diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9045 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8666137069401314,
+  "eval_steps": 500,
+  "global_step": 1500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.41025641025641e-08,
+      "loss": 2.943,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-07,
+      "loss": 2.9886,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-07,
+      "loss": 2.9708,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-07,
+      "loss": 3.0867,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.205128205128205e-07,
+      "loss": 3.1046,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 3.106,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4871794871794876e-07,
+      "loss": 3.1017,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 2.8961,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.76923076923077e-07,
+      "loss": 2.944,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.41025641025641e-07,
+      "loss": 2.9741,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.051282051282052e-07,
+      "loss": 2.9144,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 2.7468,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 2.7316,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.974358974358975e-07,
+      "loss": 2.8715,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615384615384617e-07,
+      "loss": 2.6562,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 2.4969,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0897435897435899e-06,
+      "loss": 2.5151,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 2.2567,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.217948717948718e-06,
+      "loss": 2.2057,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 2.2083,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3461538461538462e-06,
+      "loss": 2.0459,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4102564102564104e-06,
+      "loss": 2.0801,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4743589743589745e-06,
+      "loss": 2.1923,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.8564,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.602564102564103e-06,
+      "loss": 1.7679,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.9021,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7307692307692308e-06,
+      "loss": 1.5449,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.794871794871795e-06,
+      "loss": 1.6172,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8589743589743592e-06,
+      "loss": 1.7403,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 1.784,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.987179487179487e-06,
+      "loss": 1.6549,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 1.6599,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1153846153846155e-06,
+      "loss": 1.5515,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1794871794871797e-06,
+      "loss": 1.4204,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.243589743589744e-06,
+      "loss": 1.7113,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.4683,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.371794871794872e-06,
+      "loss": 1.4036,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.435897435897436e-06,
+      "loss": 1.3254,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5e-06,
+      "loss": 1.4671,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 1.4896,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6282051282051286e-06,
+      "loss": 1.494,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 1.4019,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.756410256410257e-06,
+      "loss": 1.4106,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.8205128205128207e-06,
+      "loss": 1.4735,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.8846153846153845e-06,
+      "loss": 1.3197,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.948717948717949e-06,
+      "loss": 1.2935,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.012820512820513e-06,
+      "loss": 1.4421,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.2434,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.141025641025641e-06,
+      "loss": 1.4088,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.205128205128206e-06,
+      "loss": 1.2168,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2692307692307696e-06,
+      "loss": 1.2435,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.3046,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.397435897435898e-06,
+      "loss": 1.3269,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 1.1985,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5256410256410263e-06,
+      "loss": 1.4245,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 1.2026,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.653846153846154e-06,
+      "loss": 1.3379,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7179487179487184e-06,
+      "loss": 1.2775,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.782051282051282e-06,
+      "loss": 1.24,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.4219,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.910256410256411e-06,
+      "loss": 1.2447,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.974358974358974e-06,
+      "loss": 1.1747,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.0384615384615385e-06,
+      "loss": 1.0573,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 1.4768,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.2404,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 1.1661,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.294871794871795e-06,
+      "loss": 1.038,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.358974358974359e-06,
+      "loss": 1.2422,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.423076923076924e-06,
+      "loss": 1.1458,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.487179487179488e-06,
+      "loss": 1.0668,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.551282051282052e-06,
+      "loss": 1.1216,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.1612,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.6794871794871795e-06,
+      "loss": 1.2188,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.743589743589744e-06,
+      "loss": 1.3238,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.807692307692308e-06,
+      "loss": 1.3555,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 1.2149,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935897435897436e-06,
+      "loss": 1.0732,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5e-06,
+      "loss": 1.1334,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.064102564102565e-06,
+      "loss": 1.3419,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 1.0512,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.192307692307693e-06,
+      "loss": 1.3568,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.256410256410257e-06,
+      "loss": 1.1892,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.320512820512821e-06,
+      "loss": 1.2229,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 1.1141,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.448717948717949e-06,
+      "loss": 1.1641,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.512820512820514e-06,
+      "loss": 1.213,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.576923076923077e-06,
+      "loss": 1.2778,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 1.2092,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.705128205128206e-06,
+      "loss": 1.2763,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 1.1595,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 1.2018,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.897435897435898e-06,
+      "loss": 1.2114,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.961538461538462e-06,
+      "loss": 1.4324,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.025641025641026e-06,
+      "loss": 1.2207,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.08974358974359e-06,
+      "loss": 1.2855,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.0656,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.217948717948718e-06,
+      "loss": 1.2124,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.282051282051282e-06,
+      "loss": 1.2272,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.3461538461538466e-06,
+      "loss": 1.1834,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.410256410256412e-06,
+      "loss": 1.2152,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.474358974358975e-06,
+      "loss": 1.1499,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 1.0663,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.602564102564103e-06,
+      "loss": 1.1776,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.1595,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.730769230769232e-06,
+      "loss": 1.0708,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.794871794871796e-06,
+      "loss": 1.0552,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.858974358974359e-06,
+      "loss": 1.1106,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 1.1568,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.9871794871794876e-06,
+      "loss": 1.1169,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.051282051282053e-06,
+      "loss": 1.2895,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.115384615384616e-06,
+      "loss": 1.0976,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.17948717948718e-06,
+      "loss": 1.2333,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.243589743589744e-06,
+      "loss": 1.0838,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 1.145,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.371794871794873e-06,
+      "loss": 1.2185,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.435897435897437e-06,
+      "loss": 0.9915,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.1835,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.564102564102564e-06,
+      "loss": 0.9148,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.6282051282051286e-06,
+      "loss": 0.9414,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.0405,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.756410256410258e-06,
+      "loss": 1.1872,
+      "step": 121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.820512820512822e-06,
+      "loss": 1.2754,
+      "step": 122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.884615384615384e-06,
+      "loss": 1.2621,
+      "step": 123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.948717948717949e-06,
+      "loss": 1.2202,
+      "step": 124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.012820512820515e-06,
+      "loss": 1.0307,
+      "step": 125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 1.0905,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.141025641025641e-06,
+      "loss": 1.0909,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 1.1095,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.26923076923077e-06,
+      "loss": 1.1643,
+      "step": 129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.1046,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.397435897435898e-06,
+      "loss": 1.141,
+      "step": 131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 1.1293,
+      "step": 132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.525641025641026e-06,
+      "loss": 1.3018,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.58974358974359e-06,
+      "loss": 1.0179,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.653846153846155e-06,
+      "loss": 1.0412,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 1.0517,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.782051282051283e-06,
+      "loss": 1.1737,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 1.1861,
+      "step": 138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.910256410256411e-06,
+      "loss": 1.1194,
+      "step": 139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 1.1739,
+      "step": 140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.03846153846154e-06,
+      "loss": 1.0401,
+      "step": 141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.102564102564104e-06,
+      "loss": 1.0247,
+      "step": 142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 1.2417,
+      "step": 143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.2071,
+      "step": 144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.294871794871796e-06,
+      "loss": 1.0726,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.358974358974359e-06,
+      "loss": 1.1414,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.423076923076923e-06,
+      "loss": 1.1217,
+      "step": 147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.487179487179487e-06,
+      "loss": 0.9335,
+      "step": 148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.551282051282053e-06,
+      "loss": 1.0924,
+      "step": 149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 1.2216,
+      "step": 150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.67948717948718e-06,
+      "loss": 1.1608,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 1.2329,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.807692307692308e-06,
+      "loss": 1.1358,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.871794871794872e-06,
+      "loss": 1.1753,
+      "step": 154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.935897435897437e-06,
+      "loss": 1.0027,
+      "step": 155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1e-05,
+      "loss": 1.1274,
+      "step": 156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.999999026326575e-06,
+      "loss": 1.2214,
+      "step": 157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.999996105306673e-06,
+      "loss": 1.1085,
+      "step": 158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.999991236941436e-06,
+      "loss": 1.2267,
+      "step": 159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.999984421232756e-06,
+      "loss": 1.0595,
+      "step": 160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.999975658183292e-06,
+      "loss": 0.9717,
+      "step": 161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.999964947796453e-06,
+      "loss": 1.004,
+      "step": 162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.999952290076414e-06,
+      "loss": 1.208,
+      "step": 163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.999937685028101e-06,
+      "loss": 1.0259,
+      "step": 164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999921132657206e-06,
+      "loss": 1.2556,
+      "step": 165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999902632970174e-06,
+      "loss": 1.2101,
+      "step": 166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999882185974208e-06,
+      "loss": 1.2141,
+      "step": 167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999859791677274e-06,
+      "loss": 0.9322,
+      "step": 168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999835450088093e-06,
+      "loss": 1.0845,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999809161216145e-06,
+      "loss": 1.1338,
+      "step": 170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.99978092507167e-06,
+      "loss": 1.0232,
+      "step": 171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999750741665664e-06,
+      "loss": 0.972,
+      "step": 172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999718611009884e-06,
+      "loss": 0.9978,
+      "step": 173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999684533116843e-06,
+      "loss": 1.161,
+      "step": 174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.99964850799981e-06,
+      "loss": 1.0792,
+      "step": 175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999610535672822e-06,
+      "loss": 0.9918,
+      "step": 176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999570616150664e-06,
+      "loss": 1.0092,
+      "step": 177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999528749448883e-06,
+      "loss": 1.084,
+      "step": 178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999484935583787e-06,
+      "loss": 1.1234,
+      "step": 179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999439174572441e-06,
+      "loss": 1.1591,
+      "step": 180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999391466432663e-06,
+      "loss": 1.1137,
+      "step": 181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999341811183036e-06,
+      "loss": 0.9278,
+      "step": 182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999290208842902e-06,
+      "loss": 1.1539,
+      "step": 183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999236659432357e-06,
+      "loss": 0.9506,
+      "step": 184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999181162972253e-06,
+      "loss": 1.1226,
+      "step": 185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999123719484209e-06,
+      "loss": 1.048,
+      "step": 186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999064328990596e-06,
+      "loss": 1.078,
+      "step": 187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999002991514544e-06,
+      "loss": 1.0981,
+      "step": 188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998939707079943e-06,
+      "loss": 1.2173,
+      "step": 189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998874475711439e-06,
+      "loss": 0.9361,
+      "step": 190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998807297434441e-06,
+      "loss": 1.0418,
+      "step": 191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99873817227511e-06,
+      "loss": 1.1751,
+      "step": 192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998667100260367e-06,
+      "loss": 0.9185,
+      "step": 193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998594081417894e-06,
+      "loss": 1.0563,
+      "step": 194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99851911577613e-06,
+      "loss": 1.0217,
+      "step": 195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998442203364271e-06,
+      "loss": 1.1451,
+      "step": 196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998363344212275e-06,
+      "loss": 1.1525,
+      "step": 197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998282538350849e-06,
+      "loss": 1.0686,
+      "step": 198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998199785811472e-06,
+      "loss": 1.1493,
+      "step": 199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.998115086626367e-06,
+      "loss": 1.0811,
+      "step": 200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.998028440828524e-06,
+      "loss": 1.0354,
+      "step": 201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99793984845169e-06,
+      "loss": 1.1304,
+      "step": 202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.997849309530369e-06,
+      "loss": 1.2232,
+      "step": 203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.997756824099822e-06,
+      "loss": 1.1298,
+      "step": 204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99766239219607e-06,
+      "loss": 0.9466,
+      "step": 205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.997566013855891e-06,
+      "loss": 1.0656,
+      "step": 206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99746768911682e-06,
+      "loss": 1.0893,
+      "step": 207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.997367418017154e-06,
+      "loss": 1.145,
+      "step": 208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.997265200595944e-06,
+      "loss": 0.9033,
+      "step": 209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.997161036893001e-06,
+      "loss": 1.1995,
+      "step": 210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.997054926948895e-06,
+      "loss": 1.1877,
+      "step": 211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99694687080495e-06,
+      "loss": 0.918,
+      "step": 212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996836868503253e-06,
+      "loss": 1.2247,
+      "step": 213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996724920086643e-06,
+      "loss": 1.0695,
+      "step": 214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996611025598723e-06,
+      "loss": 0.9076,
+      "step": 215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996495185083853e-06,
+      "loss": 1.0048,
+      "step": 216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.996377398587146e-06,
+      "loss": 1.1376,
+      "step": 217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.996257666154477e-06,
+      "loss": 1.0863,
+      "step": 218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.996135987832481e-06,
+      "loss": 1.2174,
+      "step": 219
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.996012363668544e-06,
+      "loss": 0.9368,
+      "step": 220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.995886793710816e-06,
+      "loss": 1.1135,
+      "step": 221
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.995759278008202e-06,
+      "loss": 1.0947,
+      "step": 222
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.995629816610365e-06,
+      "loss": 1.006,
+      "step": 223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.995498409567727e-06,
+      "loss": 1.1067,
+      "step": 224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.99536505693147e-06,
+      "loss": 1.0684,
+      "step": 225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.995229758753523e-06,
+      "loss": 1.2242,
+      "step": 226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.99509251508659e-06,
+      "loss": 1.0718,
+      "step": 227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.994953325984116e-06,
+      "loss": 1.1542,
+      "step": 228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.994812191500313e-06,
+      "loss": 1.0573,
+      "step": 229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.99466911169015e-06,
+      "loss": 1.2511,
+      "step": 230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.994524086609349e-06,
+      "loss": 0.9884,
+      "step": 231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.994377116314397e-06,
+      "loss": 0.9579,
+      "step": 232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.994228200862532e-06,
+      "loss": 1.1967,
+      "step": 233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.994077340311751e-06,
+      "loss": 1.3001,
+      "step": 234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.993924534720814e-06,
+      "loss": 1.0579,
+      "step": 235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.993769784149229e-06,
+      "loss": 1.0575,
+      "step": 236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.993613088657267e-06,
+      "loss": 1.2556,
+      "step": 237
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.993454448305961e-06,
+      "loss": 1.0135,
+      "step": 238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.993293863157092e-06,
+      "loss": 1.0949,
+      "step": 239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.993131333273203e-06,
+      "loss": 1.0404,
+      "step": 240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.992966858717597e-06,
+      "loss": 1.0815,
+      "step": 241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.992800439554331e-06,
+      "loss": 1.1136,
+      "step": 242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.992632075848218e-06,
+      "loss": 1.264,
+      "step": 243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.992461767664833e-06,
+      "loss": 1.0117,
+      "step": 244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.992289515070506e-06,
+      "loss": 1.2151,
+      "step": 245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.99211531813232e-06,
+      "loss": 1.088,
+      "step": 246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.991939176918127e-06,
+      "loss": 1.1497,
+      "step": 247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.99176109149652e-06,
+      "loss": 0.8865,
+      "step": 248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.991581061936864e-06,
+      "loss": 1.2291,
+      "step": 249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.991399088309273e-06,
+      "loss": 1.0176,
+      "step": 250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.991215170684618e-06,
+      "loss": 1.0204,
+      "step": 251
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.991029309134533e-06,
+      "loss": 1.2076,
+      "step": 252
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.990841503731403e-06,
+      "loss": 1.0757,
+      "step": 253
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.990651754548374e-06,
+      "loss": 1.0483,
+      "step": 254
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.990460061659346e-06,
+      "loss": 0.9856,
+      "step": 255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.990266425138979e-06,
+      "loss": 0.9875,
+      "step": 256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.990070845062687e-06,
+      "loss": 1.1317,
+      "step": 257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.989873321506643e-06,
+      "loss": 0.9574,
+      "step": 258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.989673854547778e-06,
+      "loss": 1.0106,
+      "step": 259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.989472444263773e-06,
+      "loss": 0.9025,
+      "step": 260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.989269090733078e-06,
+      "loss": 1.149,
+      "step": 261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.98906379403489e-06,
+      "loss": 1.1655,
+      "step": 262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.988856554249163e-06,
+      "loss": 0.9097,
+      "step": 263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.988647371456614e-06,
+      "loss": 0.92,
+      "step": 264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.988436245738714e-06,
+      "loss": 1.1595,
+      "step": 265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.988223177177688e-06,
+      "loss": 0.9177,
+      "step": 266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.98800816585652e-06,
+      "loss": 1.0043,
+      "step": 267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.98779121185895e-06,
+      "loss": 1.1646,
+      "step": 268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.987572315269475e-06,
+      "loss": 1.1096,
+      "step": 269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.987351476173352e-06,
+      "loss": 0.937,
+      "step": 270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.987128694656585e-06,
+      "loss": 1.116,
+      "step": 271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.986903970805944e-06,
+      "loss": 0.9868,
+      "step": 272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.986677304708952e-06,
+      "loss": 1.0705,
+      "step": 273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.986448696453887e-06,
+      "loss": 1.1303,
+      "step": 274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.986218146129789e-06,
+      "loss": 1.1893,
+      "step": 275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.985985653826444e-06,
+      "loss": 1.1013,
+      "step": 276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.985751219634406e-06,
+      "loss": 1.0536,
+      "step": 277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.985514843644978e-06,
+      "loss": 1.0179,
+      "step": 278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.985276525950221e-06,
+      "loss": 1.1292,
+      "step": 279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.985036266642952e-06,
+      "loss": 1.1465,
+      "step": 280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.984794065816744e-06,
+      "loss": 1.0592,
+      "step": 281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.98454992356593e-06,
+      "loss": 1.1181,
+      "step": 282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.984303839985593e-06,
+      "loss": 1.1689,
+      "step": 283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.984055815171576e-06,
+      "loss": 1.289,
+      "step": 284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.983805849220475e-06,
+      "loss": 1.0943,
+      "step": 285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.983553942229649e-06,
+      "loss": 1.0122,
+      "step": 286
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.983300094297205e-06,
+      "loss": 1.2763,
+      "step": 287
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.983044305522007e-06,
+      "loss": 0.959,
+      "step": 288
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.982786576003682e-06,
+      "loss": 0.982,
+      "step": 289
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.982526905842601e-06,
+      "loss": 0.9806,
+      "step": 290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.982265295139906e-06,
+      "loss": 1.1962,
+      "step": 291
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.982001743997477e-06,
+      "loss": 1.1179,
+      "step": 292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.981736252517967e-06,
+      "loss": 0.8824,
+      "step": 293
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.981468820804774e-06,
+      "loss": 1.0198,
+      "step": 294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.98119944896205e-06,
+      "loss": 1.0382,
+      "step": 295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.980928137094716e-06,
+      "loss": 0.9761,
+      "step": 296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.980654885308433e-06,
+      "loss": 1.1228,
+      "step": 297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.980379693709626e-06,
+      "loss": 1.1449,
+      "step": 298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.980102562405474e-06,
+      "loss": 1.199,
+      "step": 299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.979823491503909e-06,
+      "loss": 1.0364,
+      "step": 300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.979542481113625e-06,
+      "loss": 1.2125,
+      "step": 301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.979259531344062e-06,
+      "loss": 1.1563,
+      "step": 302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.978974642305424e-06,
+      "loss": 1.1152,
+      "step": 303
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.978687814108666e-06,
+      "loss": 1.0776,
+      "step": 304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.978399046865498e-06,
+      "loss": 1.0826,
+      "step": 305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.978108340688383e-06,
+      "loss": 1.07,
+      "step": 306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.977815695690547e-06,
+      "loss": 1.1216,
+      "step": 307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.977521111985965e-06,
+      "loss": 1.0197,
+      "step": 308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.977224589689366e-06,
+      "loss": 0.9882,
+      "step": 309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.976926128916242e-06,
+      "loss": 1.0157,
+      "step": 310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.976625729782827e-06,
+      "loss": 1.08,
+      "step": 311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.976323392406122e-06,
+      "loss": 1.1745,
+      "step": 312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.976019116903875e-06,
+      "loss": 1.1643,
+      "step": 313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.975712903394597e-06,
+      "loss": 1.0343,
+      "step": 314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.975404751997543e-06,
+      "loss": 1.1074,
+      "step": 315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.975094662832732e-06,
+      "loss": 0.952,
+      "step": 316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.974782636020933e-06,
+      "loss": 0.9747,
+      "step": 317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.974468671683673e-06,
+      "loss": 1.0048,
+      "step": 318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.974152769943227e-06,
+      "loss": 0.9961,
+      "step": 319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.973834930922634e-06,
+      "loss": 1.0847,
+      "step": 320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.973515154745679e-06,
+      "loss": 1.1878,
+      "step": 321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.973193441536905e-06,
+      "loss": 1.0162,
+      "step": 322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.972869791421615e-06,
+      "loss": 1.0508,
+      "step": 323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.972544204525853e-06,
+      "loss": 0.847,
+      "step": 324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.97221668097643e-06,
+      "loss": 1.0445,
+      "step": 325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.971887220900907e-06,
+      "loss": 0.8559,
+      "step": 326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.971555824427594e-06,
+      "loss": 0.9151,
+      "step": 327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.971222491685564e-06,
+      "loss": 1.1306,
+      "step": 328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.970887222804637e-06,
+      "loss": 1.0244,
+      "step": 329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.970550017915393e-06,
+      "loss": 1.1952,
+      "step": 330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.97021087714916e-06,
+      "loss": 1.1184,
+      "step": 331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.969869800638026e-06,
+      "loss": 1.1313,
+      "step": 332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.969526788514827e-06,
+      "loss": 1.016,
+      "step": 333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.969181840913159e-06,
+      "loss": 0.8947,
+      "step": 334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.968834957967363e-06,
+      "loss": 1.1373,
+      "step": 335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.968486139812544e-06,
+      "loss": 1.1395,
+      "step": 336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.968135386584554e-06,
+      "loss": 1.0279,
+      "step": 337
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.967782698420004e-06,
+      "loss": 1.0045,
+      "step": 338
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.967428075456248e-06,
+      "loss": 1.0933,
+      "step": 339
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.967071517831411e-06,
+      "loss": 1.0829,
+      "step": 340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.966713025684351e-06,
+      "loss": 1.112,
+      "step": 341
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.966352599154697e-06,
+      "loss": 0.9315,
+      "step": 342
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.96599023838282e-06,
+      "loss": 1.1093,
+      "step": 343
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.965625943509851e-06,
+      "loss": 1.0793,
+      "step": 344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.965259714677668e-06,
+      "loss": 0.9675,
+      "step": 345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.964891552028911e-06,
+      "loss": 1.0486,
+      "step": 346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.964521455706965e-06,
+      "loss": 0.958,
+      "step": 347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.964149425855971e-06,
+      "loss": 1.0303,
+      "step": 348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.963775462620825e-06,
+      "loss": 1.0611,
+      "step": 349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.963399566147172e-06,
+      "loss": 1.1357,
+      "step": 350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.963021736581413e-06,
+      "loss": 1.0651,
+      "step": 351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.962641974070702e-06,
+      "loss": 0.9568,
+      "step": 352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.962260278762946e-06,
+      "loss": 1.041,
+      "step": 353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.961876650806799e-06,
+      "loss": 1.2201,
+      "step": 354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.961491090351676e-06,
+      "loss": 0.9089,
+      "step": 355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.961103597547741e-06,
+      "loss": 1.1417,
+      "step": 356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.96071417254591e-06,
+      "loss": 1.2027,
+      "step": 357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.960322815497852e-06,
+      "loss": 0.9323,
+      "step": 358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.959929526555987e-06,
+      "loss": 0.8827,
+      "step": 359
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.95953430587349e-06,
+      "loss": 1.0305,
+      "step": 360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.95913715360429e-06,
+      "loss": 0.9028,
+      "step": 361
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.958738069903062e-06,
+      "loss": 1.0454,
+      "step": 362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.958337054925239e-06,
+      "loss": 1.1861,
+      "step": 363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.957934108827003e-06,
+      "loss": 0.8635,
+      "step": 364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.957529231765289e-06,
+      "loss": 1.1306,
+      "step": 365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.957122423897786e-06,
+      "loss": 1.0467,
+      "step": 366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.956713685382931e-06,
+      "loss": 1.0476,
+      "step": 367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.956303016379916e-06,
+      "loss": 1.128,
+      "step": 368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.955890417048686e-06,
+      "loss": 1.0823,
+      "step": 369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.955475887549933e-06,
+      "loss": 0.9169,
+      "step": 370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.955059428045102e-06,
+      "loss": 1.1195,
+      "step": 371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.954641038696395e-06,
+      "loss": 1.1197,
+      "step": 372
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.954220719666761e-06,
+      "loss": 0.9047,
+      "step": 373
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.953798471119902e-06,
+      "loss": 1.0175,
+      "step": 374
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.953374293220268e-06,
+      "loss": 1.0745,
+      "step": 375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.952948186133066e-06,
+      "loss": 1.1427,
+      "step": 376
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.952520150024251e-06,
+      "loss": 1.0006,
+      "step": 377
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.952090185060528e-06,
+      "loss": 0.9272,
+      "step": 378
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.951658291409358e-06,
+      "loss": 0.9994,
+      "step": 379
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.951224469238949e-06,
+      "loss": 1.041,
+      "step": 380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.950788718718262e-06,
+      "loss": 1.0084,
+      "step": 381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.950351040017007e-06,
+      "loss": 1.1663,
+      "step": 382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.94991143330565e-06,
+      "loss": 0.9753,
+      "step": 383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.9494698987554e-06,
+      "loss": 1.2382,
+      "step": 384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.949026436538223e-06,
+      "loss": 1.0753,
+      "step": 385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.948581046826835e-06,
+      "loss": 1.0164,
+      "step": 386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.9481337297947e-06,
+      "loss": 1.202,
+      "step": 387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.947684485616036e-06,
+      "loss": 1.2132,
+      "step": 388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.947233314465807e-06,
+      "loss": 1.0343,
+      "step": 389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.946780216519734e-06,
+      "loss": 1.1803,
+      "step": 390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.946325191954283e-06,
+      "loss": 1.0142,
+      "step": 391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.94586824094667e-06,
+      "loss": 1.1111,
+      "step": 392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.945409363674867e-06,
+      "loss": 1.2319,
+      "step": 393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.944948560317592e-06,
+      "loss": 1.0264,
+      "step": 394
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.944485831054311e-06,
+      "loss": 0.9998,
+      "step": 395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.944021176065247e-06,
+      "loss": 1.1853,
+      "step": 396
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.943554595531364e-06,
+      "loss": 0.938,
+      "step": 397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.943086089634387e-06,
+      "loss": 1.1229,
+      "step": 398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.942615658556778e-06,
+      "loss": 1.0387,
+      "step": 399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.942143302481759e-06,
+      "loss": 0.9583,
+      "step": 400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.9416690215933e-06,
+      "loss": 1.0944,
+      "step": 401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.941192816076114e-06,
+      "loss": 1.0373,
+      "step": 402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.940714686115674e-06,
+      "loss": 1.0935,
+      "step": 403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.940234631898193e-06,
+      "loss": 1.1676,
+      "step": 404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.939752653610639e-06,
+      "loss": 1.1275,
+      "step": 405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.939268751440728e-06,
+      "loss": 0.9004,
+      "step": 406
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.938782925576925e-06,
+      "loss": 1.1214,
+      "step": 407
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.938295176208441e-06,
+      "loss": 0.9776,
+      "step": 408
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.937805503525244e-06,
+      "loss": 1.0831,
+      "step": 409
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.937313907718046e-06,
+      "loss": 1.0985,
+      "step": 410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.936820388978306e-06,
+      "loss": 0.9818,
+      "step": 411
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.936324947498237e-06,
+      "loss": 1.1852,
+      "step": 412
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.935827583470793e-06,
+      "loss": 0.8824,
+      "step": 413
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.93532829708969e-06,
+      "loss": 1.0186,
+      "step": 414
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.93482708854938e-06,
+      "loss": 1.0661,
+      "step": 415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.934323958045069e-06,
+      "loss": 1.1771,
+      "step": 416
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.933818905772713e-06,
+      "loss": 1.1797,
+      "step": 417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.933311931929008e-06,
+      "loss": 0.9796,
+      "step": 418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.932803036711413e-06,
+      "loss": 1.2804,
+      "step": 419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.932292220318121e-06,
+      "loss": 0.9236,
+      "step": 420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.931779482948083e-06,
+      "loss": 0.8463,
+      "step": 421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.931264824800995e-06,
+      "loss": 1.0292,
+      "step": 422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.930748246077296e-06,
+      "loss": 1.0739,
+      "step": 423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.930229746978181e-06,
+      "loss": 1.0623,
+      "step": 424
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.92970932770559e-06,
+      "loss": 1.0589,
+      "step": 425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.929186988462208e-06,
+      "loss": 1.023,
+      "step": 426
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.928662729451472e-06,
+      "loss": 1.0123,
+      "step": 427
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.928136550877565e-06,
+      "loss": 1.1057,
+      "step": 428
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.927608452945413e-06,
+      "loss": 0.9542,
+      "step": 429
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.927078435860702e-06,
+      "loss": 1.0872,
+      "step": 430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.926546499829853e-06,
+      "loss": 0.9379,
+      "step": 431
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.926012645060037e-06,
+      "loss": 1.1088,
+      "step": 432
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.925476871759177e-06,
+      "loss": 0.9663,
+      "step": 433
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.924939180135938e-06,
+      "loss": 1.1775,
+      "step": 434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.924399570399737e-06,
+      "loss": 1.0455,
+      "step": 435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.923858042760732e-06,
+      "loss": 1.0095,
+      "step": 436
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.923314597429834e-06,
+      "loss": 1.0482,
+      "step": 437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.9227692346187e-06,
+      "loss": 0.862,
+      "step": 438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.922221954539728e-06,
+      "loss": 1.0475,
+      "step": 439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.92167275740607e-06,
+      "loss": 1.0089,
+      "step": 440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.92112164343162e-06,
+      "loss": 1.0188,
+      "step": 441
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.920568612831021e-06,
+      "loss": 0.9698,
+      "step": 442
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.92001366581966e-06,
+      "loss": 0.9601,
+      "step": 443
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.919456802613672e-06,
+      "loss": 0.8416,
+      "step": 444
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.91889802342994e-06,
+      "loss": 1.2232,
+      "step": 445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.918337328486088e-06,
+      "loss": 0.9872,
+      "step": 446
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.917774718000494e-06,
+      "loss": 1.2299,
+      "step": 447
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.917210192192273e-06,
+      "loss": 1.0318,
+      "step": 448
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.916643751281294e-06,
+      "loss": 1.144,
+      "step": 449
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.916075395488167e-06,
+      "loss": 0.9429,
+      "step": 450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.91550512503425e-06,
+      "loss": 0.9432,
+      "step": 451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.914932940141644e-06,
+      "loss": 1.1643,
+      "step": 452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.9143588410332e-06,
+      "loss": 1.1689,
+      "step": 453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.913782827932509e-06,
+      "loss": 1.0366,
+      "step": 454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.913204901063915e-06,
+      "loss": 1.0424,
+      "step": 455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.912625060652496e-06,
+      "loss": 0.9092,
+      "step": 456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.912043306924088e-06,
+      "loss": 1.0998,
+      "step": 457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.911459640105266e-06,
+      "loss": 0.8732,
+      "step": 458
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.910874060423345e-06,
+      "loss": 0.9607,
+      "step": 459
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.910286568106398e-06,
+      "loss": 1.0134,
+      "step": 460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.909697163383228e-06,
+      "loss": 0.9903,
+      "step": 461
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.909105846483394e-06,
+      "loss": 1.2032,
+      "step": 462
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.908512617637195e-06,
+      "loss": 1.0256,
+      "step": 463
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.907917477075677e-06,
+      "loss": 1.1146,
+      "step": 464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.907320425030627e-06,
+      "loss": 0.8173,
+      "step": 465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.906721461734579e-06,
+      "loss": 1.0522,
+      "step": 466
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.906120587420811e-06,
+      "loss": 1.1911,
+      "step": 467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.905517802323345e-06,
+      "loss": 1.1293,
+      "step": 468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.904913106676948e-06,
+      "loss": 1.0679,
+      "step": 469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.904306500717129e-06,
+      "loss": 1.1274,
+      "step": 470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.903697984680143e-06,
+      "loss": 1.0748,
+      "step": 471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.90308755880299e-06,
+      "loss": 1.0324,
+      "step": 472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.902475223323408e-06,
+      "loss": 1.1286,
+      "step": 473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.901860978479889e-06,
+      "loss": 0.9904,
+      "step": 474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.90124482451166e-06,
+      "loss": 0.8558,
+      "step": 475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.900626761658691e-06,
+      "loss": 1.0362,
+      "step": 476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.900006790161701e-06,
+      "loss": 1.0759,
+      "step": 477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.899384910262151e-06,
+      "loss": 1.0135,
+      "step": 478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.898761122202242e-06,
+      "loss": 0.902,
+      "step": 479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.898135426224923e-06,
+      "loss": 1.1064,
+      "step": 480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.89750782257388e-06,
+      "loss": 1.0156,
+      "step": 481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.896878311493549e-06,
+      "loss": 1.0196,
+      "step": 482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.896246893229102e-06,
+      "loss": 1.0481,
+      "step": 483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.89561356802646e-06,
+      "loss": 0.9186,
+      "step": 484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.894978336132281e-06,
+      "loss": 1.1194,
+      "step": 485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.89434119779397e-06,
+      "loss": 1.1421,
+      "step": 486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.893702153259674e-06,
+      "loss": 1.0624,
+      "step": 487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.893061202778278e-06,
+      "loss": 1.0135,
+      "step": 488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.892418346599415e-06,
+      "loss": 1.0439,
+      "step": 489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.891773584973457e-06,
+      "loss": 1.006,
+      "step": 490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.89112691815152e-06,
+      "loss": 1.1106,
+      "step": 491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.89047834638546e-06,
+      "loss": 1.0478,
+      "step": 492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.889827869927875e-06,
+      "loss": 1.0596,
+      "step": 493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.889175489032107e-06,
+      "loss": 0.991,
+      "step": 494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.888521203952238e-06,
+      "loss": 1.0785,
+      "step": 495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.887865014943093e-06,
+      "loss": 0.8694,
+      "step": 496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.887206922260234e-06,
+      "loss": 1.1631,
+      "step": 497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.886546926159972e-06,
+      "loss": 1.0418,
+      "step": 498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.885885026899353e-06,
+      "loss": 0.9247,
+      "step": 499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.885221224736168e-06,
+      "loss": 1.0714,
+      "step": 500
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 1.1758191585540771,
+      "eval_runtime": 90.1378,
+      "eval_samples_per_second": 17.107,
+      "eval_steps_per_second": 2.141,
+      "step": 500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.884555519928946e-06,
+      "loss": 1.081,
+      "step": 501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.883887912736959e-06,
+      "loss": 0.9773,
+      "step": 502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.88321840342022e-06,
+      "loss": 0.8937,
+      "step": 503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.882546992239483e-06,
+      "loss": 0.8111,
+      "step": 504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.881873679456243e-06,
+      "loss": 1.0577,
+      "step": 505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.88119846533273e-06,
+      "loss": 1.0657,
+      "step": 506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.880521350131925e-06,
+      "loss": 1.0389,
+      "step": 507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.879842334117538e-06,
+      "loss": 0.9929,
+      "step": 508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.879161417554031e-06,
+      "loss": 0.8891,
+      "step": 509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.878478600706595e-06,
+      "loss": 1.0267,
+      "step": 510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.877793883841169e-06,
+      "loss": 0.8922,
+      "step": 511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.877107267224429e-06,
+      "loss": 1.0693,
+      "step": 512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.876418751123788e-06,
+      "loss": 1.0529,
+      "step": 513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.875728335807407e-06,
+      "loss": 0.9663,
+      "step": 514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.87503602154418e-06,
+      "loss": 0.9459,
+      "step": 515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.87434180860374e-06,
+      "loss": 1.0108,
+      "step": 516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.873645697256463e-06,
+      "loss": 1.0826,
+      "step": 517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.872947687773464e-06,
+      "loss": 1.1406,
+      "step": 518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.872247780426595e-06,
+      "loss": 1.1833,
+      "step": 519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.871545975488448e-06,
+      "loss": 0.8866,
+      "step": 520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.870842273232358e-06,
+      "loss": 0.9354,
+      "step": 521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.87013667393239e-06,
+      "loss": 0.93,
+      "step": 522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.869429177863358e-06,
+      "loss": 1.1015,
+      "step": 523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.868719785300808e-06,
+      "loss": 1.0774,
+      "step": 524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.868008496521027e-06,
+      "loss": 1.0039,
+      "step": 525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.86729531180104e-06,
+      "loss": 1.1082,
+      "step": 526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.866580231418612e-06,
+      "loss": 0.8766,
+      "step": 527
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.865863255652242e-06,
+      "loss": 1.1319,
+      "step": 528
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.865144384781171e-06,
+      "loss": 1.0232,
+      "step": 529
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.86442361908538e-06,
+      "loss": 0.9866,
+      "step": 530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.863700958845582e-06,
+      "loss": 1.1781,
+      "step": 531
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.862976404343234e-06,
+      "loss": 1.0595,
+      "step": 532
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.862249955860524e-06,
+      "loss": 0.8039,
+      "step": 533
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.861521613680384e-06,
+      "loss": 1.1911,
+      "step": 534
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.86079137808648e-06,
+      "loss": 0.998,
+      "step": 535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.860059249363217e-06,
+      "loss": 1.1036,
+      "step": 536
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.859325227795736e-06,
+      "loss": 1.0584,
+      "step": 537
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.858589313669918e-06,
+      "loss": 1.0019,
+      "step": 538
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.857851507272376e-06,
+      "loss": 0.942,
+      "step": 539
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.857111808890465e-06,
+      "loss": 0.9628,
+      "step": 540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.856370218812274e-06,
+      "loss": 0.9041,
+      "step": 541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.85562673732663e-06,
+      "loss": 0.9948,
+      "step": 542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.854881364723094e-06,
+      "loss": 0.9932,
+      "step": 543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.854134101291972e-06,
+      "loss": 1.1193,
+      "step": 544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.853384947324293e-06,
+      "loss": 1.0415,
+      "step": 545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.852633903111834e-06,
+      "loss": 1.0647,
+      "step": 546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.851880968947102e-06,
+      "loss": 1.1693,
+      "step": 547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.851126145123341e-06,
+      "loss": 1.0946,
+      "step": 548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.850369431934536e-06,
+      "loss": 1.0398,
+      "step": 549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.849610829675398e-06,
+      "loss": 1.1324,
+      "step": 550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.848850338641382e-06,
+      "loss": 1.0135,
+      "step": 551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.848087959128679e-06,
+      "loss": 1.0106,
+      "step": 552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.847323691434205e-06,
+      "loss": 1.1095,
+      "step": 553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.846557535855625e-06,
+      "loss": 1.0138,
+      "step": 554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.845789492691332e-06,
+      "loss": 1.0235,
+      "step": 555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.845019562240455e-06,
+      "loss": 1.0265,
+      "step": 556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.844247744802857e-06,
+      "loss": 1.1371,
+      "step": 557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.843474040679137e-06,
+      "loss": 1.1075,
+      "step": 558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.842698450170631e-06,
+      "loss": 0.7993,
+      "step": 559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.841920973579408e-06,
+      "loss": 1.008,
+      "step": 560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.84114161120827e-06,
+      "loss": 0.9307,
+      "step": 561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.840360363360755e-06,
+      "loss": 0.9155,
+      "step": 562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.839577230341135e-06,
+      "loss": 0.9678,
+      "step": 563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.838792212454416e-06,
+      "loss": 1.1954,
+      "step": 564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.83800531000634e-06,
+      "loss": 1.0043,
+      "step": 565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.83721652330338e-06,
+      "loss": 1.0244,
+      "step": 566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.836425852652744e-06,
+      "loss": 1.0347,
+      "step": 567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.835633298362377e-06,
+      "loss": 0.9116,
+      "step": 568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.834838860740949e-06,
+      "loss": 0.9405,
+      "step": 569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.834042540097875e-06,
+      "loss": 1.1432,
+      "step": 570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.833244336743293e-06,
+      "loss": 0.9495,
+      "step": 571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.83244425098808e-06,
+      "loss": 1.1235,
+      "step": 572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.831642283143847e-06,
+      "loss": 0.8453,
+      "step": 573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.830838433522934e-06,
+      "loss": 1.2087,
+      "step": 574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.830032702438416e-06,
+      "loss": 0.9393,
+      "step": 575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.829225090204102e-06,
+      "loss": 1.0099,
+      "step": 576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.82841559713453e-06,
+      "loss": 0.7879,
+      "step": 577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.827604223544973e-06,
+      "loss": 0.9673,
+      "step": 578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.826790969751438e-06,
+      "loss": 1.0294,
+      "step": 579
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.825975836070662e-06,
+      "loss": 1.0178,
+      "step": 580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.825158822820113e-06,
+      "loss": 1.1594,
+      "step": 581
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.824339930317994e-06,
+      "loss": 1.2388,
+      "step": 582
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.823519158883237e-06,
+      "loss": 0.9907,
+      "step": 583
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.822696508835511e-06,
+      "loss": 0.8205,
+      "step": 584
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.821871980495208e-06,
+      "loss": 0.8609,
+      "step": 585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.821045574183461e-06,
+      "loss": 0.9305,
+      "step": 586
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.820217290222127e-06,
+      "loss": 1.0216,
+      "step": 587
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.819387128933799e-06,
+      "loss": 0.9592,
+      "step": 588
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.818555090641797e-06,
+      "loss": 0.9982,
+      "step": 589
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.817721175670176e-06,
+      "loss": 0.9477,
+      "step": 590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.816885384343722e-06,
+      "loss": 0.9817,
+      "step": 591
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.81604771698795e-06,
+      "loss": 1.0529,
+      "step": 592
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.8152081739291e-06,
+      "loss": 0.8294,
+      "step": 593
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.814366755494155e-06,
+      "loss": 0.9609,
+      "step": 594
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.813523462010819e-06,
+      "loss": 0.9591,
+      "step": 595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.81267829380753e-06,
+      "loss": 1.0678,
+      "step": 596
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.811831251213453e-06,
+      "loss": 0.8412,
+      "step": 597
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.810982334558487e-06,
+      "loss": 0.9476,
+      "step": 598
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.810131544173258e-06,
+      "loss": 0.9569,
+      "step": 599
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.809278880389126e-06,
+      "loss": 0.9127,
+      "step": 600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.808424343538172e-06,
+      "loss": 0.9523,
+      "step": 601
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.807567933953217e-06,
+      "loss": 0.8829,
+      "step": 602
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.806709651967802e-06,
+      "loss": 1.0579,
+      "step": 603
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.805849497916205e-06,
+      "loss": 1.0728,
+      "step": 604
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.804987472133427e-06,
+      "loss": 1.1204,
+      "step": 605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.804123574955202e-06,
+      "loss": 1.0067,
+      "step": 606
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.803257806717993e-06,
+      "loss": 1.0024,
+      "step": 607
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.802390167758987e-06,
+      "loss": 1.0424,
+      "step": 608
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.801520658416105e-06,
+      "loss": 1.0301,
+      "step": 609
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.800649279027994e-06,
+      "loss": 1.0168,
+      "step": 610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.799776029934029e-06,
+      "loss": 1.0393,
+      "step": 611
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.798900911474315e-06,
+      "loss": 1.092,
+      "step": 612
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.79802392398968e-06,
+      "loss": 0.9596,
+      "step": 613
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.797145067821689e-06,
+      "loss": 0.9582,
+      "step": 614
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.796264343312626e-06,
+      "loss": 0.9319,
+      "step": 615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.795381750805508e-06,
+      "loss": 1.1094,
+      "step": 616
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.794497290644076e-06,
+      "loss": 1.0964,
+      "step": 617
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.793610963172802e-06,
+      "loss": 0.9325,
+      "step": 618
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.792722768736885e-06,
+      "loss": 1.0106,
+      "step": 619
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.791832707682242e-06,
+      "loss": 0.9457,
+      "step": 620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.790940780355534e-06,
+      "loss": 1.008,
+      "step": 621
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.790046987104133e-06,
+      "loss": 0.9542,
+      "step": 622
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.789151328276147e-06,
+      "loss": 1.1266,
+      "step": 623
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.78825380422041e-06,
+      "loss": 0.9439,
+      "step": 624
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.787354415286472e-06,
+      "loss": 1.0654,
+      "step": 625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.786453161824625e-06,
+      "loss": 1.0264,
+      "step": 626
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.785550044185877e-06,
+      "loss": 1.0827,
+      "step": 627
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.784645062721962e-06,
+      "loss": 0.9121,
+      "step": 628
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.783738217785349e-06,
+      "loss": 1.0849,
+      "step": 629
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.78282950972922e-06,
+      "loss": 0.9767,
+      "step": 630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.781918938907494e-06,
+      "loss": 1.0719,
+      "step": 631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.781006505674807e-06,
+      "loss": 0.9645,
+      "step": 632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.780092210386522e-06,
+      "loss": 0.9634,
+      "step": 633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.779176053398736e-06,
+      "loss": 0.9141,
+      "step": 634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.778258035068258e-06,
+      "loss": 1.0154,
+      "step": 635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.77733815575263e-06,
+      "loss": 1.03,
+      "step": 636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.776416415810116e-06,
+      "loss": 0.9659,
+      "step": 637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.775492815599708e-06,
+      "loss": 1.1142,
+      "step": 638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.774567355481116e-06,
+      "loss": 0.8802,
+      "step": 639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.773640035814783e-06,
+      "loss": 1.1137,
+      "step": 640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.772710856961867e-06,
+      "loss": 1.1486,
+      "step": 641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.771779819284257e-06,
+      "loss": 1.0132,
+      "step": 642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.770846923144565e-06,
+      "loss": 0.8972,
+      "step": 643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.769912168906124e-06,
+      "loss": 1.0555,
+      "step": 644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.768975556932991e-06,
+      "loss": 0.8229,
+      "step": 645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.76803708758995e-06,
+      "loss": 0.9453,
+      "step": 646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.767096761242503e-06,
+      "loss": 0.9541,
+      "step": 647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.766154578256883e-06,
+      "loss": 1.0962,
+      "step": 648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.765210539000036e-06,
+      "loss": 1.0105,
+      "step": 649
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.76426464383964e-06,
+      "loss": 1.0209,
+      "step": 650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.763316893144092e-06,
+      "loss": 0.9617,
+      "step": 651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.76236728728251e-06,
+      "loss": 1.0085,
+      "step": 652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.761415826624737e-06,
+      "loss": 0.8521,
+      "step": 653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.76046251154134e-06,
+      "loss": 0.9602,
+      "step": 654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.759507342403604e-06,
+      "loss": 1.0567,
+      "step": 655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.758550319583537e-06,
+      "loss": 0.9179,
+      "step": 656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.757591443453872e-06,
+      "loss": 1.1019,
+      "step": 657
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.756630714388062e-06,
+      "loss": 1.0333,
+      "step": 658
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.755668132760281e-06,
+      "loss": 0.8943,
+      "step": 659
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.754703698945425e-06,
+      "loss": 1.1336,
+      "step": 660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.753737413319109e-06,
+      "loss": 0.9643,
+      "step": 661
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.752769276257676e-06,
+      "loss": 1.2361,
+      "step": 662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.751799288138184e-06,
+      "loss": 1.1315,
+      "step": 663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.750827449338414e-06,
+      "loss": 1.1689,
+      "step": 664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.749853760236866e-06,
+      "loss": 1.0509,
+      "step": 665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.748878221212763e-06,
+      "loss": 0.9875,
+      "step": 666
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.747900832646047e-06,
+      "loss": 0.9205,
+      "step": 667
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.746921594917382e-06,
+      "loss": 0.8165,
+      "step": 668
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.74594050840815e-06,
+      "loss": 0.9914,
+      "step": 669
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.744957573500455e-06,
+      "loss": 0.8208,
+      "step": 670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.74397279057712e-06,
+      "loss": 1.09,
+      "step": 671
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.742986160021688e-06,
+      "loss": 1.0669,
+      "step": 672
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.741997682218421e-06,
+      "loss": 1.0583,
+      "step": 673
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.741007357552298e-06,
+      "loss": 0.9845,
+      "step": 674
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.740015186409027e-06,
+      "loss": 1.0995,
+      "step": 675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.739021169175021e-06,
+      "loss": 0.9932,
+      "step": 676
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.738025306237424e-06,
+      "loss": 1.0708,
+      "step": 677
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.73702759798409e-06,
+      "loss": 1.0282,
+      "step": 678
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.736028044803602e-06,
+      "loss": 1.0041,
+      "step": 679
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.73502664708525e-06,
+      "loss": 1.0263,
+      "step": 680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.734023405219049e-06,
+      "loss": 1.0082,
+      "step": 681
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.73301831959573e-06,
+      "loss": 1.0097,
+      "step": 682
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.732011390606748e-06,
+      "loss": 1.0337,
+      "step": 683
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.731002618644265e-06,
+      "loss": 1.0752,
+      "step": 684
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.729992004101169e-06,
+      "loss": 1.1927,
+      "step": 685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.728979547371065e-06,
+      "loss": 0.9574,
+      "step": 686
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.727965248848273e-06,
+      "loss": 0.9339,
+      "step": 687
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.726949108927828e-06,
+      "loss": 0.9178,
+      "step": 688
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.72593112800549e-06,
+      "loss": 0.9228,
+      "step": 689
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.724911306477729e-06,
+      "loss": 0.952,
+      "step": 690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.723889644741735e-06,
+      "loss": 0.9064,
+      "step": 691
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.722866143195412e-06,
+      "loss": 0.8325,
+      "step": 692
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.721840802237388e-06,
+      "loss": 0.8084,
+      "step": 693
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.720813622266996e-06,
+      "loss": 1.0438,
+      "step": 694
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.719784603684292e-06,
+      "loss": 0.9741,
+      "step": 695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.71875374689005e-06,
+      "loss": 1.0801,
+      "step": 696
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.717721052285754e-06,
+      "loss": 1.1121,
+      "step": 697
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.716686520273611e-06,
+      "loss": 0.946,
+      "step": 698
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.715650151256534e-06,
+      "loss": 0.726,
+      "step": 699
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.714611945638162e-06,
+      "loss": 1.1074,
+      "step": 700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.713571903822842e-06,
+      "loss": 0.9112,
+      "step": 701
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.71253002621564e-06,
+      "loss": 0.9756,
+      "step": 702
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.711486313222332e-06,
+      "loss": 1.1096,
+      "step": 703
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.710440765249416e-06,
+      "loss": 1.0547,
+      "step": 704
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.709393382704098e-06,
+      "loss": 1.0299,
+      "step": 705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.708344165994304e-06,
+      "loss": 0.9274,
+      "step": 706
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.70729311552867e-06,
+      "loss": 1.0014,
+      "step": 707
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.706240231716549e-06,
+      "loss": 1.0566,
+      "step": 708
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.705185514968006e-06,
+      "loss": 0.8526,
+      "step": 709
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.70412896569382e-06,
+      "loss": 0.9816,
+      "step": 710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.703070584305487e-06,
+      "loss": 0.9615,
+      "step": 711
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.702010371215213e-06,
+      "loss": 1.0533,
+      "step": 712
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.70094832683592e-06,
+      "loss": 0.9257,
+      "step": 713
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.699884451581238e-06,
+      "loss": 0.9539,
+      "step": 714
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.698818745865518e-06,
+      "loss": 0.8565,
+      "step": 715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.697751210103817e-06,
+      "loss": 1.0449,
+      "step": 716
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.696681844711909e-06,
+      "loss": 0.9523,
+      "step": 717
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.695610650106278e-06,
+      "loss": 1.0615,
+      "step": 718
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.69453762670412e-06,
+      "loss": 1.004,
+      "step": 719
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.693462774923351e-06,
+      "loss": 1.2029,
+      "step": 720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.692386095182585e-06,
+      "loss": 0.9176,
+      "step": 721
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.691307587901161e-06,
+      "loss": 0.9886,
+      "step": 722
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.690227253499122e-06,
+      "loss": 1.0573,
+      "step": 723
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.689145092397227e-06,
+      "loss": 0.9027,
+      "step": 724
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.688061105016942e-06,
+      "loss": 1.1269,
+      "step": 725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.686975291780449e-06,
+      "loss": 0.9658,
+      "step": 726
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.68588765311064e-06,
+      "loss": 0.9108,
+      "step": 727
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.684798189431114e-06,
+      "loss": 1.0141,
+      "step": 728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.683706901166186e-06,
+      "loss": 1.2133,
+      "step": 729
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.682613788740878e-06,
+      "loss": 1.0399,
+      "step": 730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.681518852580925e-06,
+      "loss": 0.8892,
+      "step": 731
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.68042209311277e-06,
+      "loss": 1.0292,
+      "step": 732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.679323510763568e-06,
+      "loss": 1.0511,
+      "step": 733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.678223105961182e-06,
+      "loss": 1.0881,
+      "step": 734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.677120879134187e-06,
+      "loss": 1.0328,
+      "step": 735
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.676016830711868e-06,
+      "loss": 0.8984,
+      "step": 736
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.674910961124215e-06,
+      "loss": 1.0944,
+      "step": 737
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.67380327080193e-06,
+      "loss": 0.9009,
+      "step": 738
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.67269376017643e-06,
+      "loss": 1.0113,
+      "step": 739
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.671582429679829e-06,
+      "loss": 1.0413,
+      "step": 740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.670469279744959e-06,
+      "loss": 0.9889,
+      "step": 741
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.669354310805359e-06,
+      "loss": 0.9766,
+      "step": 742
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.668237523295272e-06,
+      "loss": 0.9172,
+      "step": 743
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.667118917649656e-06,
+      "loss": 1.0089,
+      "step": 744
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.665998494304172e-06,
+      "loss": 0.9186,
+      "step": 745
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.66487625369519e-06,
+      "loss": 0.9578,
+      "step": 746
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.66375219625979e-06,
+      "loss": 1.0808,
+      "step": 747
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.662626322435755e-06,
+      "loss": 0.9492,
+      "step": 748
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.661498632661582e-06,
+      "loss": 0.8191,
+      "step": 749
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.660369127376469e-06,
+      "loss": 0.8447,
+      "step": 750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.659237807020326e-06,
+      "loss": 1.2476,
+      "step": 751
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.658104672033763e-06,
+      "loss": 0.9947,
+      "step": 752
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.656969722858108e-06,
+      "loss": 1.0914,
+      "step": 753
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.655832959935382e-06,
+      "loss": 0.9976,
+      "step": 754
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.654694383708327e-06,
+      "loss": 1.0859,
+      "step": 755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.653553994620378e-06,
+      "loss": 1.0485,
+      "step": 756
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.652411793115681e-06,
+      "loss": 0.9869,
+      "step": 757
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.651267779639093e-06,
+      "loss": 1.0297,
+      "step": 758
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.650121954636169e-06,
+      "loss": 1.1313,
+      "step": 759
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.648974318553173e-06,
+      "loss": 0.9502,
+      "step": 760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.647824871837076e-06,
+      "loss": 1.1985,
+      "step": 761
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.64667361493555e-06,
+      "loss": 0.9875,
+      "step": 762
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.645520548296975e-06,
+      "loss": 0.926,
+      "step": 763
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.644365672370437e-06,
+      "loss": 1.0709,
+      "step": 764
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.643208987605722e-06,
+      "loss": 0.9883,
+      "step": 765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.642050494453323e-06,
+      "loss": 0.9756,
+      "step": 766
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.64089019336444e-06,
+      "loss": 0.8874,
+      "step": 767
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.639728084790976e-06,
+      "loss": 0.9945,
+      "step": 768
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.638564169185532e-06,
+      "loss": 0.8,
+      "step": 769
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.63739844700142e-06,
+      "loss": 0.9862,
+      "step": 770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.636230918692654e-06,
+      "loss": 0.8823,
+      "step": 771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.635061584713949e-06,
+      "loss": 1.0126,
+      "step": 772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.633890445520726e-06,
+      "loss": 0.9885,
+      "step": 773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.632717501569106e-06,
+      "loss": 0.978,
+      "step": 774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.631542753315917e-06,
+      "loss": 0.9951,
+      "step": 775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.630366201218686e-06,
+      "loss": 0.9718,
+      "step": 776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.629187845735645e-06,
+      "loss": 0.9921,
+      "step": 777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.628007687325726e-06,
+      "loss": 1.2403,
+      "step": 778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.626825726448565e-06,
+      "loss": 0.997,
+      "step": 779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.6256419635645e-06,
+      "loss": 0.9714,
+      "step": 780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.624456399134571e-06,
+      "loss": 1.0251,
+      "step": 781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.623269033620518e-06,
+      "loss": 0.7807,
+      "step": 782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.622079867484785e-06,
+      "loss": 0.9913,
+      "step": 783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.620888901190513e-06,
+      "loss": 1.079,
+      "step": 784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.619696135201549e-06,
+      "loss": 1.0543,
+      "step": 785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.618501569982437e-06,
+      "loss": 0.8452,
+      "step": 786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.617305205998427e-06,
+      "loss": 0.9145,
+      "step": 787
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.616107043715462e-06,
+      "loss": 1.1313,
+      "step": 788
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.614907083600191e-06,
+      "loss": 0.9158,
+      "step": 789
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.613705326119964e-06,
+      "loss": 0.9069,
+      "step": 790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.612501771742825e-06,
+      "loss": 1.16,
+      "step": 791
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.611296420937526e-06,
+      "loss": 0.8531,
+      "step": 792
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.61008927417351e-06,
+      "loss": 1.0885,
+      "step": 793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.608880331920927e-06,
+      "loss": 1.2345,
+      "step": 794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.60766959465062e-06,
+      "loss": 1.0343,
+      "step": 795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.606457062834137e-06,
+      "loss": 0.8784,
+      "step": 796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.605242736943718e-06,
+      "loss": 0.9922,
+      "step": 797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.60402661745231e-06,
+      "loss": 1.0144,
+      "step": 798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.602808704833553e-06,
+      "loss": 0.9958,
+      "step": 799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.601588999561784e-06,
+      "loss": 0.8585,
+      "step": 800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.600367502112044e-06,
+      "loss": 1.0404,
+      "step": 801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.59914421296007e-06,
+      "loss": 0.9155,
+      "step": 802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.59791913258229e-06,
+      "loss": 0.9832,
+      "step": 803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.59669226145584e-06,
+      "loss": 1.0073,
+      "step": 804
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.595463600058547e-06,
+      "loss": 0.9705,
+      "step": 805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.59423314886894e-06,
+      "loss": 0.8102,
+      "step": 806
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.593000908366238e-06,
+      "loss": 0.9577,
+      "step": 807
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.591766879030363e-06,
+      "loss": 0.9902,
+      "step": 808
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.59053106134193e-06,
+      "loss": 1.0644,
+      "step": 809
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.589293455782253e-06,
+      "loss": 0.9131,
+      "step": 810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.588054062833343e-06,
+      "loss": 0.9156,
+      "step": 811
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.586812882977902e-06,
+      "loss": 0.9358,
+      "step": 812
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.585569916699337e-06,
+      "loss": 1.0515,
+      "step": 813
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.58432516448174e-06,
+      "loss": 0.7793,
+      "step": 814
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.583078626809907e-06,
+      "loss": 1.1331,
+      "step": 815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.581830304169325e-06,
+      "loss": 1.082,
+      "step": 816
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.580580197046177e-06,
+      "loss": 1.0717,
+      "step": 817
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.579328305927343e-06,
+      "loss": 0.9437,
+      "step": 818
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.578074631300396e-06,
+      "loss": 1.0494,
+      "step": 819
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.576819173653602e-06,
+      "loss": 0.9864,
+      "step": 820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.575561933475926e-06,
+      "loss": 1.0391,
+      "step": 821
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.574302911257021e-06,
+      "loss": 0.8389,
+      "step": 822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.57304210748724e-06,
+      "loss": 0.8811,
+      "step": 823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.571779522657628e-06,
+      "loss": 0.935,
+      "step": 824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.57051515725992e-06,
+      "loss": 0.9695,
+      "step": 825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.569249011786552e-06,
+      "loss": 0.963,
+      "step": 826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.567981086730646e-06,
+      "loss": 1.0503,
+      "step": 827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.566711382586022e-06,
+      "loss": 0.8755,
+      "step": 828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.565439899847188e-06,
+      "loss": 1.0427,
+      "step": 829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.564166639009351e-06,
+      "loss": 0.962,
+      "step": 830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.562891600568403e-06,
+      "loss": 0.783,
+      "step": 831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.561614785020937e-06,
+      "loss": 1.0365,
+      "step": 832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.560336192864227e-06,
+      "loss": 0.9486,
+      "step": 833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.559055824596252e-06,
+      "loss": 1.1212,
+      "step": 834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.557773680715676e-06,
+      "loss": 0.8548,
+      "step": 835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.55648976172185e-06,
+      "loss": 1.0733,
+      "step": 836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.555204068114826e-06,
+      "loss": 0.8152,
+      "step": 837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.55391660039534e-06,
+      "loss": 0.9988,
+      "step": 838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.55262735906482e-06,
+      "loss": 1.0423,
+      "step": 839
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.551336344625387e-06,
+      "loss": 0.8871,
+      "step": 840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.55004355757985e-06,
+      "loss": 0.7791,
+      "step": 841
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.548748998431715e-06,
+      "loss": 0.7771,
+      "step": 842
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.54745266768517e-06,
+      "loss": 0.8488,
+      "step": 843
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.546154565845095e-06,
+      "loss": 0.8832,
+      "step": 844
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.544854693417061e-06,
+      "loss": 1.1125,
+      "step": 845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.543553050907332e-06,
+      "loss": 1.114,
+      "step": 846
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.542249638822854e-06,
+      "loss": 1.1414,
+      "step": 847
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.540944457671267e-06,
+      "loss": 0.8935,
+      "step": 848
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.5396375079609e-06,
+      "loss": 0.9468,
+      "step": 849
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.53832879020077e-06,
+      "loss": 0.8347,
+      "step": 850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.537018304900583e-06,
+      "loss": 1.0666,
+      "step": 851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.53570605257073e-06,
+      "loss": 0.9733,
+      "step": 852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.534392033722295e-06,
+      "loss": 1.0,
+      "step": 853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.533076248867048e-06,
+      "loss": 1.1306,
+      "step": 854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.531758698517449e-06,
+      "loss": 0.9219,
+      "step": 855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.530439383186637e-06,
+      "loss": 1.0814,
+      "step": 856
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.529118303388454e-06,
+      "loss": 1.184,
+      "step": 857
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.527795459637413e-06,
+      "loss": 0.9596,
+      "step": 858
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.526470852448724e-06,
+      "loss": 0.9735,
+      "step": 859
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.52514448233828e-06,
+      "loss": 0.9911,
+      "step": 860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.523816349822663e-06,
+      "loss": 0.8469,
+      "step": 861
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.522486455419137e-06,
+      "loss": 0.9735,
+      "step": 862
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.52115479964566e-06,
+      "loss": 1.026,
+      "step": 863
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.519821383020866e-06,
+      "loss": 1.0261,
+      "step": 864
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.518486206064084e-06,
+      "loss": 1.1446,
+      "step": 865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.517149269295321e-06,
+      "loss": 0.9989,
+      "step": 866
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.515810573235275e-06,
+      "loss": 1.113,
+      "step": 867
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.514470118405328e-06,
+      "loss": 0.8882,
+      "step": 868
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.513127905327545e-06,
+      "loss": 1.076,
+      "step": 869
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.511783934524674e-06,
+      "loss": 0.9609,
+      "step": 870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.510438206520155e-06,
+      "loss": 0.9213,
+      "step": 871
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.509090721838106e-06,
+      "loss": 0.9279,
+      "step": 872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.50774148100333e-06,
+      "loss": 1.0727,
+      "step": 873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.506390484541317e-06,
+      "loss": 0.9794,
+      "step": 874
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.505037732978237e-06,
+      "loss": 0.9279,
+      "step": 875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.503683226840948e-06,
+      "loss": 1.064,
+      "step": 876
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.502326966656984e-06,
+      "loss": 1.0275,
+      "step": 877
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.500968952954572e-06,
+      "loss": 1.0561,
+      "step": 878
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.499609186262612e-06,
+      "loss": 1.0854,
+      "step": 879
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.498247667110696e-06,
+      "loss": 1.1747,
+      "step": 880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.49688439602909e-06,
+      "loss": 1.029,
+      "step": 881
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.495519373548748e-06,
+      "loss": 0.8993,
+      "step": 882
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.494152600201306e-06,
+      "loss": 0.8862,
+      "step": 883
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.492784076519077e-06,
+      "loss": 0.9915,
+      "step": 884
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.491413803035062e-06,
+      "loss": 1.1195,
+      "step": 885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.490041780282939e-06,
+      "loss": 0.987,
+      "step": 886
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.488668008797071e-06,
+      "loss": 1.1364,
+      "step": 887
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.487292489112497e-06,
+      "loss": 1.0206,
+      "step": 888
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.48591522176494e-06,
+      "loss": 1.0684,
+      "step": 889
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.484536207290804e-06,
+      "loss": 0.7775,
+      "step": 890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.483155446227176e-06,
+      "loss": 1.0717,
+      "step": 891
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.481772939111815e-06,
+      "loss": 1.0833,
+      "step": 892
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.48038868648317e-06,
+      "loss": 1.0001,
+      "step": 893
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.479002688880362e-06,
+      "loss": 0.8678,
+      "step": 894
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.477614946843194e-06,
+      "loss": 0.9319,
+      "step": 895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.47622546091215e-06,
+      "loss": 1.0058,
+      "step": 896
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.474834231628394e-06,
+      "loss": 1.0518,
+      "step": 897
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.473441259533765e-06,
+      "loss": 0.9858,
+      "step": 898
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.472046545170784e-06,
+      "loss": 0.9304,
+      "step": 899
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.470650089082649e-06,
+      "loss": 0.9344,
+      "step": 900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.469251891813238e-06,
+      "loss": 1.0551,
+      "step": 901
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.467851953907103e-06,
+      "loss": 0.8394,
+      "step": 902
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.466450275909481e-06,
+      "loss": 0.9246,
+      "step": 903
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.46504685836628e-06,
+      "loss": 1.1245,
+      "step": 904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.46364170182409e-06,
+      "loss": 1.042,
+      "step": 905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.462234806830172e-06,
+      "loss": 0.9936,
+      "step": 906
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.460826173932475e-06,
+      "loss": 1.045,
+      "step": 907
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.459415803679613e-06,
+      "loss": 0.9055,
+      "step": 908
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458003696620885e-06,
+      "loss": 0.8422,
+      "step": 909
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.456589853306263e-06,
+      "loss": 0.9626,
+      "step": 910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.455174274286396e-06,
+      "loss": 0.9061,
+      "step": 911
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.453756960112605e-06,
+      "loss": 1.0182,
+      "step": 912
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.452337911336896e-06,
+      "loss": 0.9314,
+      "step": 913
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.450917128511942e-06,
+      "loss": 0.9824,
+      "step": 914
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.449494612191094e-06,
+      "loss": 1.0699,
+      "step": 915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44807036292838e-06,
+      "loss": 0.9369,
+      "step": 916
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4466443812785e-06,
+      "loss": 1.0849,
+      "step": 917
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.445216667796833e-06,
+      "loss": 1.0361,
+      "step": 918
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.443787223039425e-06,
+      "loss": 1.0555,
+      "step": 919
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442356047563007e-06,
+      "loss": 1.0686,
+      "step": 920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44092314192497e-06,
+      "loss": 1.0136,
+      "step": 921
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.439488506683393e-06,
+      "loss": 1.0277,
+      "step": 922
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.438052142397024e-06,
+      "loss": 1.0419,
+      "step": 923
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.436614049625277e-06,
+      "loss": 0.9637,
+      "step": 924
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.435174228928248e-06,
+      "loss": 0.9622,
+      "step": 925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.433732680866704e-06,
+      "loss": 0.8612,
+      "step": 926
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.432289406002085e-06,
+      "loss": 1.1018,
+      "step": 927
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.430844404896496e-06,
+      "loss": 0.9159,
+      "step": 928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.429397678112726e-06,
+      "loss": 0.992,
+      "step": 929
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.42794922621423e-06,
+      "loss": 1.0421,
+      "step": 930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.426499049765136e-06,
+      "loss": 0.8513,
+      "step": 931
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.425047149330242e-06,
+      "loss": 0.9309,
+      "step": 932
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.423593525475019e-06,
+      "loss": 0.9437,
+      "step": 933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.42213817876561e-06,
+      "loss": 1.0481,
+      "step": 934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.420681109768827e-06,
+      "loss": 1.0543,
+      "step": 935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.419222319052154e-06,
+      "loss": 0.7497,
+      "step": 936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.417761807183745e-06,
+      "loss": 0.9544,
+      "step": 937
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.416299574732425e-06,
+      "loss": 1.067,
+      "step": 938
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.414835622267689e-06,
+      "loss": 1.0091,
+      "step": 939
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.4133699503597e-06,
+      "loss": 1.1458,
+      "step": 940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.411902559579295e-06,
+      "loss": 1.0347,
+      "step": 941
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.410433450497977e-06,
+      "loss": 1.003,
+      "step": 942
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.408962623687918e-06,
+      "loss": 0.7984,
+      "step": 943
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.407490079721959e-06,
+      "loss": 1.0298,
+      "step": 944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.406015819173614e-06,
+      "loss": 0.9763,
+      "step": 945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.404539842617059e-06,
+      "loss": 0.9516,
+      "step": 946
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.403062150627145e-06,
+      "loss": 1.0415,
+      "step": 947
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.401582743779384e-06,
+      "loss": 0.8496,
+      "step": 948
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.400101622649962e-06,
+      "loss": 0.8754,
+      "step": 949
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.398618787815731e-06,
+      "loss": 0.9275,
+      "step": 950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.39713423985421e-06,
+      "loss": 1.0016,
+      "step": 951
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.39564797934358e-06,
+      "loss": 0.9654,
+      "step": 952
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.394160006862698e-06,
+      "loss": 0.9456,
+      "step": 953
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.392670322991085e-06,
+      "loss": 0.971,
+      "step": 954
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.391178928308925e-06,
+      "loss": 1.1241,
+      "step": 955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.38968582339707e-06,
+      "loss": 1.0109,
+      "step": 956
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.388191008837042e-06,
+      "loss": 1.0672,
+      "step": 957
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.386694485211021e-06,
+      "loss": 1.0224,
+      "step": 958
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.385196253101861e-06,
+      "loss": 0.9187,
+      "step": 959
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.383696313093073e-06,
+      "loss": 1.1279,
+      "step": 960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.382194665768844e-06,
+      "loss": 0.8455,
+      "step": 961
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.380691311714013e-06,
+      "loss": 0.9372,
+      "step": 962
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.379186251514096e-06,
+      "loss": 0.9143,
+      "step": 963
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.377679485755265e-06,
+      "loss": 0.9776,
+      "step": 964
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.376171015024358e-06,
+      "loss": 0.9982,
+      "step": 965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.374660839908881e-06,
+      "loss": 0.8045,
+      "step": 966
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.373148960996998e-06,
+      "loss": 0.9794,
+      "step": 967
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.371635378877542e-06,
+      "loss": 1.0392,
+      "step": 968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.370120094140006e-06,
+      "loss": 0.9256,
+      "step": 969
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.368603107374547e-06,
+      "loss": 0.9179,
+      "step": 970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.367084419171984e-06,
+      "loss": 0.9725,
+      "step": 971
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.365564030123802e-06,
+      "loss": 0.9556,
+      "step": 972
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.364041940822143e-06,
+      "loss": 0.9437,
+      "step": 973
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.362518151859816e-06,
+      "loss": 1.047,
+      "step": 974
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.360992663830289e-06,
+      "loss": 0.8546,
+      "step": 975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.359465477327694e-06,
+      "loss": 1.1282,
+      "step": 976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.357936592946823e-06,
+      "loss": 1.0022,
+      "step": 977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.356406011283128e-06,
+      "loss": 0.9689,
+      "step": 978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.354873732932727e-06,
+      "loss": 0.8217,
+      "step": 979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.353339758492394e-06,
+      "loss": 1.0406,
+      "step": 980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.351804088559565e-06,
+      "loss": 1.1586,
+      "step": 981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.350266723732333e-06,
+      "loss": 1.0181,
+      "step": 982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.34872766460946e-06,
+      "loss": 0.962,
+      "step": 983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.34718691179036e-06,
+      "loss": 0.9141,
+      "step": 984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.345644465875108e-06,
+      "loss": 0.8637,
+      "step": 985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.344100327464441e-06,
+      "loss": 0.9987,
+      "step": 986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.342554497159753e-06,
+      "loss": 1.1295,
+      "step": 987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.341006975563098e-06,
+      "loss": 0.8159,
+      "step": 988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.339457763277188e-06,
+      "loss": 1.0408,
+      "step": 989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.337906860905394e-06,
+      "loss": 1.1985,
+      "step": 990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.336354269051744e-06,
+      "loss": 0.9744,
+      "step": 991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.334799988320926e-06,
+      "loss": 1.0857,
+      "step": 992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.333244019318283e-06,
+      "loss": 1.0269,
+      "step": 993
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.331686362649819e-06,
+      "loss": 0.9672,
+      "step": 994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.330127018922195e-06,
+      "loss": 0.9154,
+      "step": 995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.328565988742723e-06,
+      "loss": 1.0965,
+      "step": 996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.327003272719382e-06,
+      "loss": 0.9059,
+      "step": 997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.325438871460798e-06,
+      "loss": 0.9214,
+      "step": 998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.323872785576257e-06,
+      "loss": 1.0141,
+      "step": 999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.322305015675706e-06,
+      "loss": 0.9246,
+      "step": 1000
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 1.1130287647247314,
+      "eval_runtime": 89.4192,
+      "eval_samples_per_second": 17.245,
+      "eval_steps_per_second": 2.158,
+      "step": 1000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.320735562369739e-06,
+      "loss": 1.0262,
+      "step": 1001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.31916442626961e-06,
+      "loss": 0.9005,
+      "step": 1002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.317591607987233e-06,
+      "loss": 0.9305,
+      "step": 1003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.316017108135167e-06,
+      "loss": 1.0341,
+      "step": 1004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.314440927326635e-06,
+      "loss": 1.0179,
+      "step": 1005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.31286306617551e-06,
+      "loss": 0.8485,
+      "step": 1006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.31128352529632e-06,
+      "loss": 0.9813,
+      "step": 1007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.30970230530425e-06,
+      "loss": 1.0792,
+      "step": 1008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.308119406815132e-06,
+      "loss": 1.024,
+      "step": 1009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.306534830445462e-06,
+      "loss": 0.9247,
+      "step": 1010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.30494857681238e-06,
+      "loss": 1.0072,
+      "step": 1011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.303360646533686e-06,
+      "loss": 0.9068,
+      "step": 1012
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.30177104022783e-06,
+      "loss": 0.9445,
+      "step": 1013
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.300179758513912e-06,
+      "loss": 0.8823,
+      "step": 1014
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.29858680201169e-06,
+      "loss": 0.927,
+      "step": 1015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.296992171341573e-06,
+      "loss": 0.8278,
+      "step": 1016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.295395867124617e-06,
+      "loss": 1.1282,
+      "step": 1017
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.293797889982535e-06,
+      "loss": 0.8373,
+      "step": 1018
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.292198240537692e-06,
+      "loss": 1.0767,
+      "step": 1019
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.290596919413101e-06,
+      "loss": 0.9809,
+      "step": 1020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.288993927232428e-06,
+      "loss": 0.9832,
+      "step": 1021
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.287389264619988e-06,
+      "loss": 0.9975,
+      "step": 1022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.28578293220075e-06,
+      "loss": 0.989,
+      "step": 1023
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.28417493060033e-06,
+      "loss": 1.0948,
+      "step": 1024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.282565260444997e-06,
+      "loss": 0.9299,
+      "step": 1025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.280953922361667e-06,
+      "loss": 1.0251,
+      "step": 1026
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.279340916977905e-06,
+      "loss": 0.925,
+      "step": 1027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.27772624492193e-06,
+      "loss": 1.0595,
+      "step": 1028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.276109906822606e-06,
+      "loss": 0.9993,
+      "step": 1029
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.274491903309446e-06,
+      "loss": 1.1356,
+      "step": 1030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.272872235012616e-06,
+      "loss": 0.9312,
+      "step": 1031
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.271250902562925e-06,
+      "loss": 1.0092,
+      "step": 1032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.269627906591832e-06,
+      "loss": 0.9611,
+      "step": 1033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.268003247731446e-06,
+      "loss": 1.1149,
+      "step": 1034
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.266376926614521e-06,
+      "loss": 0.8319,
+      "step": 1035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.264748943874459e-06,
+      "loss": 0.9669,
+      "step": 1036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.26311930014531e-06,
+      "loss": 0.718,
+      "step": 1037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.26148799606177e-06,
+      "loss": 1.0643,
+      "step": 1038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.25985503225918e-06,
+      "loss": 1.0405,
+      "step": 1039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.258220409373531e-06,
+      "loss": 0.9853,
+      "step": 1040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.256584128041461e-06,
+      "loss": 1.0318,
+      "step": 1041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.254946188900247e-06,
+      "loss": 1.0672,
+      "step": 1042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.253306592587819e-06,
+      "loss": 0.9124,
+      "step": 1043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.251665339742751e-06,
+      "loss": 1.0134,
+      "step": 1044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.250022431004256e-06,
+      "loss": 0.9459,
+      "step": 1045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.248377867012201e-06,
+      "loss": 1.0033,
+      "step": 1046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.24673164840709e-06,
+      "loss": 0.9705,
+      "step": 1047
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.245083775830078e-06,
+      "loss": 0.9757,
+      "step": 1048
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.243434249922956e-06,
+      "loss": 0.9737,
+      "step": 1049
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.24178307132817e-06,
+      "loss": 1.0314,
+      "step": 1050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.2401302406888e-06,
+      "loss": 1.0993,
+      "step": 1051
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.238475758648572e-06,
+      "loss": 0.9867,
+      "step": 1052
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.236819625851858e-06,
+      "loss": 0.9707,
+      "step": 1053
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.235161842943671e-06,
+      "loss": 0.9254,
+      "step": 1054
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.233502410569666e-06,
+      "loss": 1.0468,
+      "step": 1055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.231841329376142e-06,
+      "loss": 0.8706,
+      "step": 1056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.230178600010039e-06,
+      "loss": 0.9673,
+      "step": 1057
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.228514223118936e-06,
+      "loss": 0.976,
+      "step": 1058
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.22684819935106e-06,
+      "loss": 0.8862,
+      "step": 1059
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.225180529355275e-06,
+      "loss": 1.013,
+      "step": 1060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.223511213781091e-06,
+      "loss": 0.9399,
+      "step": 1061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.22184025327865e-06,
+      "loss": 1.0251,
+      "step": 1062
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.220167648498743e-06,
+      "loss": 1.0655,
+      "step": 1063
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.218493400092796e-06,
+      "loss": 1.0431,
+      "step": 1064
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.216817508712882e-06,
+      "loss": 1.0286,
+      "step": 1065
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.215139975011705e-06,
+      "loss": 1.0957,
+      "step": 1066
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.213460799642612e-06,
+      "loss": 1.0327,
+      "step": 1067
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.211779983259597e-06,
+      "loss": 0.9459,
+      "step": 1068
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.210097526517281e-06,
+      "loss": 0.9187,
+      "step": 1069
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.20841343007093e-06,
+      "loss": 0.8113,
+      "step": 1070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.206727694576448e-06,
+      "loss": 1.1715,
+      "step": 1071
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.205040320690379e-06,
+      "loss": 1.156,
+      "step": 1072
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.203351309069902e-06,
+      "loss": 0.9687,
+      "step": 1073
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.201660660372835e-06,
+      "loss": 1.1257,
+      "step": 1074
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.199968375257635e-06,
+      "loss": 0.9322,
+      "step": 1075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.198274454383396e-06,
+      "loss": 0.9005,
+      "step": 1076
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.196578898409847e-06,
+      "loss": 0.9231,
+      "step": 1077
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.194881707997355e-06,
+      "loss": 0.9364,
+      "step": 1078
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.193182883806921e-06,
+      "loss": 1.0337,
+      "step": 1079
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.191482426500192e-06,
+      "loss": 0.9184,
+      "step": 1080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.189780336739437e-06,
+      "loss": 1.0979,
+      "step": 1081
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.188076615187573e-06,
+      "loss": 0.8412,
+      "step": 1082
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.186371262508144e-06,
+      "loss": 1.0163,
+      "step": 1083
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.184664279365334e-06,
+      "loss": 0.9226,
+      "step": 1084
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.182955666423961e-06,
+      "loss": 1.0751,
+      "step": 1085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.181245424349477e-06,
+      "loss": 0.9414,
+      "step": 1086
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.179533553807967e-06,
+      "loss": 1.0504,
+      "step": 1087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.177820055466155e-06,
+      "loss": 1.0274,
+      "step": 1088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.176104929991396e-06,
+      "loss": 0.8466,
+      "step": 1089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.174388178051676e-06,
+      "loss": 1.0,
+      "step": 1090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.172669800315619e-06,
+      "loss": 0.9134,
+      "step": 1091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.170949797452481e-06,
+      "loss": 0.9939,
+      "step": 1092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.169228170132151e-06,
+      "loss": 0.9452,
+      "step": 1093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.16750491902515e-06,
+      "loss": 0.9147,
+      "step": 1094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.165780044802629e-06,
+      "loss": 0.9883,
+      "step": 1095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.164053548136376e-06,
+      "loss": 0.8092,
+      "step": 1096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.162325429698807e-06,
+      "loss": 0.9919,
+      "step": 1097
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.160595690162974e-06,
+      "loss": 1.1023,
+      "step": 1098
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.158864330202553e-06,
+      "loss": 1.1119,
+      "step": 1099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.15713135049186e-06,
+      "loss": 1.0518,
+      "step": 1100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.155396751705836e-06,
+      "loss": 1.0735,
+      "step": 1101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.153660534520054e-06,
+      "loss": 0.9277,
+      "step": 1102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.151922699610718e-06,
+      "loss": 0.957,
+      "step": 1103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.15018324765466e-06,
+      "loss": 1.0271,
+      "step": 1104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.148442179329344e-06,
+      "loss": 1.2147,
+      "step": 1105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.146699495312862e-06,
+      "loss": 0.8248,
+      "step": 1106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.14495519628394e-06,
+      "loss": 0.8633,
+      "step": 1107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.143209282921924e-06,
+      "loss": 0.9277,
+      "step": 1108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.141461755906795e-06,
+      "loss": 0.9786,
+      "step": 1109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.139712615919163e-06,
+      "loss": 0.8974,
+      "step": 1110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.137961863640263e-06,
+      "loss": 0.9786,
+      "step": 1111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.136209499751962e-06,
+      "loss": 1.0536,
+      "step": 1112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.134455524936749e-06,
+      "loss": 0.9478,
+      "step": 1113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.132699939877744e-06,
+      "loss": 1.0421,
+      "step": 1114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.130942745258694e-06,
+      "loss": 1.0761,
+      "step": 1115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.129183941763971e-06,
+      "loss": 0.8273,
+      "step": 1116
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.127423530078578e-06,
+      "loss": 0.8243,
+      "step": 1117
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.12566151088814e-06,
+      "loss": 0.9616,
+      "step": 1118
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.123897884878909e-06,
+      "loss": 0.9063,
+      "step": 1119
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.122132652737765e-06,
+      "loss": 0.8652,
+      "step": 1120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.12036581515221e-06,
+      "loss": 0.9421,
+      "step": 1121
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.118597372810374e-06,
+      "loss": 0.8445,
+      "step": 1122
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.11682732640101e-06,
+      "loss": 0.9457,
+      "step": 1123
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.115055676613498e-06,
+      "loss": 0.9335,
+      "step": 1124
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.113282424137842e-06,
+      "loss": 0.9681,
+      "step": 1125
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.111507569664668e-06,
+      "loss": 0.9651,
+      "step": 1126
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.10973111388523e-06,
+      "loss": 1.034,
+      "step": 1127
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.107953057491399e-06,
+      "loss": 0.8958,
+      "step": 1128
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.106173401175677e-06,
+      "loss": 0.9818,
+      "step": 1129
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.104392145631183e-06,
+      "loss": 0.9704,
+      "step": 1130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.102609291551664e-06,
+      "loss": 1.0282,
+      "step": 1131
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.100824839631485e-06,
+      "loss": 0.8351,
+      "step": 1132
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.099038790565638e-06,
+      "loss": 1.0102,
+      "step": 1133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.09725114504973e-06,
+      "loss": 0.9433,
+      "step": 1134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.095461903779999e-06,
+      "loss": 0.8263,
+      "step": 1135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.093671067453296e-06,
+      "loss": 1.0467,
+      "step": 1136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.0918786367671e-06,
+      "loss": 1.0342,
+      "step": 1137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.090084612419506e-06,
+      "loss": 1.0448,
+      "step": 1138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.08828899510923e-06,
+      "loss": 1.0285,
+      "step": 1139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.086491785535613e-06,
+      "loss": 1.0512,
+      "step": 1140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.084692984398612e-06,
+      "loss": 0.964,
+      "step": 1141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.082892592398804e-06,
+      "loss": 1.1244,
+      "step": 1142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.081090610237388e-06,
+      "loss": 0.9588,
+      "step": 1143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.07928703861618e-06,
+      "loss": 0.7332,
+      "step": 1144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.077481878237618e-06,
+      "loss": 0.9741,
+      "step": 1145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.07567512980475e-06,
+      "loss": 0.9613,
+      "step": 1146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.073866794021258e-06,
+      "loss": 0.9809,
+      "step": 1147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.072056871591428e-06,
+      "loss": 0.9366,
+      "step": 1148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.070245363220172e-06,
+      "loss": 0.9632,
+      "step": 1149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.068432269613012e-06,
+      "loss": 0.9609,
+      "step": 1150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.066617591476099e-06,
+      "loss": 0.8978,
+      "step": 1151
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.064801329516192e-06,
+      "loss": 1.1538,
+      "step": 1152
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.062983484440668e-06,
+      "loss": 0.9779,
+      "step": 1153
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.061164056957523e-06,
+      "loss": 0.8917,
+      "step": 1154
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.059343047775368e-06,
+      "loss": 0.9073,
+      "step": 1155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.057520457603433e-06,
+      "loss": 1.0209,
+      "step": 1156
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.055696287151556e-06,
+      "loss": 0.7582,
+      "step": 1157
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.053870537130198e-06,
+      "loss": 1.1078,
+      "step": 1158
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.052043208250435e-06,
+      "loss": 1.1298,
+      "step": 1159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.050214301223952e-06,
+      "loss": 0.9329,
+      "step": 1160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.048383816763056e-06,
+      "loss": 0.9385,
+      "step": 1161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.04655175558066e-06,
+      "loss": 0.9825,
+      "step": 1162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.044718118390299e-06,
+      "loss": 1.0473,
+      "step": 1163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.042882905906118e-06,
+      "loss": 1.006,
+      "step": 1164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.041046118842874e-06,
+      "loss": 0.8925,
+      "step": 1165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.039207757915942e-06,
+      "loss": 0.9181,
+      "step": 1166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.037367823841308e-06,
+      "loss": 0.97,
+      "step": 1167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.035526317335567e-06,
+      "loss": 0.9156,
+      "step": 1168
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.03368323911593e-06,
+      "loss": 0.6677,
+      "step": 1169
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.03183858990022e-06,
+      "loss": 1.0573,
+      "step": 1170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.029992370406872e-06,
+      "loss": 0.8438,
+      "step": 1171
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.028144581354932e-06,
+      "loss": 0.7961,
+      "step": 1172
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.026295223464056e-06,
+      "loss": 0.8662,
+      "step": 1173
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.024444297454515e-06,
+      "loss": 0.8332,
+      "step": 1174
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.022591804047183e-06,
+      "loss": 1.0643,
+      "step": 1175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.020737743963555e-06,
+      "loss": 0.9569,
+      "step": 1176
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.018882117925726e-06,
+      "loss": 0.8465,
+      "step": 1177
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.017024926656408e-06,
+      "loss": 0.836,
+      "step": 1178
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.015166170878919e-06,
+      "loss": 0.9639,
+      "step": 1179
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.01330585131719e-06,
+      "loss": 1.0215,
+      "step": 1180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.011443968695756e-06,
+      "loss": 1.0461,
+      "step": 1181
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.009580523739763e-06,
+      "loss": 0.8797,
+      "step": 1182
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.007715517174966e-06,
+      "loss": 0.9169,
+      "step": 1183
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.005848949727729e-06,
+      "loss": 0.8447,
+      "step": 1184
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.003980822125024e-06,
+      "loss": 1.0077,
+      "step": 1185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.002111135094426e-06,
+      "loss": 0.8241,
+      "step": 1186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.000239889364123e-06,
+      "loss": 1.1505,
+      "step": 1187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.998367085662908e-06,
+      "loss": 0.8957,
+      "step": 1188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.996492724720181e-06,
+      "loss": 1.0677,
+      "step": 1189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.994616807265945e-06,
+      "loss": 0.8511,
+      "step": 1190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.992739334030815e-06,
+      "loss": 0.927,
+      "step": 1191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.990860305746012e-06,
+      "loss": 1.1384,
+      "step": 1192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.988979723143354e-06,
+      "loss": 0.748,
+      "step": 1193
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.987097586955276e-06,
+      "loss": 1.0348,
+      "step": 1194
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.985213897914808e-06,
+      "loss": 0.9461,
+      "step": 1195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.98332865675559e-06,
+      "loss": 0.9058,
+      "step": 1196
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.981441864211869e-06,
+      "loss": 1.0126,
+      "step": 1197
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.97955352101849e-06,
+      "loss": 0.8629,
+      "step": 1198
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.977663627910903e-06,
+      "loss": 0.9692,
+      "step": 1199
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.97577218562517e-06,
+      "loss": 0.8966,
+      "step": 1200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.973879194897942e-06,
+      "loss": 1.0587,
+      "step": 1201
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.971984656466488e-06,
+      "loss": 0.9766,
+      "step": 1202
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.970088571068667e-06,
+      "loss": 0.9177,
+      "step": 1203
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.968190939442948e-06,
+      "loss": 1.0443,
+      "step": 1204
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.966291762328404e-06,
+      "loss": 1.0181,
+      "step": 1205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.964391040464699e-06,
+      "loss": 0.9556,
+      "step": 1206
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.962488774592113e-06,
+      "loss": 0.9506,
+      "step": 1207
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.960584965451517e-06,
+      "loss": 1.0374,
+      "step": 1208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.958679613784387e-06,
+      "loss": 0.8593,
+      "step": 1209
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.956772720332799e-06,
+      "loss": 0.9204,
+      "step": 1210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.954864285839428e-06,
+      "loss": 0.8278,
+      "step": 1211
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.952954311047554e-06,
+      "loss": 0.9701,
+      "step": 1212
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.951042796701051e-06,
+      "loss": 0.9464,
+      "step": 1213
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.949129743544396e-06,
+      "loss": 0.758,
+      "step": 1214
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.947215152322666e-06,
+      "loss": 0.9965,
+      "step": 1215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.945299023781534e-06,
+      "loss": 1.0842,
+      "step": 1216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.943381358667273e-06,
+      "loss": 0.8665,
+      "step": 1217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.941462157726757e-06,
+      "loss": 0.9714,
+      "step": 1218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.939541421707454e-06,
+      "loss": 0.9738,
+      "step": 1219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.937619151357433e-06,
+      "loss": 1.0803,
+      "step": 1220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.935695347425359e-06,
+      "loss": 0.7428,
+      "step": 1221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.933770010660494e-06,
+      "loss": 0.8451,
+      "step": 1222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.9318431418127e-06,
+      "loss": 0.9917,
+      "step": 1223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.92991474163243e-06,
+      "loss": 0.9201,
+      "step": 1224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.927984810870739e-06,
+      "loss": 1.1316,
+      "step": 1225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.926053350279276e-06,
+      "loss": 0.9683,
+      "step": 1226
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.924120360610284e-06,
+      "loss": 0.8985,
+      "step": 1227
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.922185842616607e-06,
+      "loss": 0.9449,
+      "step": 1228
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.920249797051675e-06,
+      "loss": 1.1098,
+      "step": 1229
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.918312224669523e-06,
+      "loss": 1.0074,
+      "step": 1230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.916373126224775e-06,
+      "loss": 1.127,
+      "step": 1231
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.914432502472648e-06,
+      "loss": 0.9567,
+      "step": 1232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.912490354168959e-06,
+      "loss": 0.9672,
+      "step": 1233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.910546682070114e-06,
+      "loss": 1.0179,
+      "step": 1234
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.908601486933113e-06,
+      "loss": 0.8757,
+      "step": 1235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.906654769515551e-06,
+      "loss": 0.9275,
+      "step": 1236
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.904706530575613e-06,
+      "loss": 1.0786,
+      "step": 1237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.902756770872082e-06,
+      "loss": 0.9729,
+      "step": 1238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.900805491164326e-06,
+      "loss": 1.0915,
+      "step": 1239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.89885269221231e-06,
+      "loss": 0.9051,
+      "step": 1240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.89689837477659e-06,
+      "loss": 1.0547,
+      "step": 1241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.89494253961831e-06,
+      "loss": 1.0363,
+      "step": 1242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.892985187499212e-06,
+      "loss": 1.0334,
+      "step": 1243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.891026319181625e-06,
+      "loss": 0.9801,
+      "step": 1244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.889065935428462e-06,
+      "loss": 1.0754,
+      "step": 1245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.887104037003238e-06,
+      "loss": 1.1574,
+      "step": 1246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.88514062467005e-06,
+      "loss": 0.9881,
+      "step": 1247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.883175699193589e-06,
+      "loss": 0.9397,
+      "step": 1248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.881209261339133e-06,
+      "loss": 0.8106,
+      "step": 1249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.879241311872546e-06,
+      "loss": 0.9982,
+      "step": 1250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.877271851560287e-06,
+      "loss": 1.0277,
+      "step": 1251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.8753008811694e-06,
+      "loss": 0.8966,
+      "step": 1252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.873328401467517e-06,
+      "loss": 1.0737,
+      "step": 1253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.871354413222859e-06,
+      "loss": 0.8758,
+      "step": 1254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.869378917204235e-06,
+      "loss": 1.0245,
+      "step": 1255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.867401914181037e-06,
+      "loss": 1.0042,
+      "step": 1256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.86542340492325e-06,
+      "loss": 0.93,
+      "step": 1257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.863443390201443e-06,
+      "loss": 0.8435,
+      "step": 1258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.86146187078677e-06,
+      "loss": 1.042,
+      "step": 1259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.85947884745097e-06,
+      "loss": 1.0231,
+      "step": 1260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.857494320966374e-06,
+      "loss": 0.8047,
+      "step": 1261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.85550829210589e-06,
+      "loss": 1.0889,
+      "step": 1262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.85352076164302e-06,
+      "loss": 0.8959,
+      "step": 1263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.851531730351843e-06,
+      "loss": 0.9234,
+      "step": 1264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.849541199007028e-06,
+      "loss": 1.0993,
+      "step": 1265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.847549168383823e-06,
+      "loss": 0.9614,
+      "step": 1266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.845555639258065e-06,
+      "loss": 0.8619,
+      "step": 1267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.843560612406171e-06,
+      "loss": 0.917,
+      "step": 1268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.841564088605145e-06,
+      "loss": 0.9658,
+      "step": 1269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.83956606863257e-06,
+      "loss": 0.8081,
+      "step": 1270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.837566553266617e-06,
+      "loss": 1.0067,
+      "step": 1271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.835565543286031e-06,
+      "loss": 0.8255,
+      "step": 1272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.833563039470146e-06,
+      "loss": 1.0527,
+      "step": 1273
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.831559042598879e-06,
+      "loss": 0.8809,
+      "step": 1274
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.82955355345272e-06,
+      "loss": 0.8077,
+      "step": 1275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.82754657281275e-06,
+      "loss": 1.0206,
+      "step": 1276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.825538101460625e-06,
+      "loss": 0.9175,
+      "step": 1277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.82352814017858e-06,
+      "loss": 0.9291,
+      "step": 1278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.82151668974944e-06,
+      "loss": 0.8901,
+      "step": 1279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.819503750956598e-06,
+      "loss": 0.9898,
+      "step": 1280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.817489324584035e-06,
+      "loss": 0.8457,
+      "step": 1281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.815473411416305e-06,
+      "loss": 0.9643,
+      "step": 1282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.813456012238548e-06,
+      "loss": 0.7832,
+      "step": 1283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.811437127836477e-06,
+      "loss": 0.7861,
+      "step": 1284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.809416758996386e-06,
+      "loss": 1.09,
+      "step": 1285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.807394906505146e-06,
+      "loss": 0.9607,
+      "step": 1286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.80537157115021e-06,
+      "loss": 0.9299,
+      "step": 1287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.803346753719602e-06,
+      "loss": 1.1431,
+      "step": 1288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.801320455001928e-06,
+      "loss": 0.9957,
+      "step": 1289
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.799292675786365e-06,
+      "loss": 0.897,
+      "step": 1290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.797263416862678e-06,
+      "loss": 0.8123,
+      "step": 1291
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.795232679021197e-06,
+      "loss": 0.9541,
+      "step": 1292
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.793200463052832e-06,
+      "loss": 0.8835,
+      "step": 1293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.791166769749068e-06,
+      "loss": 1.0021,
+      "step": 1294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.78913159990197e-06,
+      "loss": 0.8914,
+      "step": 1295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.787094954304172e-06,
+      "loss": 1.0029,
+      "step": 1296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.785056833748883e-06,
+      "loss": 0.9944,
+      "step": 1297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.783017239029894e-06,
+      "loss": 1.0457,
+      "step": 1298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.780976170941561e-06,
+      "loss": 0.9125,
+      "step": 1299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.778933630278817e-06,
+      "loss": 0.9951,
+      "step": 1300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.77688961783717e-06,
+      "loss": 0.9831,
+      "step": 1301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.7748441344127e-06,
+      "loss": 0.9971,
+      "step": 1302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.772797180802061e-06,
+      "loss": 1.1294,
+      "step": 1303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.770748757802478e-06,
+      "loss": 1.0253,
+      "step": 1304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.76869886621175e-06,
+      "loss": 0.9648,
+      "step": 1305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.766647506828246e-06,
+      "loss": 1.0831,
+      "step": 1306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.764594680450907e-06,
+      "loss": 0.9608,
+      "step": 1307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.762540387879245e-06,
+      "loss": 1.0421,
+      "step": 1308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.760484629913349e-06,
+      "loss": 0.9144,
+      "step": 1309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.758427407353868e-06,
+      "loss": 0.9735,
+      "step": 1310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.756368721002031e-06,
+      "loss": 0.8375,
+      "step": 1311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.754308571659633e-06,
+      "loss": 0.9071,
+      "step": 1312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.752246960129037e-06,
+      "loss": 0.882,
+      "step": 1313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.75018388721318e-06,
+      "loss": 0.8478,
+      "step": 1314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.748119353715562e-06,
+      "loss": 0.9091,
+      "step": 1315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.746053360440258e-06,
+      "loss": 0.9208,
+      "step": 1316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.74398590819191e-06,
+      "loss": 1.0387,
+      "step": 1317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.741916997775727e-06,
+      "loss": 0.9518,
+      "step": 1318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.739846629997483e-06,
+      "loss": 0.8994,
+      "step": 1319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.73777480566353e-06,
+      "loss": 1.0833,
+      "step": 1320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.73570152558077e-06,
+      "loss": 0.9483,
+      "step": 1321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.733626790556692e-06,
+      "loss": 1.049,
+      "step": 1322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.731550601399336e-06,
+      "loss": 0.9837,
+      "step": 1323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.729472958917316e-06,
+      "loss": 0.9904,
+      "step": 1324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.727393863919809e-06,
+      "loss": 1.1107,
+      "step": 1325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.725313317216558e-06,
+      "loss": 1.0005,
+      "step": 1326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.723231319617875e-06,
+      "loss": 0.9344,
+      "step": 1327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.721147871934632e-06,
+      "loss": 1.0561,
+      "step": 1328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.71906297497827e-06,
+      "loss": 0.92,
+      "step": 1329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.716976629560791e-06,
+      "loss": 0.995,
+      "step": 1330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.714888836494762e-06,
+      "loss": 0.881,
+      "step": 1331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.712799596593317e-06,
+      "loss": 1.0066,
+      "step": 1332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.71070891067015e-06,
+      "loss": 0.9815,
+      "step": 1333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.708616779539518e-06,
+      "loss": 0.947,
+      "step": 1334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.706523204016242e-06,
+      "loss": 0.9087,
+      "step": 1335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.704428184915706e-06,
+      "loss": 1.0734,
+      "step": 1336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.702331723053856e-06,
+      "loss": 0.946,
+      "step": 1337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.7002338192472e-06,
+      "loss": 0.9879,
+      "step": 1338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.698134474312808e-06,
+      "loss": 0.9318,
+      "step": 1339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.696033689068309e-06,
+      "loss": 0.8354,
+      "step": 1340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.693931464331893e-06,
+      "loss": 1.0353,
+      "step": 1341
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.691827800922314e-06,
+      "loss": 0.9885,
+      "step": 1342
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.689722699658887e-06,
+      "loss": 0.9467,
+      "step": 1343
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.68761616136148e-06,
+      "loss": 1.0466,
+      "step": 1344
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.685508186850529e-06,
+      "loss": 0.8944,
+      "step": 1345
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.683398776947022e-06,
+      "loss": 0.7941,
+      "step": 1346
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.681287932472512e-06,
+      "loss": 0.8459,
+      "step": 1347
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.679175654249107e-06,
+      "loss": 0.9956,
+      "step": 1348
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.677061943099476e-06,
+      "loss": 1.0268,
+      "step": 1349
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.674946799846844e-06,
+      "loss": 0.986,
+      "step": 1350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.672830225314994e-06,
+      "loss": 0.9477,
+      "step": 1351
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.670712220328268e-06,
+      "loss": 0.8761,
+      "step": 1352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.668592785711562e-06,
+      "loss": 0.8886,
+      "step": 1353
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.666471922290334e-06,
+      "loss": 1.1705,
+      "step": 1354
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.664349630890592e-06,
+      "loss": 1.0659,
+      "step": 1355
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.662225912338906e-06,
+      "loss": 0.933,
+      "step": 1356
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.660100767462398e-06,
+      "loss": 0.7246,
+      "step": 1357
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.657974197088747e-06,
+      "loss": 0.9659,
+      "step": 1358
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.65584620204619e-06,
+      "loss": 1.0416,
+      "step": 1359
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.65371678316351e-06,
+      "loss": 0.8401,
+      "step": 1360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.651585941270056e-06,
+      "loss": 0.9712,
+      "step": 1361
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.64945367719572e-06,
+      "loss": 0.8428,
+      "step": 1362
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.647319991770959e-06,
+      "loss": 0.9584,
+      "step": 1363
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.645184885826776e-06,
+      "loss": 0.798,
+      "step": 1364
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.64304836019473e-06,
+      "loss": 1.0165,
+      "step": 1365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.640910415706931e-06,
+      "loss": 0.8217,
+      "step": 1366
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.638771053196043e-06,
+      "loss": 0.973,
+      "step": 1367
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.636630273495284e-06,
+      "loss": 1.1034,
+      "step": 1368
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.634488077438419e-06,
+      "loss": 1.0471,
+      "step": 1369
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.63234446585977e-06,
+      "loss": 0.8644,
+      "step": 1370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.630199439594209e-06,
+      "loss": 0.8391,
+      "step": 1371
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.628052999477156e-06,
+      "loss": 0.8289,
+      "step": 1372
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.625905146344585e-06,
+      "loss": 0.7755,
+      "step": 1373
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.623755881033016e-06,
+      "loss": 0.8259,
+      "step": 1374
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.621605204379525e-06,
+      "loss": 0.8174,
+      "step": 1375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.619453117221737e-06,
+      "loss": 0.9404,
+      "step": 1376
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.617299620397817e-06,
+      "loss": 1.1008,
+      "step": 1377
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.615144714746493e-06,
+      "loss": 1.0099,
+      "step": 1378
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.61298840110703e-06,
+      "loss": 0.7762,
+      "step": 1379
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.61083068031925e-06,
+      "loss": 0.9883,
+      "step": 1380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.608671553223515e-06,
+      "loss": 0.9078,
+      "step": 1381
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.606511020660743e-06,
+      "loss": 0.9825,
+      "step": 1382
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.604349083472392e-06,
+      "loss": 0.9048,
+      "step": 1383
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.602185742500473e-06,
+      "loss": 0.9336,
+      "step": 1384
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.600020998587537e-06,
+      "loss": 0.8575,
+      "step": 1385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.59785485257669e-06,
+      "loss": 1.0398,
+      "step": 1386
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.595687305311578e-06,
+      "loss": 0.8378,
+      "step": 1387
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.593518357636391e-06,
+      "loss": 0.9388,
+      "step": 1388
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.591348010395874e-06,
+      "loss": 1.0285,
+      "step": 1389
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.589176264435306e-06,
+      "loss": 0.9243,
+      "step": 1390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.587003120600516e-06,
+      "loss": 0.9461,
+      "step": 1391
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.58482857973788e-06,
+      "loss": 0.8934,
+      "step": 1392
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.582652642694311e-06,
+      "loss": 1.0649,
+      "step": 1393
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.580475310317273e-06,
+      "loss": 0.9417,
+      "step": 1394
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.578296583454767e-06,
+      "loss": 1.0228,
+      "step": 1395
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.576116462955345e-06,
+      "loss": 0.9974,
+      "step": 1396
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.573934949668093e-06,
+      "loss": 0.8364,
+      "step": 1397
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.571752044442645e-06,
+      "loss": 0.92,
+      "step": 1398
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.569567748129177e-06,
+      "loss": 0.8771,
+      "step": 1399
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.567382061578404e-06,
+      "loss": 0.8692,
+      "step": 1400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.565194985641582e-06,
+      "loss": 0.984,
+      "step": 1401
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.563006521170516e-06,
+      "loss": 0.8858,
+      "step": 1402
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.56081666901754e-06,
+      "loss": 1.0462,
+      "step": 1403
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.558625430035537e-06,
+      "loss": 0.9489,
+      "step": 1404
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.556432805077927e-06,
+      "loss": 0.8713,
+      "step": 1405
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.554238794998672e-06,
+      "loss": 1.0007,
+      "step": 1406
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.552043400652269e-06,
+      "loss": 1.0622,
+      "step": 1407
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.549846622893758e-06,
+      "loss": 0.8984,
+      "step": 1408
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.547648462578717e-06,
+      "loss": 1.0005,
+      "step": 1409
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.54544892056326e-06,
+      "loss": 0.9246,
+      "step": 1410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.543247997704045e-06,
+      "loss": 1.0462,
+      "step": 1411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.54104569485826e-06,
+      "loss": 0.8148,
+      "step": 1412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.538842012883639e-06,
+      "loss": 0.9098,
+      "step": 1413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.536636952638444e-06,
+      "loss": 0.9557,
+      "step": 1414
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.534430514981483e-06,
+      "loss": 0.9795,
+      "step": 1415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.53222270077209e-06,
+      "loss": 0.873,
+      "step": 1416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.530013510870148e-06,
+      "loss": 0.9734,
+      "step": 1417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.527802946136063e-06,
+      "loss": 0.9432,
+      "step": 1418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.525591007430785e-06,
+      "loss": 0.8709,
+      "step": 1419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.523377695615797e-06,
+      "loss": 0.9138,
+      "step": 1420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.521163011553114e-06,
+      "loss": 1.0264,
+      "step": 1421
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.518946956105288e-06,
+      "loss": 0.8116,
+      "step": 1422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.516729530135406e-06,
+      "loss": 0.9081,
+      "step": 1423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.514510734507087e-06,
+      "loss": 0.9908,
+      "step": 1424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.512290570084486e-06,
+      "loss": 1.0439,
+      "step": 1425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.510069037732284e-06,
+      "loss": 1.0257,
+      "step": 1426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.507846138315702e-06,
+      "loss": 0.8882,
+      "step": 1427
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.505621872700493e-06,
+      "loss": 1.0474,
+      "step": 1428
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.503396241752942e-06,
+      "loss": 0.8826,
+      "step": 1429
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.501169246339857e-06,
+      "loss": 1.1818,
+      "step": 1430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.498940887328592e-06,
+      "loss": 1.0188,
+      "step": 1431
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.496711165587021e-06,
+      "loss": 0.9732,
+      "step": 1432
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.494480081983552e-06,
+      "loss": 0.984,
+      "step": 1433
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.492247637387123e-06,
+      "loss": 0.8419,
+      "step": 1434
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.490013832667205e-06,
+      "loss": 0.8972,
+      "step": 1435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.487778668693797e-06,
+      "loss": 0.9877,
+      "step": 1436
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.485542146337424e-06,
+      "loss": 0.8958,
+      "step": 1437
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.483304266469148e-06,
+      "loss": 1.0236,
+      "step": 1438
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.481065029960548e-06,
+      "loss": 0.936,
+      "step": 1439
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.478824437683742e-06,
+      "loss": 0.9118,
+      "step": 1440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.476582490511373e-06,
+      "loss": 0.9584,
+      "step": 1441
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.474339189316607e-06,
+      "loss": 0.9937,
+      "step": 1442
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.472094534973143e-06,
+      "loss": 0.8702,
+      "step": 1443
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.469848528355208e-06,
+      "loss": 0.9191,
+      "step": 1444
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.467601170337548e-06,
+      "loss": 0.9978,
+      "step": 1445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.465352461795443e-06,
+      "loss": 0.9037,
+      "step": 1446
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.463102403604697e-06,
+      "loss": 1.0339,
+      "step": 1447
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.460850996641635e-06,
+      "loss": 1.0317,
+      "step": 1448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.458598241783112e-06,
+      "loss": 1.015,
+      "step": 1449
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.45634413990651e-06,
+      "loss": 1.0457,
+      "step": 1450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.45408869188973e-06,
+      "loss": 0.8485,
+      "step": 1451
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.451831898611202e-06,
+      "loss": 0.9281,
+      "step": 1452
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.449573760949876e-06,
+      "loss": 1.0014,
+      "step": 1453
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.447314279785228e-06,
+      "loss": 0.9047,
+      "step": 1454
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.445053455997256e-06,
+      "loss": 0.8937,
+      "step": 1455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.442791290466483e-06,
+      "loss": 0.8027,
+      "step": 1456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.440527784073953e-06,
+      "loss": 0.9251,
+      "step": 1457
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.438262937701232e-06,
+      "loss": 0.9703,
+      "step": 1458
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.435996752230408e-06,
+      "loss": 0.9422,
+      "step": 1459
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.433729228544092e-06,
+      "loss": 0.911,
+      "step": 1460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.431460367525412e-06,
+      "loss": 0.8487,
+      "step": 1461
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.429190170058024e-06,
+      "loss": 1.1255,
+      "step": 1462
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.426918637026097e-06,
+      "loss": 0.8608,
+      "step": 1463
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.424645769314324e-06,
+      "loss": 0.9629,
+      "step": 1464
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.422371567807921e-06,
+      "loss": 1.0172,
+      "step": 1465
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.420096033392615e-06,
+      "loss": 1.0182,
+      "step": 1466
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.417819166954659e-06,
+      "loss": 0.8028,
+      "step": 1467
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.415540969380824e-06,
+      "loss": 0.8301,
+      "step": 1468
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.413261441558397e-06,
+      "loss": 0.8579,
+      "step": 1469
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.410980584375184e-06,
+      "loss": 1.0516,
+      "step": 1470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.408698398719509e-06,
+      "loss": 0.929,
+      "step": 1471
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.406414885480215e-06,
+      "loss": 0.956,
+      "step": 1472
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.404130045546657e-06,
+      "loss": 1.0098,
+      "step": 1473
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.401843879808713e-06,
+      "loss": 0.796,
+      "step": 1474
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.399556389156773e-06,
+      "loss": 0.8273,
+      "step": 1475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.397267574481746e-06,
+      "loss": 0.9059,
+      "step": 1476
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.394977436675056e-06,
+      "loss": 0.9197,
+      "step": 1477
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.392685976628638e-06,
+      "loss": 0.9879,
+      "step": 1478
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.39039319523495e-06,
+      "loss": 0.8722,
+      "step": 1479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.388099093386957e-06,
+      "loss": 0.9692,
+      "step": 1480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.38580367197814e-06,
+      "loss": 1.0045,
+      "step": 1481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.3835069319025e-06,
+      "loss": 0.9617,
+      "step": 1482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.381208874054542e-06,
+      "loss": 0.8805,
+      "step": 1483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.378909499329294e-06,
+      "loss": 0.9755,
+      "step": 1484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.376608808622287e-06,
+      "loss": 0.9124,
+      "step": 1485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.374306802829574e-06,
+      "loss": 0.9246,
+      "step": 1486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.372003482847715e-06,
+      "loss": 0.7819,
+      "step": 1487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.369698849573778e-06,
+      "loss": 0.9715,
+      "step": 1488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.367392903905354e-06,
+      "loss": 0.9585,
+      "step": 1489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.365085646740533e-06,
+      "loss": 0.9815,
+      "step": 1490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.362777078977922e-06,
+      "loss": 1.0392,
+      "step": 1491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.36046720151664e-06,
+      "loss": 1.0063,
+      "step": 1492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.35815601525631e-06,
+      "loss": 1.1032,
+      "step": 1493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.355843521097071e-06,
+      "loss": 1.0424,
+      "step": 1494
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.353529719939568e-06,
+      "loss": 0.7762,
+      "step": 1495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.351214612684954e-06,
+      "loss": 0.9752,
+      "step": 1496
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.348898200234894e-06,
+      "loss": 1.0171,
+      "step": 1497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.34658048349156e-06,
+      "loss": 0.964,
+      "step": 1498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.34426146335763e-06,
+      "loss": 0.9035,
+      "step": 1499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.341941140736292e-06,
+      "loss": 0.9203,
+      "step": 1500
+    },
+    {
+      "epoch": 0.87,
+      "eval_loss": 1.064848780632019,
+      "eval_runtime": 93.4679,
+      "eval_samples_per_second": 16.498,
+      "eval_steps_per_second": 2.065,
+      "step": 1500
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5190,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 8314415961866240.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}