      "loss": 0.7877,
+      "step": 4972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.480920892200096e-08,
+      "loss": 0.7687,
+      "step": 4973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.395449126177291e-08,
+      "loss": 0.8183,
+      "step": 4974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.310362557888642e-08,
+      "loss": 0.8628,
+      "step": 4975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.225661220420234e-08,
+      "loss": 0.8256,
+      "step": 4976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.141345146708169e-08,
+      "loss": 0.774,
+      "step": 4977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.057414369538887e-08,
+      "loss": 0.8115,
+      "step": 4978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.973868921548834e-08,
+      "loss": 0.8331,
+      "step": 4979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.890708835225026e-08,
+      "loss": 0.8003,
+      "step": 4980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.807934142904262e-08,
+      "loss": 0.7888,
+      "step": 4981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.72554487677335e-08,
+      "loss": 0.8565,
+      "step": 4982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.643541068869776e-08,
+      "loss": 0.7992,
+      "step": 4983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.561922751080697e-08,
+      "loss": 0.7543,
+      "step": 4984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.480689955143395e-08,
+      "loss": 0.8291,
+      "step": 4985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.399842712645378e-08,
+      "loss": 0.7673,
+      "step": 4986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.319381055024278e-08,
+      "loss": 0.7232,
+      "step": 4987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.239305013567734e-08,
+      "loss": 0.8144,
+      "step": 4988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.159614619413169e-08,
+      "loss": 0.7817,
+      "step": 4989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.080309903548577e-08,
+      "loss": 0.8318,
+      "step": 4990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.0013908968114e-08,
+      "loss": 0.8286,
+      "step": 4991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.922857629889646e-08,
+      "loss": 0.7634,
+      "step": 4992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.844710133320887e-08,
+      "loss": 0.7934,
+      "step": 4993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.766948437493039e-08,
+      "loss": 0.7888,
+      "step": 4994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.68957257264369e-08,
+      "loss": 0.7941,
+      "step": 4995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.61258256886055e-08,
+      "loss": 0.7616,
+      "step": 4996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.535978456081339e-08,
+      "loss": 0.8161,
+      "step": 4997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.459760264093563e-08,
+      "loss": 0.8154,
+      "step": 4998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.383928022534847e-08,
+      "loss": 0.7995,
+      "step": 4999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.308481760892605e-08,
+      "loss": 0.8437,
+      "step": 5000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.233421508504257e-08,
+      "loss": 0.8112,
+      "step": 5001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.1587472945569e-08,
+      "loss": 0.7834,
+      "step": 5002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.084459148087864e-08,
+      "loss": 0.7626,
+      "step": 5003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.01055709798415e-08,
+      "loss": 0.7882,
+      "step": 5004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.937041172982662e-08,
+      "loss": 0.7742,
+      "step": 5005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.863911401669976e-08,
+      "loss": 0.83,
+      "step": 5006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.791167812483013e-08,
+      "loss": 0.6963,
+      "step": 5007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.718810433707924e-08,
+      "loss": 0.7904,
+      "step": 5008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.646839293481088e-08,
+      "loss": 0.8455,
+      "step": 5009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.575254419788457e-08,
+      "loss": 0.8564,
+      "step": 5010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.504055840465983e-08,
+      "loss": 0.837,
+      "step": 5011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.433243583199412e-08,
+      "loss": 0.7811,
+      "step": 5012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.362817675523936e-08,
+      "loss": 0.7783,
+      "step": 5013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.292778144824984e-08,
+      "loss": 0.7759,
+      "step": 5014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.223125018337328e-08,
+      "loss": 0.8578,
+      "step": 5015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.153858323145856e-08,
+      "loss": 0.7799,
+      "step": 5016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.0849780861848e-08,
+      "loss": 0.8054,
+      "step": 5017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.016484334238515e-08,
+      "loss": 0.753,
+      "step": 5018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.948377093940694e-08,
+      "loss": 0.3274,
+      "step": 5019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.880656391775041e-08,
+      "loss": 0.8169,
+      "step": 5020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.813322254074827e-08,
+      "loss": 0.8109,
+      "step": 5021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.746374707022884e-08,
+      "loss": 0.7566,
+      "step": 5022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.679813776651943e-08,
+      "loss": 0.7774,
+      "step": 5023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6136394888443025e-08,
+      "loss": 0.7477,
+      "step": 5024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5478518693319326e-08,
+      "loss": 0.8274,
+      "step": 5025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.482450943696371e-08,
+      "loss": 0.7658,
+      "step": 5026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4174367373688305e-08,
+      "loss": 0.8188,
+      "step": 5027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3528092756302e-08,
+      "loss": 0.8203,
+      "step": 5028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2885685836109316e-08,
+      "loss": 0.3053,
+      "step": 5029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2247146862911544e-08,
+      "loss": 0.829,
+      "step": 5030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1612476085004526e-08,
+      "loss": 0.7795,
+      "step": 5031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.098167374917973e-08,
+      "loss": 0.8073,
+      "step": 5032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.03547401007276e-08,
+      "loss": 0.7634,
+      "step": 5033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9731675383432046e-08,
+      "loss": 0.779,
+      "step": 5034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.911247983957035e-08,
+      "loss": 0.8238,
+      "step": 5035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.84971537099177e-08,
+      "loss": 0.8511,
+      "step": 5036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.788569723374603e-08,
+      "loss": 0.827,
+      "step": 5037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.727811064882071e-08,
+      "loss": 0.8025,
+      "step": 5038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.667439419140163e-08,
+      "loss": 0.7447,
+      "step": 5039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6074548096244346e-08,
+      "loss": 0.729,
+      "step": 5040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5478572596601154e-08,
+      "loss": 0.7852,
+      "step": 5041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4886467924215584e-08,
+      "loss": 0.7454,
+      "step": 5042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4298234309331224e-08,
+      "loss": 0.801,
+      "step": 5043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.371387198068067e-08,
+      "loss": 0.7204,
+      "step": 5044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.313338116549437e-08,
+      "loss": 0.8207,
+      "step": 5045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.255676208949844e-08,
+      "loss": 0.2976,
+      "step": 5046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1984014976911293e-08,
+      "loss": 0.7632,
+      "step": 5047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.141514005044589e-08,
+      "loss": 0.8457,
+      "step": 5048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.085013753130862e-08,
+      "loss": 0.7883,
+      "step": 5049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0289007639203736e-08,
+      "loss": 0.7891,
+      "step": 5050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.973175059232559e-08,
+      "loss": 0.7931,
+      "step": 5051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.917836660736529e-08,
+      "loss": 0.8298,
+      "step": 5052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8628855899506266e-08,
+      "loss": 0.7767,
+      "step": 5053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.808321868242648e-08,
+      "loss": 0.7826,
+      "step": 5054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.754145516829733e-08,
+      "loss": 0.776,
+      "step": 5055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.700356556778473e-08,
+      "loss": 0.8098,
+      "step": 5056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6469550090048043e-08,
+      "loss": 0.7755,
+      "step": 5057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5939408942738955e-08,
+      "loss": 0.3335,
+      "step": 5058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.541314233200255e-08,
+      "loss": 0.7606,
+      "step": 5059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4890750462480695e-08,
+      "loss": 0.8173,
+      "step": 5060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4372233537304236e-08,
+      "loss": 0.7884,
+      "step": 5061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385759175809966e-08,
+      "loss": 0.7894,
+      "step": 5062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3346825324986896e-08,
+      "loss": 0.7455,
+      "step": 5063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.283993443657596e-08,
+      "loss": 0.7687,
+      "step": 5064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.233691928997362e-08,
+      "loss": 0.8423,
+      "step": 5065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.183778008077787e-08,
+      "loss": 0.8244,
+      "step": 5066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1342517003079e-08,
+      "loss": 0.8018,
+      "step": 5067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0851130249460737e-08,
+      "loss": 0.8068,
+      "step": 5068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.036362001099913e-08,
+      "loss": 0.7654,
+      "step": 5069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.987998647726365e-08,
+      "loss": 0.8148,
+      "step": 5070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9400229836314965e-08,
+      "loss": 0.8176,
+      "step": 5071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8924350274708303e-08,
+      "loss": 0.7641,
+      "step": 5072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.845234797748897e-08,
+      "loss": 0.7436,
+      "step": 5073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.798422312819571e-08,
+      "loss": 0.7423,
+      "step": 5074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7519975908859576e-08,
+      "loss": 0.729,
+      "step": 5075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7059606500003945e-08,
+      "loss": 0.7581,
+      "step": 5076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6603115080643393e-08,
+      "loss": 0.7983,
+      "step": 5077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6150501828284826e-08,
+      "loss": 0.7739,
+      "step": 5078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5701766918929672e-08,
+      "loss": 0.8263,
+      "step": 5079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5256910527066136e-08,
+      "loss": 0.7869,
+      "step": 5080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4815932825679177e-08,
+      "loss": 0.7985,
+      "step": 5081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4378833986242745e-08,
+      "loss": 0.3525,
+      "step": 5082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.39456141787231e-08,
+      "loss": 0.8164,
+      "step": 5083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3516273571577708e-08,
+      "loss": 0.7952,
+      "step": 5084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.309081233175747e-08,
+      "loss": 0.313,
+      "step": 5085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2669230624702277e-08,
+      "loss": 0.3382,
+      "step": 5086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.225152861434654e-08,
+      "loss": 0.7896,
+      "step": 5087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.183770646311145e-08,
+      "loss": 0.7862,
+      "step": 5088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.142776433191496e-08,
+      "loss": 0.7967,
+      "step": 5089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.102170238016177e-08,
+      "loss": 0.3351,
+      "step": 5090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0619520765750022e-08,
+      "loss": 0.7527,
+      "step": 5091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0221219645067956e-08,
+      "loss": 0.7612,
+      "step": 5092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9826799172996125e-08,
+      "loss": 0.2714,
+      "step": 5093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9436259502906284e-08,
+      "loss": 0.8242,
+      "step": 5094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9049600786658073e-08,
+      "loss": 0.7789,
+      "step": 5095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8666823174605665e-08,
+      "loss": 0.7811,
+      "step": 5096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8287926815592217e-08,
+      "loss": 0.7691,
+      "step": 5097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7912911856952098e-08,
+      "loss": 0.7746,
+      "step": 5098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7541778444509773e-08,
+      "loss": 0.7928,
+      "step": 5099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.717452672258202e-08,
+      "loss": 0.8187,
+      "step": 5100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6811156833974605e-08,
+      "loss": 0.7474,
+      "step": 5101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.64516689199834e-08,
+      "loss": 0.7854,
+      "step": 5102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6096063120396576e-08,
+      "loss": 0.8206,
+      "step": 5103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574433957349242e-08,
+      "loss": 0.842,
+      "step": 5104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5396498416038186e-08,
+      "loss": 0.7628,
+      "step": 5105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5052539783292353e-08,
+      "loss": 0.8432,
+      "step": 5106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4712463809004596e-08,
+      "loss": 0.7391,
+      "step": 5107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.437627062541136e-08,
+      "loss": 0.3068,
+      "step": 5108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4043960363244735e-08,
+      "loss": 0.8114,
+      "step": 5109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.371553315172247e-08,
+      "loss": 0.7619,
+      "step": 5110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3390989118554631e-08,
+      "loss": 0.773,
+      "step": 5111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3070328389939157e-08,
+      "loss": 0.8151,
+      "step": 5112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2753551090566307e-08,
+      "loss": 0.8392,
+      "step": 5113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2440657343615325e-08,
+      "loss": 0.7669,
+      "step": 5114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2131647270754443e-08,
+      "loss": 0.3555,
+      "step": 5115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1826520992144208e-08,
+      "loss": 0.7591,
+      "step": 5116
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1525278626431935e-08,
+      "loss": 0.7373,
+      "step": 5117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1227920290757255e-08,
+      "loss": 0.7799,
+      "step": 5118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0934446100746566e-08,
+      "loss": 0.7939,
+      "step": 5119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.064485617051969e-08,
+      "loss": 0.7287,
+      "step": 5120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0359150612682112e-08,
+      "loss": 0.805,
+      "step": 5121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0077329538331626e-08,
+      "loss": 0.8008,
+      "step": 5122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.799393057056128e-09,
+      "loss": 0.8016,
+      "step": 5123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.525341276929167e-09,
+      "loss": 0.8182,
+      "step": 5124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.255174304516611e-09,
+      "loss": 0.8201,
+      "step": 5125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.988892244874425e-09,
+      "loss": 0.7964,
+      "step": 5126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.726495201545338e-09,
+      "loss": 0.8356,
+      "step": 5127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.467983276563285e-09,
+      "loss": 0.8634,
+      "step": 5128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.213356570452303e-09,
+      "loss": 0.8526,
+      "step": 5129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.962615182222078e-09,
+      "loss": 0.7405,
+      "step": 5130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.715759209373509e-09,
+      "loss": 0.8421,
+      "step": 5131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.472788747899806e-09,
+      "loss": 0.779,
+      "step": 5132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.233703892277622e-09,
+      "loss": 0.87,
+      "step": 5133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.998504735475919e-09,
+      "loss": 0.7686,
+      "step": 5134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.767191368952653e-09,
+      "loss": 0.7608,
+      "step": 5135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.539763882653649e-09,
+      "loss": 0.8287,
+      "step": 5136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.316222365014834e-09,
+      "loss": 0.8255,
+      "step": 5137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.0965669029611205e-09,
+      "loss": 0.7718,
+      "step": 5138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.880797581904185e-09,
+      "loss": 0.3216,
+      "step": 5139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.668914485748022e-09,
+      "loss": 0.7948,
+      "step": 5140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.460917696882284e-09,
+      "loss": 0.7612,
+      "step": 5141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.256807296187827e-09,
+      "loss": 0.7544,
+      "step": 5142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.056583363032275e-09,
+      "loss": 0.8106,
+      "step": 5143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.860245975273348e-09,
+      "loss": 0.8648,
+      "step": 5144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6677952092577525e-09,
+      "loss": 0.8034,
+      "step": 5145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4792311398178525e-09,
+      "loss": 0.29,
+      "step": 5146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2945538402794365e-09,
+      "loss": 0.7785,
+      "step": 5147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.11376338245395e-09,
+      "loss": 0.8463,
+      "step": 5148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.936859836641826e-09,
+      "loss": 0.7554,
+      "step": 5149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.763843271631373e-09,
+      "loss": 0.806,
+      "step": 5150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.594713754702106e-09,
+      "loss": 0.7559,
+      "step": 5151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4294713516180854e-09,
+      "loss": 0.7826,
+      "step": 5152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2681161266356898e-09,
+      "loss": 0.762,
+      "step": 5153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1106481424969524e-09,
+      "loss": 0.7184,
+      "step": 5154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9570674604340046e-09,
+      "loss": 0.7798,
+      "step": 5155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.807374140166852e-09,
+      "loss": 0.8422,
+      "step": 5156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6615682399044883e-09,
+      "loss": 0.7458,
+      "step": 5157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.519649816342673e-09,
+      "loss": 0.7683,
+      "step": 5158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3816189246672616e-09,
+      "loss": 0.7614,
+      "step": 5159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.247475618550876e-09,
+      "loss": 0.8144,
+      "step": 5160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1172199501573453e-09,
+      "loss": 0.8295,
+      "step": 5161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9908519701339337e-09,
+      "loss": 0.7285,
+      "step": 5162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.868371727620222e-09,
+      "loss": 0.3355,
+      "step": 5163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7497792702436677e-09,
+      "loss": 0.7638,
+      "step": 5164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6350746441173848e-09,
+      "loss": 0.7711,
+      "step": 5165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5242578938468034e-09,
+      "loss": 0.7626,
+      "step": 5166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4173290625207891e-09,
+      "loss": 0.8063,
+      "step": 5167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.314288191719415e-09,
+      "loss": 0.6762,
+      "step": 5168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.21513532151063e-09,
+      "loss": 0.8363,
+      "step": 5169
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.119870490450259e-09,
+      "loss": 0.7803,
+      "step": 5170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0284937355820035e-09,
+      "loss": 0.8354,
+      "step": 5171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.410050924374414e-10,
+      "loss": 0.7942,
+      "step": 5172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.574045950360266e-10,
+      "loss": 0.7756,
+      "step": 5173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.776922758884198e-10,
+      "loss": 0.7867,
+      "step": 5174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.01868165987607e-10,
+      "loss": 0.8023,
+      "step": 5175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.299322948211118e-10,
+      "loss": 0.7894,
+      "step": 5176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.618846903576724e-10,
+      "loss": 0.8131,
+      "step": 5177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.977253790605652e-10,
+      "loss": 0.7823,
+      "step": 5178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3745438587761145e-10,
+      "loss": 0.8398,
+      "step": 5179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8107173424450915e-10,
+      "loss": 0.7631,
+      "step": 5180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.285774460870528e-10,
+      "loss": 0.7882,
+      "step": 5181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.799715418155824e-10,
+      "loss": 0.7883,
+      "step": 5182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.352540403327552e-10,
+      "loss": 0.7678,
+      "step": 5183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9442495902577408e-10,
+      "loss": 0.7756,
+      "step": 5184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5748431377082817e-10,
+      "loss": 0.7821,
+      "step": 5185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2443211893309326e-10,
+      "loss": 0.7268,
+      "step": 5186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.526838736451105e-11,
+      "loss": 0.7709,
+      "step": 5187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.999313040600974e-11,
+      "loss": 0.751,
+      "step": 5188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.860635788528356e-11,
+      "loss": 0.7522,
+      "step": 5189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1108078119013175e-11,
+      "loss": 0.8107,
+      "step": 5190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.749829791064528e-11,
+      "loss": 0.7929,
+      "step": 5191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.777022553723258e-12,
+      "loss": 0.8684,
+      "step": 5192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.944255827446284e-12,
+      "loss": 0.8356,
+      "step": 5193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.5325,
+      "step": 5194
+    },
+    {
+      "epoch": 1.0,
+      "step": 5194,
+      "total_flos": 1.0343410282921984e+16,
+      "train_loss": 0.8150010557876076,
+      "train_runtime": 69657.047,
+      "train_samples_per_second": 9.545,
+      "train_steps_per_second": 0.075
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5194,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 5000,
+  "total_flos": 1.0343410282921984e+16,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}