diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12922 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9992992291520673,
+  "global_step": 2139,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-07,
+      "loss": 0.0185,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-06,
+      "loss": 0.0047,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-06,
+      "loss": 0.0375,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-06,
+      "loss": 0.0069,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 0.0189,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-06,
+      "loss": 0.0194,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-06,
+      "loss": 0.0251,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-06,
+      "loss": 0.0217,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.399999999999999e-06,
+      "loss": 0.0132,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 0.0209,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.599999999999999e-06,
+      "loss": 0.015,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2e-06,
+      "loss": 0.0331,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.799999999999998e-06,
+      "loss": 0.0119,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.4e-06,
+      "loss": 0.0339,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 0.046,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.6e-06,
+      "loss": 0.0613,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.02e-05,
+      "loss": 0.0245,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0799999999999998e-05,
+      "loss": 0.1201,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.14e-05,
+      "loss": 0.0402,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 0.0104,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.26e-05,
+      "loss": 0.0161,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3199999999999997e-05,
+      "loss": 0.0087,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3799999999999998e-05,
+      "loss": 0.0596,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.44e-05,
+      "loss": 0.0145,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 0.0148,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 0.0128,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6199999999999997e-05,
+      "loss": 0.0086,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.68e-05,
+      "loss": 0.01,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.74e-05,
+      "loss": 0.0448,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 0.0119,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 0.0298,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.92e-05,
+      "loss": 0.0188,
+      "step": 32
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98e-05,
+      "loss": 0.0066,
+      "step": 33
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.04e-05,
+      "loss": 0.0174,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1e-05,
+      "loss": 0.0338,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1599999999999996e-05,
+      "loss": 0.0108,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2199999999999998e-05,
+      "loss": 0.0295,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.28e-05,
+      "loss": 0.005,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.34e-05,
+      "loss": 0.0038,
+      "step": 39
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 0.0102,
+      "step": 40
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 0.003,
+      "step": 41
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.52e-05,
+      "loss": 0.1315,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5799999999999997e-05,
+      "loss": 0.1411,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 0.04,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 0.1223,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7599999999999997e-05,
+      "loss": 0.0138,
+      "step": 46
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 0.0098,
+      "step": 47
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.88e-05,
+      "loss": 0.04,
+      "step": 48
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.94e-05,
+      "loss": 0.0371,
+      "step": 49
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 0.351,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.06e-05,
+      "loss": 0.0136,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.119999999999999e-05,
+      "loss": 0.0092,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1799999999999994e-05,
+      "loss": 0.0125,
+      "step": 53
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2399999999999995e-05,
+      "loss": 0.0538,
+      "step": 54
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 0.0117,
+      "step": 55
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.36e-05,
+      "loss": 0.0142,
+      "step": 56
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.42e-05,
+      "loss": 0.0025,
+      "step": 57
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.48e-05,
+      "loss": 0.0567,
+      "step": 58
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.539999999999999e-05,
+      "loss": 0.0236,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 0.0223,
+      "step": 60
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6599999999999995e-05,
+      "loss": 0.0312,
+      "step": 61
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7199999999999996e-05,
+      "loss": 0.0112,
+      "step": 62
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.78e-05,
+      "loss": 0.0183,
+      "step": 63
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.84e-05,
+      "loss": 0.0171,
+      "step": 64
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9e-05,
+      "loss": 0.0361,
+      "step": 65
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96e-05,
+      "loss": 0.0528,
+      "step": 66
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.02e-05,
+      "loss": 0.0127,
+      "step": 67
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.08e-05,
+      "loss": 0.0148,
+      "step": 68
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.14e-05,
+      "loss": 0.0157,
+      "step": 69
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.2e-05,
+      "loss": 0.0578,
+      "step": 70
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.259999999999999e-05,
+      "loss": 0.0134,
+      "step": 71
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.319999999999999e-05,
+      "loss": 0.0128,
+      "step": 72
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.3799999999999994e-05,
+      "loss": 0.0089,
+      "step": 73
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4399999999999995e-05,
+      "loss": 0.0179,
+      "step": 74
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 0.0256,
+      "step": 75
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.56e-05,
+      "loss": 0.0308,
+      "step": 76
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.62e-05,
+      "loss": 0.0081,
+      "step": 77
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.68e-05,
+      "loss": 0.0109,
+      "step": 78
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7399999999999993e-05,
+      "loss": 0.0973,
+      "step": 79
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 0.0037,
+      "step": 80
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8599999999999995e-05,
+      "loss": 0.0311,
+      "step": 81
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 0.0416,
+      "step": 82
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.98e-05,
+      "loss": 0.0087,
+      "step": 83
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.04e-05,
+      "loss": 0.0019,
+      "step": 84
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.1e-05,
+      "loss": 0.0221,
+      "step": 85
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.1599999999999994e-05,
+      "loss": 0.0105,
+      "step": 86
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.2199999999999995e-05,
+      "loss": 0.0044,
+      "step": 87
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.279999999999999e-05,
+      "loss": 0.061,
+      "step": 88
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.339999999999999e-05,
+      "loss": 0.0087,
+      "step": 89
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 0.0456,
+      "step": 90
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.459999999999999e-05,
+      "loss": 0.0401,
+      "step": 91
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.519999999999999e-05,
+      "loss": 0.0159,
+      "step": 92
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.5799999999999994e-05,
+      "loss": 0.006,
+      "step": 93
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 0.0429,
+      "step": 94
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 0.0395,
+      "step": 95
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.76e-05,
+      "loss": 0.0021,
+      "step": 96
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.82e-05,
+      "loss": 0.0035,
+      "step": 97
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.88e-05,
+      "loss": 0.0074,
+      "step": 98
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.94e-05,
+      "loss": 0.0212,
+      "step": 99
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 0.1271,
+      "step": 100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.0599999999999996e-05,
+      "loss": 0.0279,
+      "step": 101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.12e-05,
+      "loss": 0.0243,
+      "step": 102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.18e-05,
+      "loss": 0.0226,
+      "step": 103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.239999999999999e-05,
+      "loss": 0.0034,
+      "step": 104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 0.017,
+      "step": 105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.359999999999999e-05,
+      "loss": 0.033,
+      "step": 106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.419999999999999e-05,
+      "loss": 0.0183,
+      "step": 107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.479999999999999e-05,
+      "loss": 0.0422,
+      "step": 108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.539999999999999e-05,
+      "loss": 0.0124,
+      "step": 109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 0.0348,
+      "step": 110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.659999999999999e-05,
+      "loss": 0.0534,
+      "step": 111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.72e-05,
+      "loss": 0.029,
+      "step": 112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.78e-05,
+      "loss": 0.0114,
+      "step": 113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.84e-05,
+      "loss": 0.0077,
+      "step": 114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.9e-05,
+      "loss": 0.0038,
+      "step": 115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.96e-05,
+      "loss": 0.0055,
+      "step": 116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.02e-05,
+      "loss": 0.0269,
+      "step": 117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.079999999999999e-05,
+      "loss": 0.0277,
+      "step": 118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.139999999999999e-05,
+      "loss": 0.0274,
+      "step": 119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 0.0252,
+      "step": 120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.259999999999999e-05,
+      "loss": 0.0254,
+      "step": 121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.319999999999999e-05,
+      "loss": 0.0068,
+      "step": 122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.379999999999999e-05,
+      "loss": 0.023,
+      "step": 123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.439999999999999e-05,
+      "loss": 0.0153,
+      "step": 124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.5e-05,
+      "loss": 0.0422,
+      "step": 125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.56e-05,
+      "loss": 0.0483,
+      "step": 126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.62e-05,
+      "loss": 0.0112,
+      "step": 127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.68e-05,
+      "loss": 0.0227,
+      "step": 128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.74e-05,
+      "loss": 0.036,
+      "step": 129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.8e-05,
+      "loss": 0.0086,
+      "step": 130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.86e-05,
+      "loss": 0.0035,
+      "step": 131
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.92e-05,
+      "loss": 0.0539,
+      "step": 132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.98e-05,
+      "loss": 0.0284,
+      "step": 133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.04e-05,
+      "loss": 0.0067,
+      "step": 134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.1e-05,
+      "loss": 0.0636,
+      "step": 135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.16e-05,
+      "loss": 0.0235,
+      "step": 136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.22e-05,
+      "loss": 0.0065,
+      "step": 137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.28e-05,
+      "loss": 0.1075,
+      "step": 138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.34e-05,
+      "loss": 0.0199,
+      "step": 139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.4e-05,
+      "loss": 0.0177,
+      "step": 140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.459999999999998e-05,
+      "loss": 0.0151,
+      "step": 141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.519999999999998e-05,
+      "loss": 0.0114,
+      "step": 142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.579999999999998e-05,
+      "loss": 0.0615,
+      "step": 143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.639999999999999e-05,
+      "loss": 0.0593,
+      "step": 144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 0.0491,
+      "step": 145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.759999999999999e-05,
+      "loss": 0.0699,
+      "step": 146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.819999999999999e-05,
+      "loss": 0.0241,
+      "step": 147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.879999999999999e-05,
+      "loss": 0.0289,
+      "step": 148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.939999999999999e-05,
+      "loss": 0.0397,
+      "step": 149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 0.0226,
+      "step": 150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.059999999999999e-05,
+      "loss": 0.0327,
+      "step": 151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.12e-05,
+      "loss": 0.0201,
+      "step": 152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.18e-05,
+      "loss": 0.0115,
+      "step": 153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.24e-05,
+      "loss": 0.0203,
+      "step": 154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.3e-05,
+      "loss": 0.0317,
+      "step": 155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.36e-05,
+      "loss": 0.0232,
+      "step": 156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.419999999999999e-05,
+      "loss": 0.0181,
+      "step": 157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.479999999999999e-05,
+      "loss": 0.0428,
+      "step": 158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.539999999999999e-05,
+      "loss": 0.0034,
+      "step": 159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 0.0417,
+      "step": 160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.659999999999999e-05,
+      "loss": 0.0309,
+      "step": 161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.719999999999999e-05,
+      "loss": 0.0221,
+      "step": 162
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.779999999999999e-05,
+      "loss": 0.0264,
+      "step": 163
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.839999999999999e-05,
+      "loss": 0.0139,
+      "step": 164
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.9e-05,
+      "loss": 0.0102,
+      "step": 165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.96e-05,
+      "loss": 0.0239,
+      "step": 166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001002,
+      "loss": 0.0096,
+      "step": 167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001008,
+      "loss": 0.0123,
+      "step": 168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001014,
+      "loss": 0.015,
+      "step": 169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000102,
+      "loss": 0.0345,
+      "step": 170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001026,
+      "loss": 0.0288,
+      "step": 171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00010319999999999999,
+      "loss": 0.0145,
+      "step": 172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00010379999999999999,
+      "loss": 0.0264,
+      "step": 173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00010439999999999999,
+      "loss": 0.0597,
+      "step": 174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 0.0676,
+      "step": 175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00010559999999999998,
+      "loss": 0.039,
+      "step": 176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00010619999999999998,
+      "loss": 0.0426,
+      "step": 177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00010679999999999998,
+      "loss": 0.0203,
+      "step": 178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00010739999999999998,
+      "loss": 0.033,
+      "step": 179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 0.0067,
+      "step": 180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00010859999999999998,
+      "loss": 0.0213,
+      "step": 181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00010919999999999998,
+      "loss": 0.0191,
+      "step": 182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00010979999999999999,
+      "loss": 0.0033,
+      "step": 183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011039999999999999,
+      "loss": 0.003,
+      "step": 184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011099999999999999,
+      "loss": 0.0054,
+      "step": 185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011159999999999999,
+      "loss": 0.0388,
+      "step": 186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011219999999999999,
+      "loss": 0.0124,
+      "step": 187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011279999999999999,
+      "loss": 0.0467,
+      "step": 188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011339999999999999,
+      "loss": 0.0328,
+      "step": 189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 0.0331,
+      "step": 190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001146,
+      "loss": 0.0312,
+      "step": 191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001152,
+      "loss": 0.0092,
+      "step": 192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001158,
+      "loss": 0.0372,
+      "step": 193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001164,
+      "loss": 0.0171,
+      "step": 194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000117,
+      "loss": 0.0336,
+      "step": 195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001176,
+      "loss": 0.0213,
+      "step": 196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001182,
+      "loss": 0.0422,
+      "step": 197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001188,
+      "loss": 0.0252,
+      "step": 198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001194,
+      "loss": 0.0219,
+      "step": 199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 0.0639,
+      "step": 200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00012059999999999999,
+      "loss": 0.0324,
+      "step": 201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00012119999999999999,
+      "loss": 0.0153,
+      "step": 202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00012179999999999999,
+      "loss": 0.0155,
+      "step": 203
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001224,
+      "loss": 0.0307,
+      "step": 204
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00012299999999999998,
+      "loss": 0.0317,
+      "step": 205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001236,
+      "loss": 0.0205,
+      "step": 206
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00012419999999999998,
+      "loss": 0.0398,
+      "step": 207
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00012479999999999997,
+      "loss": 0.0502,
+      "step": 208
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00012539999999999999,
+      "loss": 0.0164,
+      "step": 209
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 0.0447,
+      "step": 210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001266,
+      "loss": 0.0335,
+      "step": 211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00012719999999999997,
+      "loss": 0.0434,
+      "step": 212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001278,
+      "loss": 0.0487,
+      "step": 213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00012839999999999998,
+      "loss": 0.011,
+      "step": 214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000129,
+      "loss": 0.0566,
+      "step": 215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00012959999999999998,
+      "loss": 0.0396,
+      "step": 216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001302,
+      "loss": 0.0111,
+      "step": 217
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00013079999999999998,
+      "loss": 0.0222,
+      "step": 218
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001314,
+      "loss": 0.0223,
+      "step": 219
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 0.0391,
+      "step": 220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001326,
+      "loss": 0.0285,
+      "step": 221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00013319999999999999,
+      "loss": 0.0137,
+      "step": 222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001338,
+      "loss": 0.0031,
+      "step": 223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001344,
+      "loss": 0.0713,
+      "step": 224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000135,
+      "loss": 0.0897,
+      "step": 225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001356,
+      "loss": 0.1005,
+      "step": 226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001362,
+      "loss": 0.037,
+      "step": 227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001368,
+      "loss": 0.0208,
+      "step": 228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001374,
+      "loss": 0.0253,
+      "step": 229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000138,
+      "loss": 0.0359,
+      "step": 230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001386,
+      "loss": 0.0175,
+      "step": 231
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001392,
+      "loss": 0.0579,
+      "step": 232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00013979999999999998,
+      "loss": 0.0149,
+      "step": 233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001404,
+      "loss": 0.0566,
+      "step": 234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014099999999999998,
+      "loss": 0.0317,
+      "step": 235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014159999999999997,
+      "loss": 0.0418,
+      "step": 236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001422,
+      "loss": 0.0273,
+      "step": 237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014279999999999997,
+      "loss": 0.0034,
+      "step": 238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001434,
+      "loss": 0.0385,
+      "step": 239
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 0.0808,
+      "step": 240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001446,
+      "loss": 0.0585,
+      "step": 241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014519999999999998,
+      "loss": 0.086,
+      "step": 242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001458,
+      "loss": 0.0506,
+      "step": 243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014639999999999998,
+      "loss": 0.0313,
+      "step": 244
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000147,
+      "loss": 0.0445,
+      "step": 245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014759999999999998,
+      "loss": 0.0515,
+      "step": 246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001482,
+      "loss": 0.0371,
+      "step": 247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014879999999999998,
+      "loss": 0.0567,
+      "step": 248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001494,
+      "loss": 0.0289,
+      "step": 249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015,
+      "loss": 0.1516,
+      "step": 250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015059999999999997,
+      "loss": 0.0726,
+      "step": 251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001512,
+      "loss": 0.0262,
+      "step": 252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015179999999999998,
+      "loss": 0.0704,
+      "step": 253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001524,
+      "loss": 0.019,
+      "step": 254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015299999999999998,
+      "loss": 0.0283,
+      "step": 255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001536,
+      "loss": 0.0285,
+      "step": 256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015419999999999998,
+      "loss": 0.0103,
+      "step": 257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001548,
+      "loss": 0.012,
+      "step": 258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015539999999999998,
+      "loss": 0.0414,
+      "step": 259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000156,
+      "loss": 0.0188,
+      "step": 260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015659999999999998,
+      "loss": 0.027,
+      "step": 261
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001572,
+      "loss": 0.075,
+      "step": 262
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001578,
+      "loss": 0.0702,
+      "step": 263
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001584,
+      "loss": 0.0742,
+      "step": 264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000159,
+      "loss": 0.0371,
+      "step": 265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001596,
+      "loss": 0.0436,
+      "step": 266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001602,
+      "loss": 0.0906,
+      "step": 267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001608,
+      "loss": 0.0286,
+      "step": 268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001614,
+      "loss": 0.0268,
+      "step": 269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000162,
+      "loss": 0.0452,
+      "step": 270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001626,
+      "loss": 0.059,
+      "step": 271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001632,
+      "loss": 0.0411,
+      "step": 272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001638,
+      "loss": 0.0262,
+      "step": 273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001644,
+      "loss": 0.0558,
+      "step": 274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000165,
+      "loss": 0.0638,
+      "step": 275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001656,
+      "loss": 0.0502,
+      "step": 276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001662,
+      "loss": 0.022,
+      "step": 277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001668,
+      "loss": 0.067,
+      "step": 278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001674,
+      "loss": 0.0809,
+      "step": 279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000168,
+      "loss": 0.046,
+      "step": 280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001686,
+      "loss": 0.0201,
+      "step": 281
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016919999999999997,
+      "loss": 0.0544,
+      "step": 282
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016979999999999998,
+      "loss": 0.0186,
+      "step": 283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017039999999999997,
+      "loss": 0.0318,
+      "step": 284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017099999999999998,
+      "loss": 0.044,
+      "step": 285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017159999999999997,
+      "loss": 0.0318,
+      "step": 286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017219999999999998,
+      "loss": 0.0564,
+      "step": 287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017279999999999997,
+      "loss": 0.0456,
+      "step": 288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017339999999999996,
+      "loss": 0.0522,
+      "step": 289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 0.0429,
+      "step": 290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017459999999999996,
+      "loss": 0.0143,
+      "step": 291
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017519999999999998,
+      "loss": 0.0763,
+      "step": 292
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017579999999999996,
+      "loss": 0.0593,
+      "step": 293
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017639999999999998,
+      "loss": 0.024,
+      "step": 294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 0.0108,
+      "step": 295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017759999999999998,
+      "loss": 0.0115,
+      "step": 296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017819999999999997,
+      "loss": 0.0152,
+      "step": 297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017879999999999998,
+      "loss": 0.1313,
+      "step": 298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017939999999999997,
+      "loss": 0.0546,
+      "step": 299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 0.0144,
+      "step": 300
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 0.0947057232260704,
+      "eval_runtime": 12867.7645,
+      "eval_samples_per_second": 0.205,
+      "eval_steps_per_second": 0.103,
+      "eval_wer": 0.074910749702499,
+      "step": 300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018059999999999997,
+      "loss": 0.0439,
+      "step": 301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018119999999999999,
+      "loss": 0.0363,
+      "step": 302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018179999999999997,
+      "loss": 0.0768,
+      "step": 303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001824,
+      "loss": 0.0381,
+      "step": 304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018299999999999998,
+      "loss": 0.0701,
+      "step": 305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001836,
+      "loss": 0.019,
+      "step": 306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018419999999999998,
+      "loss": 0.0439,
+      "step": 307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001848,
+      "loss": 0.066,
+      "step": 308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018539999999999998,
+      "loss": 0.0226,
+      "step": 309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000186,
+      "loss": 0.0244,
+      "step": 310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018659999999999998,
+      "loss": 0.0758,
+      "step": 311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001872,
+      "loss": 0.0534,
+      "step": 312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018779999999999998,
+      "loss": 0.0327,
+      "step": 313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018839999999999997,
+      "loss": 0.0554,
+      "step": 314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018899999999999999,
+      "loss": 0.0471,
+      "step": 315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018959999999999997,
+      "loss": 0.0253,
+      "step": 316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001902,
+      "loss": 0.0157,
+      "step": 317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019079999999999998,
+      "loss": 0.0248,
+      "step": 318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001914,
+      "loss": 0.0489,
+      "step": 319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 0.0303,
+      "step": 320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001926,
+      "loss": 0.0459,
+      "step": 321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019319999999999998,
+      "loss": 0.0707,
+      "step": 322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001938,
+      "loss": 0.019,
+      "step": 323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019439999999999998,
+      "loss": 0.126,
+      "step": 324
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000195,
+      "loss": 0.0396,
+      "step": 325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019559999999999998,
+      "loss": 0.0287,
+      "step": 326
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001962,
+      "loss": 0.0538,
+      "step": 327
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019679999999999999,
+      "loss": 0.0413,
+      "step": 328
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001974,
+      "loss": 0.0505,
+      "step": 329
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000198,
+      "loss": 0.0654,
+      "step": 330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001986,
+      "loss": 0.0537,
+      "step": 331
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001992,
+      "loss": 0.0388,
+      "step": 332
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001998,
+      "loss": 0.0214,
+      "step": 333
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002004,
+      "loss": 0.0611,
+      "step": 334
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000201,
+      "loss": 0.0215,
+      "step": 335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002016,
+      "loss": 0.0735,
+      "step": 336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002022,
+      "loss": 0.0451,
+      "step": 337
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002028,
+      "loss": 0.0775,
+      "step": 338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00020339999999999998,
+      "loss": 0.0408,
+      "step": 339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000204,
+      "loss": 0.023,
+      "step": 340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00020459999999999999,
+      "loss": 0.0824,
+      "step": 341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002052,
+      "loss": 0.0596,
+      "step": 342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002058,
+      "loss": 0.1135,
+      "step": 343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00020639999999999998,
+      "loss": 0.0232,
+      "step": 344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 0.0086,
+      "step": 345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00020759999999999998,
+      "loss": 0.0213,
+      "step": 346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020819999999999996,
+      "loss": 0.0248,
+      "step": 347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020879999999999998,
+      "loss": 0.4063,
+      "step": 348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020939999999999997,
+      "loss": 0.0521,
+      "step": 349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 0.0528,
+      "step": 350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021059999999999997,
+      "loss": 0.0701,
+      "step": 351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021119999999999996,
+      "loss": 0.0515,
+      "step": 352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021179999999999997,
+      "loss": 0.081,
+      "step": 353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021239999999999996,
+      "loss": 0.0491,
+      "step": 354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 0.0438,
+      "step": 355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021359999999999996,
+      "loss": 0.0588,
+      "step": 356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021419999999999998,
+      "loss": 0.1017,
+      "step": 357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021479999999999996,
+      "loss": 0.0522,
+      "step": 358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021539999999999998,
+      "loss": 0.0205,
+      "step": 359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 0.1065,
+      "step": 360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021659999999999998,
+      "loss": 0.0536,
+      "step": 361
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021719999999999997,
+      "loss": 0.0994,
+      "step": 362
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021779999999999998,
+      "loss": 0.036,
+      "step": 363
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021839999999999997,
+      "loss": 0.0691,
+      "step": 364
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 0.0356,
+      "step": 365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021959999999999997,
+      "loss": 0.0286,
+      "step": 366
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00022019999999999999,
+      "loss": 0.0414,
+      "step": 367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00022079999999999997,
+      "loss": 0.1357,
+      "step": 368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002214,
+      "loss": 0.0669,
+      "step": 369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 0.0572,
+      "step": 370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002226,
+      "loss": 0.0989,
+      "step": 371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00022319999999999998,
+      "loss": 0.0434,
+      "step": 372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002238,
+      "loss": 0.0683,
+      "step": 373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00022439999999999998,
+      "loss": 0.0594,
+      "step": 374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000225,
+      "loss": 0.0066,
+      "step": 375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022559999999999998,
+      "loss": 0.0704,
+      "step": 376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022619999999999997,
+      "loss": 0.0437,
+      "step": 377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022679999999999998,
+      "loss": 0.0711,
+      "step": 378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022739999999999997,
+      "loss": 0.0243,
+      "step": 379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 0.1348,
+      "step": 380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022859999999999997,
+      "loss": 0.0579,
+      "step": 381
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002292,
+      "loss": 0.094,
+      "step": 382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022979999999999997,
+      "loss": 0.0824,
+      "step": 383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002304,
+      "loss": 0.0335,
+      "step": 384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00023099999999999998,
+      "loss": 0.1456,
+      "step": 385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002316,
+      "loss": 0.0271,
+      "step": 386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00023219999999999998,
+      "loss": 0.1348,
+      "step": 387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002328,
+      "loss": 0.0561,
+      "step": 388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00023339999999999998,
+      "loss": 0.0469,
+      "step": 389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000234,
+      "loss": 0.0807,
+      "step": 390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00023459999999999998,
+      "loss": 0.0484,
+      "step": 391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002352,
+      "loss": 0.0529,
+      "step": 392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00023579999999999999,
+      "loss": 0.0463,
+      "step": 393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002364,
+      "loss": 0.0724,
+      "step": 394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000237,
+      "loss": 0.1588,
+      "step": 395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002376,
+      "loss": 0.0982,
+      "step": 396
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002382,
+      "loss": 0.0477,
+      "step": 397
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002388,
+      "loss": 0.0722,
+      "step": 398
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002394,
+      "loss": 0.1838,
+      "step": 399
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 0.0389,
+      "step": 400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002406,
+      "loss": 0.0655,
+      "step": 401
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024119999999999998,
+      "loss": 0.0617,
+      "step": 402
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002418,
+      "loss": 0.0927,
+      "step": 403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024239999999999998,
+      "loss": 0.0759,
+      "step": 404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000243,
+      "loss": 0.117,
+      "step": 405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024359999999999999,
+      "loss": 0.1214,
+      "step": 406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024419999999999997,
+      "loss": 0.1468,
+      "step": 407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002448,
+      "loss": 0.0662,
+      "step": 408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024539999999999995,
+      "loss": 0.0835,
+      "step": 409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 0.0867,
+      "step": 410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002466,
+      "loss": 0.1268,
+      "step": 411
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002472,
+      "loss": 0.058,
+      "step": 412
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00024779999999999995,
+      "loss": 0.1017,
+      "step": 413
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00024839999999999997,
+      "loss": 0.0532,
+      "step": 414
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000249,
+      "loss": 0.1115,
+      "step": 415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00024959999999999994,
+      "loss": 0.0753,
+      "step": 416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00025019999999999996,
+      "loss": 0.1401,
+      "step": 417
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00025079999999999997,
+      "loss": 0.0633,
+      "step": 418
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002514,
+      "loss": 0.0659,
+      "step": 419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 0.0691,
+      "step": 420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00025259999999999996,
+      "loss": 0.0637,
+      "step": 421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002532,
+      "loss": 0.0801,
+      "step": 422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002538,
+      "loss": 0.0314,
+      "step": 423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00025439999999999995,
+      "loss": 0.0737,
+      "step": 424
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 0.039,
+      "step": 425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002556,
+      "loss": 0.1474,
+      "step": 426
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002562,
+      "loss": 0.0729,
+      "step": 427
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00025679999999999995,
+      "loss": 0.1171,
+      "step": 428
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00025739999999999997,
+      "loss": 0.0601,
+      "step": 429
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000258,
+      "loss": 0.0789,
+      "step": 430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002586,
+      "loss": 0.1003,
+      "step": 431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00025919999999999996,
+      "loss": 0.0658,
+      "step": 432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00025979999999999997,
+      "loss": 0.0462,
+      "step": 433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002604,
+      "loss": 0.2161,
+      "step": 434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000261,
+      "loss": 0.0805,
+      "step": 435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026159999999999996,
+      "loss": 0.0881,
+      "step": 436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002622,
+      "loss": 0.1052,
+      "step": 437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002628,
+      "loss": 0.0791,
+      "step": 438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026339999999999995,
+      "loss": 0.0665,
+      "step": 439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 0.0786,
+      "step": 440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002646,
+      "loss": 0.1186,
+      "step": 441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002652,
+      "loss": 0.0622,
+      "step": 442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026579999999999996,
+      "loss": 0.0375,
+      "step": 443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026639999999999997,
+      "loss": 0.0831,
+      "step": 444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000267,
+      "loss": 0.0671,
+      "step": 445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002676,
+      "loss": 0.0669,
+      "step": 446
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026819999999999996,
+      "loss": 0.1293,
+      "step": 447
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002688,
+      "loss": 0.0603,
+      "step": 448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002694,
+      "loss": 0.1086,
+      "step": 449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00027,
+      "loss": 0.1944,
+      "step": 450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00027059999999999996,
+      "loss": 0.1334,
+      "step": 451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002712,
+      "loss": 0.0443,
+      "step": 452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002718,
+      "loss": 0.0384,
+      "step": 453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002724,
+      "loss": 0.1385,
+      "step": 454
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00027299999999999997,
+      "loss": 0.1088,
+      "step": 455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002736,
+      "loss": 0.0389,
+      "step": 456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002742,
+      "loss": 0.0905,
+      "step": 457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002748,
+      "loss": 0.1157,
+      "step": 458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00027539999999999997,
+      "loss": 0.0481,
+      "step": 459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000276,
+      "loss": 0.1397,
+      "step": 460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002766,
+      "loss": 0.13,
+      "step": 461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002772,
+      "loss": 0.0712,
+      "step": 462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002778,
+      "loss": 0.0954,
+      "step": 463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002784,
+      "loss": 0.1214,
+      "step": 464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000279,
+      "loss": 0.0376,
+      "step": 465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027959999999999997,
+      "loss": 0.0969,
+      "step": 466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002802,
+      "loss": 0.0635,
+      "step": 467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002808,
+      "loss": 0.0783,
+      "step": 468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028139999999999996,
+      "loss": 0.0727,
+      "step": 469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 0.1053,
+      "step": 470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002826,
+      "loss": 0.0659,
+      "step": 471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028319999999999994,
+      "loss": 0.0651,
+      "step": 472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028379999999999996,
+      "loss": 0.0464,
+      "step": 473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002844,
+      "loss": 0.2091,
+      "step": 474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000285,
+      "loss": 0.1503,
+      "step": 475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028559999999999995,
+      "loss": 0.0902,
+      "step": 476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028619999999999996,
+      "loss": 0.0747,
+      "step": 477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002868,
+      "loss": 0.0374,
+      "step": 478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028739999999999994,
+      "loss": 0.066,
+      "step": 479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 0.0282,
+      "step": 480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028859999999999997,
+      "loss": 0.0661,
+      "step": 481
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002892,
+      "loss": 0.0436,
+      "step": 482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028979999999999994,
+      "loss": 0.0801,
+      "step": 483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00029039999999999996,
+      "loss": 0.0713,
+      "step": 484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00029099999999999997,
+      "loss": 0.0844,
+      "step": 485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002916,
+      "loss": 0.0634,
+      "step": 486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00029219999999999995,
+      "loss": 0.0679,
+      "step": 487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00029279999999999996,
+      "loss": 0.0732,
+      "step": 488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002934,
+      "loss": 0.0361,
+      "step": 489
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000294,
+      "loss": 0.0236,
+      "step": 490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00029459999999999995,
+      "loss": 0.0823,
+      "step": 491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00029519999999999997,
+      "loss": 0.0978,
+      "step": 492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002958,
+      "loss": 0.0217,
+      "step": 493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002964,
+      "loss": 0.0274,
+      "step": 494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 0.3879,
+      "step": 495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00029759999999999997,
+      "loss": 0.4896,
+      "step": 496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002982,
+      "loss": 0.0552,
+      "step": 497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002988,
+      "loss": 0.0916,
+      "step": 498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00029939999999999996,
+      "loss": 0.121,
+      "step": 499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003,
+      "loss": 0.4469,
+      "step": 500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00029981696156192796,
+      "loss": 0.1052,
+      "step": 501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000299633923123856,
+      "loss": 0.1047,
+      "step": 502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000299450884685784,
+      "loss": 0.0617,
+      "step": 503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000299267846247712,
+      "loss": 0.1185,
+      "step": 504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00029908480780964,
+      "loss": 0.1395,
+      "step": 505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000298901769371568,
+      "loss": 0.2166,
+      "step": 506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000298718730933496,
+      "loss": 0.1048,
+      "step": 507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000298535692495424,
+      "loss": 0.1246,
+      "step": 508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000298352654057352,
+      "loss": 0.0566,
+      "step": 509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00029816961561928004,
+      "loss": 0.0744,
+      "step": 510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000297986577181208,
+      "loss": 0.1782,
+      "step": 511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00029780353874313606,
+      "loss": 0.0552,
+      "step": 512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00029762050030506405,
+      "loss": 0.0459,
+      "step": 513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00029743746186699203,
+      "loss": 0.1028,
+      "step": 514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00029725442342892007,
+      "loss": 0.0491,
+      "step": 515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00029707138499084806,
+      "loss": 0.0585,
+      "step": 516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00029688834655277604,
+      "loss": 0.136,
+      "step": 517
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002967053081147041,
+      "loss": 0.0672,
+      "step": 518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00029652226967663206,
+      "loss": 0.0317,
+      "step": 519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00029633923123856005,
+      "loss": 0.1165,
+      "step": 520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002961561928004881,
+      "loss": 0.0999,
+      "step": 521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00029597315436241607,
+      "loss": 0.0706,
+      "step": 522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002957901159243441,
+      "loss": 0.0831,
+      "step": 523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002956070774862721,
+      "loss": 0.0792,
+      "step": 524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029542403904820013,
+      "loss": 0.1255,
+      "step": 525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029524100061012807,
+      "loss": 0.1076,
+      "step": 526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002950579621720561,
+      "loss": 0.0908,
+      "step": 527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002948749237339841,
+      "loss": 0.0418,
+      "step": 528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029469188529591213,
+      "loss": 0.0905,
+      "step": 529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002945088468578401,
+      "loss": 0.075,
+      "step": 530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029432580841976815,
+      "loss": 0.0578,
+      "step": 531
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029414276998169614,
+      "loss": 0.0622,
+      "step": 532
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002939597315436241,
+      "loss": 0.0878,
+      "step": 533
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029377669310555216,
+      "loss": 0.1033,
+      "step": 534
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029359365466748014,
+      "loss": 0.0297,
+      "step": 535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029341061622940813,
+      "loss": 0.1305,
+      "step": 536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029322757779133617,
+      "loss": 0.1155,
+      "step": 537
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029304453935326415,
+      "loss": 0.0762,
+      "step": 538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029286150091519214,
+      "loss": 0.1419,
+      "step": 539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002926784624771202,
+      "loss": 0.1764,
+      "step": 540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029249542403904816,
+      "loss": 0.1484,
+      "step": 541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002923123856009762,
+      "loss": 0.0829,
+      "step": 542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002921293471629042,
+      "loss": 0.0947,
+      "step": 543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002919463087248322,
+      "loss": 0.1027,
+      "step": 544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002917632702867602,
+      "loss": 0.0589,
+      "step": 545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002915802318486882,
+      "loss": 0.288,
+      "step": 546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002913971934106162,
+      "loss": 0.0829,
+      "step": 547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002912141549725442,
+      "loss": 0.0346,
+      "step": 548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002910311165344722,
+      "loss": 0.1314,
+      "step": 549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029084807809640024,
+      "loss": 0.0952,
+      "step": 550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002906650396583282,
+      "loss": 0.1314,
+      "step": 551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002904820012202562,
+      "loss": 0.1065,
+      "step": 552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029029896278218425,
+      "loss": 0.0256,
+      "step": 553
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029011592434411223,
+      "loss": 0.1172,
+      "step": 554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002899328859060403,
+      "loss": 0.1415,
+      "step": 555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028974984746796826,
+      "loss": 0.0847,
+      "step": 556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028956680902989624,
+      "loss": 0.0813,
+      "step": 557
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028938377059182423,
+      "loss": 0.0661,
+      "step": 558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028920073215375227,
+      "loss": 0.0749,
+      "step": 559
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028901769371568025,
+      "loss": 0.0456,
+      "step": 560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002888346552776083,
+      "loss": 0.0544,
+      "step": 561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002886516168395363,
+      "loss": 0.0409,
+      "step": 562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002884685784014643,
+      "loss": 0.0799,
+      "step": 563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002882855399633923,
+      "loss": 0.0969,
+      "step": 564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002881025015253203,
+      "loss": 0.0656,
+      "step": 565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028791946308724827,
+      "loss": 0.0876,
+      "step": 566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002877364246491763,
+      "loss": 0.0948,
+      "step": 567
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002875533862111043,
+      "loss": 0.0883,
+      "step": 568
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028737034777303233,
+      "loss": 0.0842,
+      "step": 569
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002871873093349603,
+      "loss": 0.1123,
+      "step": 570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002870042708968883,
+      "loss": 0.0639,
+      "step": 571
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028682123245881634,
+      "loss": 0.1019,
+      "step": 572
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002866381940207443,
+      "loss": 0.0715,
+      "step": 573
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028645515558267236,
+      "loss": 0.0645,
+      "step": 574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028627211714460035,
+      "loss": 0.0628,
+      "step": 575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028608907870652833,
+      "loss": 0.0659,
+      "step": 576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002859060402684563,
+      "loss": 0.1341,
+      "step": 577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028572300183038436,
+      "loss": 0.1853,
+      "step": 578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028553996339231234,
+      "loss": 0.0789,
+      "step": 579
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002853569249542404,
+      "loss": 0.171,
+      "step": 580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028517388651616837,
+      "loss": 0.0676,
+      "step": 581
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002849908480780964,
+      "loss": 0.0426,
+      "step": 582
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002848078096400244,
+      "loss": 0.0767,
+      "step": 583
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002846247712019524,
+      "loss": 0.0849,
+      "step": 584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002844417327638804,
+      "loss": 0.1138,
+      "step": 585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002842586943258084,
+      "loss": 0.0723,
+      "step": 586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002840756558877364,
+      "loss": 0.0698,
+      "step": 587
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002838926174496644,
+      "loss": 0.0902,
+      "step": 588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002837095790115924,
+      "loss": 0.0403,
+      "step": 589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002835265405735204,
+      "loss": 0.0637,
+      "step": 590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028334350213544843,
+      "loss": 0.1116,
+      "step": 591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002831604636973764,
+      "loss": 0.0976,
+      "step": 592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028297742525930445,
+      "loss": 0.0749,
+      "step": 593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028279438682123244,
+      "loss": 0.0675,
+      "step": 594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002826113483831605,
+      "loss": 0.0474,
+      "step": 595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028242830994508846,
+      "loss": 0.4751,
+      "step": 596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028224527150701645,
+      "loss": 0.1161,
+      "step": 597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028206223306894443,
+      "loss": 0.1717,
+      "step": 598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028187919463087247,
+      "loss": 0.15,
+      "step": 599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028169615619280045,
+      "loss": 0.1408,
+      "step": 600
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 0.1346682608127594,
+      "eval_runtime": 13252.1056,
+      "eval_samples_per_second": 0.199,
+      "eval_steps_per_second": 0.1,
+      "eval_wer": 0.13627528758429194,
+      "step": 600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002815131177547285,
+      "loss": 0.0911,
+      "step": 601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002813300793166565,
+      "loss": 0.1032,
+      "step": 602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028114704087858446,
+      "loss": 0.0994,
+      "step": 603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002809640024405125,
+      "loss": 0.0355,
+      "step": 604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002807809640024405,
+      "loss": 0.1308,
+      "step": 605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002805979255643685,
+      "loss": 0.0735,
+      "step": 606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028041488712629646,
+      "loss": 0.1396,
+      "step": 607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002802318486882245,
+      "loss": 0.1022,
+      "step": 608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002800488102501525,
+      "loss": 0.0684,
+      "step": 609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002798657718120805,
+      "loss": 0.1135,
+      "step": 610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002796827333740085,
+      "loss": 0.0965,
+      "step": 611
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00027949969493593654,
+      "loss": 0.094,
+      "step": 612
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00027931665649786453,
+      "loss": 0.107,
+      "step": 613
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00027913361805979257,
+      "loss": 0.1137,
+      "step": 614
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00027895057962172055,
+      "loss": 0.0582,
+      "step": 615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00027876754118364854,
+      "loss": 0.0478,
+      "step": 616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002785845027455765,
+      "loss": 0.0961,
+      "step": 617
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00027840146430750456,
+      "loss": 0.0641,
+      "step": 618
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00027821842586943254,
+      "loss": 0.0559,
+      "step": 619
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002780353874313606,
+      "loss": 0.107,
+      "step": 620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00027785234899328857,
+      "loss": 0.0405,
+      "step": 621
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00027766931055521655,
+      "loss": 0.0844,
+      "step": 622
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002774862721171446,
+      "loss": 0.0934,
+      "step": 623
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002773032336790726,
+      "loss": 0.0654,
+      "step": 624
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002771201952410006,
+      "loss": 0.0768,
+      "step": 625
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002769371568029286,
+      "loss": 0.0752,
+      "step": 626
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002767541183648566,
+      "loss": 0.0742,
+      "step": 627
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027657107992678457,
+      "loss": 0.1625,
+      "step": 628
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002763880414887126,
+      "loss": 0.0883,
+      "step": 629
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002762050030506406,
+      "loss": 0.067,
+      "step": 630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027602196461256863,
+      "loss": 0.101,
+      "step": 631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002758389261744966,
+      "loss": 0.0738,
+      "step": 632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027565588773642466,
+      "loss": 0.0673,
+      "step": 633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027547284929835264,
+      "loss": 0.0898,
+      "step": 634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002752898108602806,
+      "loss": 0.0525,
+      "step": 635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027510677242220867,
+      "loss": 0.0329,
+      "step": 636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027492373398413665,
+      "loss": 0.1215,
+      "step": 637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027474069554606463,
+      "loss": 0.0539,
+      "step": 638
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002745576571079926,
+      "loss": 0.0715,
+      "step": 639
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027437461866992066,
+      "loss": 0.0288,
+      "step": 640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027419158023184864,
+      "loss": 0.0754,
+      "step": 641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002740085417937767,
+      "loss": 0.1115,
+      "step": 642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027382550335570467,
+      "loss": 0.0903,
+      "step": 643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002736424649176327,
+      "loss": 0.1069,
+      "step": 644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002734594264795607,
+      "loss": 0.0659,
+      "step": 645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027327638804148873,
+      "loss": 0.1134,
+      "step": 646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027309334960341666,
+      "loss": 0.0664,
+      "step": 647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002729103111653447,
+      "loss": 0.1328,
+      "step": 648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002727272727272727,
+      "loss": 0.1596,
+      "step": 649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002725442342892007,
+      "loss": 0.1375,
+      "step": 650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002723611958511287,
+      "loss": 0.1053,
+      "step": 651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002721781574130567,
+      "loss": 0.092,
+      "step": 652
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027199511897498473,
+      "loss": 0.102,
+      "step": 653
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002718120805369127,
+      "loss": 0.1063,
+      "step": 654
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027162904209884075,
+      "loss": 0.1538,
+      "step": 655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027144600366076874,
+      "loss": 0.0768,
+      "step": 656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002712629652226967,
+      "loss": 0.1355,
+      "step": 657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002710799267846247,
+      "loss": 0.0878,
+      "step": 658
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027089688834655275,
+      "loss": 0.0832,
+      "step": 659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027071384990848073,
+      "loss": 0.0959,
+      "step": 660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027053081147040877,
+      "loss": 0.1058,
+      "step": 661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027034777303233676,
+      "loss": 0.0616,
+      "step": 662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002701647345942648,
+      "loss": 0.0427,
+      "step": 663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002699816961561928,
+      "loss": 0.0579,
+      "step": 664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00026979865771812076,
+      "loss": 0.0904,
+      "step": 665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002696156192800488,
+      "loss": 0.1317,
+      "step": 666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002694325808419768,
+      "loss": 0.1728,
+      "step": 667
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002692495424039048,
+      "loss": 0.0858,
+      "step": 668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002690665039658328,
+      "loss": 0.1369,
+      "step": 669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002688834655277608,
+      "loss": 0.1039,
+      "step": 670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002687004270896888,
+      "loss": 0.0744,
+      "step": 671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002685173886516168,
+      "loss": 0.0424,
+      "step": 672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002683343502135448,
+      "loss": 0.113,
+      "step": 673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00026815131177547284,
+      "loss": 0.0583,
+      "step": 674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00026796827333740083,
+      "loss": 0.1103,
+      "step": 675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00026778523489932887,
+      "loss": 0.0541,
+      "step": 676
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00026760219646125685,
+      "loss": 0.0909,
+      "step": 677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00026741915802318484,
+      "loss": 0.0476,
+      "step": 678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002672361195851128,
+      "loss": 0.0458,
+      "step": 679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00026705308114704086,
+      "loss": 0.0756,
+      "step": 680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00026687004270896885,
+      "loss": 0.1846,
+      "step": 681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002666870042708969,
+      "loss": 0.0355,
+      "step": 682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00026650396583282487,
+      "loss": 0.0868,
+      "step": 683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00026632092739475285,
+      "loss": 0.048,
+      "step": 684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002661378889566809,
+      "loss": 0.0486,
+      "step": 685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002659548505186089,
+      "loss": 0.1232,
+      "step": 686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002657718120805369,
+      "loss": 0.1694,
+      "step": 687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002655887736424649,
+      "loss": 0.0859,
+      "step": 688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002654057352043929,
+      "loss": 0.0517,
+      "step": 689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00026522269676632087,
+      "loss": 0.1551,
+      "step": 690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002650396583282489,
+      "loss": 0.1003,
+      "step": 691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002648566198901769,
+      "loss": 0.1365,
+      "step": 692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00026467358145210493,
+      "loss": 0.0579,
+      "step": 693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002644905430140329,
+      "loss": 0.0305,
+      "step": 694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00026430750457596096,
+      "loss": 0.069,
+      "step": 695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00026412446613788894,
+      "loss": 0.0295,
+      "step": 696
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00026394142769981693,
+      "loss": 0.4127,
+      "step": 697
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002637583892617449,
+      "loss": 0.073,
+      "step": 698
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00026357535082367295,
+      "loss": 0.2176,
+      "step": 699
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00026339231238560094,
+      "loss": 0.1491,
+      "step": 700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000263209273947529,
+      "loss": 0.1073,
+      "step": 701
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00026302623550945696,
+      "loss": 0.0754,
+      "step": 702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00026284319707138494,
+      "loss": 0.0485,
+      "step": 703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000262660158633313,
+      "loss": 0.1084,
+      "step": 704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00026247712019524097,
+      "loss": 0.1025,
+      "step": 705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000262294081757169,
+      "loss": 0.0475,
+      "step": 706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000262111043319097,
+      "loss": 0.1344,
+      "step": 707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000261928004881025,
+      "loss": 0.0816,
+      "step": 708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00026174496644295296,
+      "loss": 0.0867,
+      "step": 709
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000261561928004881,
+      "loss": 0.058,
+      "step": 710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000261378889566809,
+      "loss": 0.0387,
+      "step": 711
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000261195851128737,
+      "loss": 0.092,
+      "step": 712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000261012812690665,
+      "loss": 0.0348,
+      "step": 713
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00026082977425259305,
+      "loss": 0.0705,
+      "step": 714
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00026064673581452103,
+      "loss": 0.0785,
+      "step": 715
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000260463697376449,
+      "loss": 0.0538,
+      "step": 716
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00026028065893837706,
+      "loss": 0.0484,
+      "step": 717
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00026009762050030504,
+      "loss": 0.0551,
+      "step": 718
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000259914582062233,
+      "loss": 0.064,
+      "step": 719
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00025973154362416106,
+      "loss": 0.0504,
+      "step": 720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00025954850518608905,
+      "loss": 0.0479,
+      "step": 721
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00025936546674801703,
+      "loss": 0.1121,
+      "step": 722
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002591824283099451,
+      "loss": 0.0537,
+      "step": 723
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00025899938987187306,
+      "loss": 0.0654,
+      "step": 724
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002588163514338011,
+      "loss": 0.0844,
+      "step": 725
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002586333129957291,
+      "loss": 0.1242,
+      "step": 726
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002584502745576571,
+      "loss": 0.0653,
+      "step": 727
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00025826723611958505,
+      "loss": 0.0722,
+      "step": 728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002580841976815131,
+      "loss": 0.0833,
+      "step": 729
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002579011592434411,
+      "loss": 0.0498,
+      "step": 730
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002577181208053691,
+      "loss": 0.034,
+      "step": 731
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002575350823672971,
+      "loss": 0.0546,
+      "step": 732
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00025735204392922514,
+      "loss": 0.0344,
+      "step": 733
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002571690054911531,
+      "loss": 0.0698,
+      "step": 734
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002569859670530811,
+      "loss": 0.058,
+      "step": 735
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00025680292861500915,
+      "loss": 0.0629,
+      "step": 736
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00025661989017693713,
+      "loss": 0.1012,
+      "step": 737
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002564368517388651,
+      "loss": 0.0977,
+      "step": 738
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00025625381330079315,
+      "loss": 0.0705,
+      "step": 739
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00025607077486272114,
+      "loss": 0.0563,
+      "step": 740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002558877364246491,
+      "loss": 0.068,
+      "step": 741
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00025570469798657716,
+      "loss": 0.0361,
+      "step": 742
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00025552165954850515,
+      "loss": 0.0524,
+      "step": 743
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002553386211104332,
+      "loss": 0.0625,
+      "step": 744
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00025515558267236117,
+      "loss": 0.1073,
+      "step": 745
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002549725442342892,
+      "loss": 0.1454,
+      "step": 746
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002547895057962172,
+      "loss": 0.0658,
+      "step": 747
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002546064673581452,
+      "loss": 0.0487,
+      "step": 748
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00025442342892007316,
+      "loss": 0.035,
+      "step": 749
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002542403904820012,
+      "loss": 0.066,
+      "step": 750
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002540573520439292,
+      "loss": 0.0507,
+      "step": 751
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00025387431360585723,
+      "loss": 0.0659,
+      "step": 752
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002536912751677852,
+      "loss": 0.0417,
+      "step": 753
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002535082367297132,
+      "loss": 0.0798,
+      "step": 754
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00025332519829164124,
+      "loss": 0.0315,
+      "step": 755
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002531421598535692,
+      "loss": 0.0215,
+      "step": 756
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00025295912141549726,
+      "loss": 0.0457,
+      "step": 757
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00025277608297742524,
+      "loss": 0.1791,
+      "step": 758
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00025259304453935323,
+      "loss": 0.0186,
+      "step": 759
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002524100061012812,
+      "loss": 0.0654,
+      "step": 760
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00025222696766320925,
+      "loss": 0.0205,
+      "step": 761
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00025204392922513724,
+      "loss": 0.1292,
+      "step": 762
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002518608907870653,
+      "loss": 0.2472,
+      "step": 763
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00025167785234899326,
+      "loss": 0.1329,
+      "step": 764
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002514948139109213,
+      "loss": 0.04,
+      "step": 765
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002513117754728493,
+      "loss": 0.1389,
+      "step": 766
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00025112873703477727,
+      "loss": 0.0638,
+      "step": 767
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002509456985967053,
+      "loss": 0.0579,
+      "step": 768
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002507626601586333,
+      "loss": 0.1253,
+      "step": 769
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002505796217205613,
+      "loss": 0.0782,
+      "step": 770
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002503965832824893,
+      "loss": 0.0805,
+      "step": 771
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002502135448444173,
+      "loss": 0.0981,
+      "step": 772
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002500305064063453,
+      "loss": 0.0855,
+      "step": 773
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002498474679682733,
+      "loss": 0.0478,
+      "step": 774
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002496644295302013,
+      "loss": 0.0537,
+      "step": 775
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00024948139109212935,
+      "loss": 0.0612,
+      "step": 776
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00024929835265405733,
+      "loss": 0.0416,
+      "step": 777
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002491153142159854,
+      "loss": 0.0531,
+      "step": 778
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002489322757779133,
+      "loss": 0.0547,
+      "step": 779
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00024874923733984134,
+      "loss": 0.0271,
+      "step": 780
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00024856619890176933,
+      "loss": 0.0308,
+      "step": 781
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00024838316046369737,
+      "loss": 0.0743,
+      "step": 782
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00024820012202562535,
+      "loss": 0.0776,
+      "step": 783
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002480170835875534,
+      "loss": 0.0294,
+      "step": 784
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002478340451494814,
+      "loss": 0.0777,
+      "step": 785
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00024765100671140936,
+      "loss": 0.0479,
+      "step": 786
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002474679682733374,
+      "loss": 0.0267,
+      "step": 787
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002472849298352654,
+      "loss": 0.0859,
+      "step": 788
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00024710189139719337,
+      "loss": 0.0907,
+      "step": 789
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002469188529591214,
+      "loss": 0.0226,
+      "step": 790
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002467358145210494,
+      "loss": 0.0458,
+      "step": 791
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002465527760829774,
+      "loss": 0.0656,
+      "step": 792
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002463697376449054,
+      "loss": 0.0422,
+      "step": 793
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002461866992068334,
+      "loss": 0.0601,
+      "step": 794
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00024600366076876144,
+      "loss": 0.0302,
+      "step": 795
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002458206223306894,
+      "loss": 0.0295,
+      "step": 796
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00024563758389261746,
+      "loss": 0.0627,
+      "step": 797
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00024545454545454545,
+      "loss": 0.0581,
+      "step": 798
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00024527150701647343,
+      "loss": 0.2136,
+      "step": 799
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002450884685784014,
+      "loss": 0.0761,
+      "step": 800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00024490543014032946,
+      "loss": 0.0786,
+      "step": 801
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00024472239170225744,
+      "loss": 0.0677,
+      "step": 802
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002445393532641855,
+      "loss": 0.0258,
+      "step": 803
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00024435631482611346,
+      "loss": 0.0514,
+      "step": 804
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00024417327638804145,
+      "loss": 0.0736,
+      "step": 805
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002439902379499695,
+      "loss": 0.0423,
+      "step": 806
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002438071995118975,
+      "loss": 0.2245,
+      "step": 807
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00024362416107382549,
+      "loss": 0.185,
+      "step": 808
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00024344112263575347,
+      "loss": 0.0528,
+      "step": 809
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00024325808419768148,
+      "loss": 0.1677,
+      "step": 810
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002430750457596095,
+      "loss": 0.0704,
+      "step": 811
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002428920073215375,
+      "loss": 0.0539,
+      "step": 812
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00024270896888346552,
+      "loss": 0.115,
+      "step": 813
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002425259304453935,
+      "loss": 0.0527,
+      "step": 814
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00024234289200732151,
+      "loss": 0.0501,
+      "step": 815
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00024215985356924953,
+      "loss": 0.1152,
+      "step": 816
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00024197681513117754,
+      "loss": 0.0817,
+      "step": 817
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00024179377669310555,
+      "loss": 0.035,
+      "step": 818
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002416107382550335,
+      "loss": 0.0662,
+      "step": 819
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00024142769981696152,
+      "loss": 0.0972,
+      "step": 820
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00024124466137888953,
+      "loss": 0.1003,
+      "step": 821
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00024106162294081754,
+      "loss": 0.0668,
+      "step": 822
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00024087858450274555,
+      "loss": 0.1209,
+      "step": 823
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00024069554606467357,
+      "loss": 0.0907,
+      "step": 824
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00024051250762660158,
+      "loss": 0.0576,
+      "step": 825
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002403294691885296,
+      "loss": 0.0483,
+      "step": 826
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00024014643075045757,
+      "loss": 0.0597,
+      "step": 827
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002399633923123856,
+      "loss": 0.0581,
+      "step": 828
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00023978035387431357,
+      "loss": 0.0512,
+      "step": 829
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00023959731543624158,
+      "loss": 0.1029,
+      "step": 830
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002394142769981696,
+      "loss": 0.0738,
+      "step": 831
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00023923123856009758,
+      "loss": 0.1748,
+      "step": 832
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002390482001220256,
+      "loss": 0.0705,
+      "step": 833
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002388651616839536,
+      "loss": 0.0776,
+      "step": 834
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00023868212324588162,
+      "loss": 0.0848,
+      "step": 835
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00023849908480780963,
+      "loss": 0.068,
+      "step": 836
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00023831604636973764,
+      "loss": 0.0417,
+      "step": 837
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00023813300793166565,
+      "loss": 0.1186,
+      "step": 838
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002379499694935936,
+      "loss": 0.0866,
+      "step": 839
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00023776693105552162,
+      "loss": 0.0435,
+      "step": 840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00023758389261744963,
+      "loss": 0.0464,
+      "step": 841
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00023740085417937764,
+      "loss": 0.0683,
+      "step": 842
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00023721781574130566,
+      "loss": 0.1008,
+      "step": 843
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00023703477730323367,
+      "loss": 0.0169,
+      "step": 844
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00023685173886516168,
+      "loss": 0.0305,
+      "step": 845
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00023666870042708966,
+      "loss": 0.0957,
+      "step": 846
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00023648566198901768,
+      "loss": 0.0602,
+      "step": 847
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002363026235509457,
+      "loss": 0.0303,
+      "step": 848
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002361195851128737,
+      "loss": 0.042,
+      "step": 849
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00023593654667480169,
+      "loss": 0.0768,
+      "step": 850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00023575350823672967,
+      "loss": 0.0689,
+      "step": 851
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00023557046979865768,
+      "loss": 0.0112,
+      "step": 852
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002353874313605857,
+      "loss": 0.0486,
+      "step": 853
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002352043929225137,
+      "loss": 0.0246,
+      "step": 854
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00023502135448444172,
+      "loss": 0.1299,
+      "step": 855
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00023483831604636973,
+      "loss": 0.0446,
+      "step": 856
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00023465527760829774,
+      "loss": 0.053,
+      "step": 857
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00023447223917022575,
+      "loss": 0.064,
+      "step": 858
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00023428920073215374,
+      "loss": 0.0812,
+      "step": 859
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00023410616229408172,
+      "loss": 0.0726,
+      "step": 860
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00023392312385600973,
+      "loss": 0.0929,
+      "step": 861
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00023374008541793775,
+      "loss": 0.1138,
+      "step": 862
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00023355704697986576,
+      "loss": 0.1475,
+      "step": 863
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00023337400854179374,
+      "loss": 0.0502,
+      "step": 864
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00023319097010372175,
+      "loss": 0.1031,
+      "step": 865
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00023300793166564977,
+      "loss": 0.0865,
+      "step": 866
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00023282489322757778,
+      "loss": 0.0924,
+      "step": 867
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002326418547895058,
+      "loss": 0.092,
+      "step": 868
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002324588163514338,
+      "loss": 0.1272,
+      "step": 869
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00023227577791336176,
+      "loss": 0.0727,
+      "step": 870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00023209273947528977,
+      "loss": 0.0926,
+      "step": 871
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00023190970103721778,
+      "loss": 0.0541,
+      "step": 872
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002317266625991458,
+      "loss": 0.058,
+      "step": 873
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002315436241610738,
+      "loss": 0.1574,
+      "step": 874
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00023136058572300182,
+      "loss": 0.0796,
+      "step": 875
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00023117754728492983,
+      "loss": 0.0853,
+      "step": 876
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00023099450884685782,
+      "loss": 0.0478,
+      "step": 877
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00023081147040878583,
+      "loss": 0.0417,
+      "step": 878
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00023062843197071384,
+      "loss": 0.051,
+      "step": 879
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00023044539353264182,
+      "loss": 0.0456,
+      "step": 880
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00023026235509456984,
+      "loss": 0.0712,
+      "step": 881
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00023007931665649785,
+      "loss": 0.1145,
+      "step": 882
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00022989627821842583,
+      "loss": 0.0327,
+      "step": 883
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00022971323978035384,
+      "loss": 0.0612,
+      "step": 884
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00022953020134228186,
+      "loss": 0.0572,
+      "step": 885
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00022934716290420987,
+      "loss": 0.0959,
+      "step": 886
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00022916412446613788,
+      "loss": 0.0377,
+      "step": 887
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002289810860280659,
+      "loss": 0.0396,
+      "step": 888
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002287980475899939,
+      "loss": 0.0798,
+      "step": 889
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00022861500915192186,
+      "loss": 0.1248,
+      "step": 890
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00022843197071384987,
+      "loss": 0.0404,
+      "step": 891
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00022824893227577788,
+      "loss": 0.0244,
+      "step": 892
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002280658938377059,
+      "loss": 0.0492,
+      "step": 893
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002278828553996339,
+      "loss": 0.0621,
+      "step": 894
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00022769981696156192,
+      "loss": 0.0542,
+      "step": 895
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002275167785234899,
+      "loss": 0.092,
+      "step": 896
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00022733374008541792,
+      "loss": 0.0342,
+      "step": 897
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00022715070164734593,
+      "loss": 0.0734,
+      "step": 898
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00022696766320927394,
+      "loss": 0.0769,
+      "step": 899
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00022678462477120193,
+      "loss": 0.0396,
+      "step": 900
+    },
+    {
+      "epoch": 1.26,
+      "eval_loss": 0.10897424817085266,
+      "eval_runtime": 12985.513,
+      "eval_samples_per_second": 0.203,
+      "eval_steps_per_second": 0.102,
+      "eval_wer": 0.09347481158270528,
+      "step": 900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00022660158633312994,
+      "loss": 0.0403,
+      "step": 901
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00022641854789505792,
+      "loss": 0.051,
+      "step": 902
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00022623550945698593,
+      "loss": 0.0492,
+      "step": 903
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00022605247101891395,
+      "loss": 0.0916,
+      "step": 904
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00022586943258084196,
+      "loss": 0.029,
+      "step": 905
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00022568639414276997,
+      "loss": 0.1532,
+      "step": 906
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00022550335570469798,
+      "loss": 0.0493,
+      "step": 907
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000225320317266626,
+      "loss": 0.0586,
+      "step": 908
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00022513727882855398,
+      "loss": 0.0237,
+      "step": 909
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00022495424039048196,
+      "loss": 0.1168,
+      "step": 910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00022477120195240997,
+      "loss": 0.2013,
+      "step": 911
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000224588163514338,
+      "loss": 0.0923,
+      "step": 912
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000224405125076266,
+      "loss": 0.2901,
+      "step": 913
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000224222086638194,
+      "loss": 0.0616,
+      "step": 914
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000224039048200122,
+      "loss": 0.0337,
+      "step": 915
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00022385600976205,
+      "loss": 0.078,
+      "step": 916
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00022367297132397802,
+      "loss": 0.0674,
+      "step": 917
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00022348993288590603,
+      "loss": 0.1511,
+      "step": 918
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00022330689444783404,
+      "loss": 0.0892,
+      "step": 919
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00022312385600976203,
+      "loss": 0.0623,
+      "step": 920
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00022294081757169,
+      "loss": 0.0958,
+      "step": 921
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00022275777913361802,
+      "loss": 0.0467,
+      "step": 922
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00022257474069554604,
+      "loss": 0.119,
+      "step": 923
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00022239170225747405,
+      "loss": 0.0432,
+      "step": 924
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00022220866381940206,
+      "loss": 0.0619,
+      "step": 925
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00022202562538133007,
+      "loss": 0.027,
+      "step": 926
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00022184258694325808,
+      "loss": 0.0426,
+      "step": 927
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00022165954850518607,
+      "loss": 0.0582,
+      "step": 928
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00022147651006711408,
+      "loss": 0.0431,
+      "step": 929
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00022129347162904206,
+      "loss": 0.0544,
+      "step": 930
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00022111043319097008,
+      "loss": 0.0566,
+      "step": 931
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002209273947528981,
+      "loss": 0.075,
+      "step": 932
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002207443563148261,
+      "loss": 0.0554,
+      "step": 933
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00022056131787675408,
+      "loss": 0.0639,
+      "step": 934
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002203782794386821,
+      "loss": 0.0236,
+      "step": 935
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002201952410006101,
+      "loss": 0.0179,
+      "step": 936
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00022001220256253812,
+      "loss": 0.0825,
+      "step": 937
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00021982916412446613,
+      "loss": 0.0843,
+      "step": 938
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00021964612568639414,
+      "loss": 0.0412,
+      "step": 939
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00021946308724832216,
+      "loss": 0.0627,
+      "step": 940
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00021928004881025011,
+      "loss": 0.0558,
+      "step": 941
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00021909701037217813,
+      "loss": 0.0158,
+      "step": 942
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00021891397193410614,
+      "loss": 0.053,
+      "step": 943
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00021873093349603415,
+      "loss": 0.0568,
+      "step": 944
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00021854789505796216,
+      "loss": 0.0313,
+      "step": 945
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00021836485661989017,
+      "loss": 0.0332,
+      "step": 946
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00021818181818181816,
+      "loss": 0.1076,
+      "step": 947
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00021799877974374617,
+      "loss": 0.0713,
+      "step": 948
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00021781574130567418,
+      "loss": 0.0602,
+      "step": 949
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002176327028676022,
+      "loss": 0.0594,
+      "step": 950
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00021744966442953018,
+      "loss": 0.0603,
+      "step": 951
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00021726662599145816,
+      "loss": 0.0795,
+      "step": 952
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00021708358755338617,
+      "loss": 0.1373,
+      "step": 953
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002169005491153142,
+      "loss": 0.0683,
+      "step": 954
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002167175106772422,
+      "loss": 0.0465,
+      "step": 955
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002165344722391702,
+      "loss": 0.0952,
+      "step": 956
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00021635143380109822,
+      "loss": 0.0687,
+      "step": 957
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00021616839536302623,
+      "loss": 0.0879,
+      "step": 958
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00021598535692495425,
+      "loss": 0.0699,
+      "step": 959
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00021580231848688223,
+      "loss": 0.0513,
+      "step": 960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00021561928004881022,
+      "loss": 0.2458,
+      "step": 961
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00021543624161073823,
+      "loss": 0.2328,
+      "step": 962
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00021525320317266624,
+      "loss": 0.1161,
+      "step": 963
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00021507016473459425,
+      "loss": 0.0653,
+      "step": 964
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00021488712629652226,
+      "loss": 0.0247,
+      "step": 965
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00021470408785845025,
+      "loss": 0.0375,
+      "step": 966
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00021452104942037826,
+      "loss": 0.0723,
+      "step": 967
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00021433801098230627,
+      "loss": 0.0494,
+      "step": 968
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00021415497254423428,
+      "loss": 0.0825,
+      "step": 969
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002139719341061623,
+      "loss": 0.1024,
+      "step": 970
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00021378889566809025,
+      "loss": 0.0661,
+      "step": 971
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00021360585723001826,
+      "loss": 0.0361,
+      "step": 972
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00021342281879194628,
+      "loss": 0.0757,
+      "step": 973
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002132397803538743,
+      "loss": 0.0875,
+      "step": 974
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002130567419158023,
+      "loss": 0.0778,
+      "step": 975
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002128737034777303,
+      "loss": 0.0399,
+      "step": 976
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00021269066503965832,
+      "loss": 0.0476,
+      "step": 977
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00021250762660158634,
+      "loss": 0.0519,
+      "step": 978
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00021232458816351432,
+      "loss": 0.0136,
+      "step": 979
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00021214154972544233,
+      "loss": 0.0905,
+      "step": 980
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00021195851128737032,
+      "loss": 0.058,
+      "step": 981
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00021177547284929833,
+      "loss": 0.038,
+      "step": 982
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00021159243441122634,
+      "loss": 0.0604,
+      "step": 983
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00021140939597315433,
+      "loss": 0.0279,
+      "step": 984
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00021122635753508234,
+      "loss": 0.0682,
+      "step": 985
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00021104331909701035,
+      "loss": 0.0285,
+      "step": 986
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00021086028065893836,
+      "loss": 0.1053,
+      "step": 987
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00021067724222086637,
+      "loss": 0.05,
+      "step": 988
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00021049420378279438,
+      "loss": 0.0737,
+      "step": 989
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002103111653447224,
+      "loss": 0.0782,
+      "step": 990
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00021012812690665035,
+      "loss": 0.0232,
+      "step": 991
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00020994508846857837,
+      "loss": 0.0332,
+      "step": 992
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00020976205003050638,
+      "loss": 0.0173,
+      "step": 993
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002095790115924344,
+      "loss": 0.0304,
+      "step": 994
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002093959731543624,
+      "loss": 0.0527,
+      "step": 995
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00020921293471629041,
+      "loss": 0.0721,
+      "step": 996
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002090298962782184,
+      "loss": 0.0593,
+      "step": 997
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002088468578401464,
+      "loss": 0.029,
+      "step": 998
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00020866381940207442,
+      "loss": 0.1074,
+      "step": 999
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00020848078096400243,
+      "loss": 0.0614,
+      "step": 1000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00020829774252593042,
+      "loss": 0.0485,
+      "step": 1001
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00020811470408785843,
+      "loss": 0.0834,
+      "step": 1002
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00020793166564978642,
+      "loss": 0.0603,
+      "step": 1003
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00020774862721171443,
+      "loss": 0.024,
+      "step": 1004
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00020756558877364244,
+      "loss": 0.0185,
+      "step": 1005
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00020738255033557045,
+      "loss": 0.0899,
+      "step": 1006
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00020719951189749846,
+      "loss": 0.068,
+      "step": 1007
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00020701647345942647,
+      "loss": 0.0546,
+      "step": 1008
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002068334350213545,
+      "loss": 0.0403,
+      "step": 1009
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002066503965832825,
+      "loss": 0.0272,
+      "step": 1010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020646735814521046,
+      "loss": 0.0273,
+      "step": 1011
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020628431970713847,
+      "loss": 0.0284,
+      "step": 1012
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020610128126906648,
+      "loss": 0.0913,
+      "step": 1013
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002059182428309945,
+      "loss": 0.0668,
+      "step": 1014
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002057352043929225,
+      "loss": 0.0165,
+      "step": 1015
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002055521659548505,
+      "loss": 0.0786,
+      "step": 1016
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002053691275167785,
+      "loss": 0.048,
+      "step": 1017
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002051860890787065,
+      "loss": 0.0367,
+      "step": 1018
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020500305064063452,
+      "loss": 0.0931,
+      "step": 1019
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020482001220256254,
+      "loss": 0.0535,
+      "step": 1020
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020463697376449055,
+      "loss": 0.0594,
+      "step": 1021
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002044539353264185,
+      "loss": 0.027,
+      "step": 1022
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020427089688834652,
+      "loss": 0.0262,
+      "step": 1023
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00020408785845027453,
+      "loss": 0.0752,
+      "step": 1024
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00020390482001220254,
+      "loss": 0.0576,
+      "step": 1025
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00020372178157413055,
+      "loss": 0.0441,
+      "step": 1026
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00020353874313605856,
+      "loss": 0.0299,
+      "step": 1027
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00020335570469798658,
+      "loss": 0.0596,
+      "step": 1028
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00020317266625991456,
+      "loss": 0.0337,
+      "step": 1029
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00020298962782184257,
+      "loss": 0.0757,
+      "step": 1030
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00020280658938377058,
+      "loss": 0.0431,
+      "step": 1031
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00020262355094569857,
+      "loss": 0.0346,
+      "step": 1032
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00020244051250762658,
+      "loss": 0.1306,
+      "step": 1033
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002022574740695546,
+      "loss": 0.0635,
+      "step": 1034
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00020207443563148258,
+      "loss": 0.0306,
+      "step": 1035
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002018913971934106,
+      "loss": 0.053,
+      "step": 1036
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002017083587553386,
+      "loss": 0.0555,
+      "step": 1037
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00020152532031726661,
+      "loss": 0.025,
+      "step": 1038
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00020134228187919463,
+      "loss": 0.0791,
+      "step": 1039
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00020115924344112264,
+      "loss": 0.0398,
+      "step": 1040
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00020097620500305065,
+      "loss": 0.0202,
+      "step": 1041
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002007931665649786,
+      "loss": 0.0299,
+      "step": 1042
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00020061012812690662,
+      "loss": 0.0127,
+      "step": 1043
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00020042708968883463,
+      "loss": 0.0388,
+      "step": 1044
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00020024405125076264,
+      "loss": 0.0665,
+      "step": 1045
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00020006101281269065,
+      "loss": 0.0513,
+      "step": 1046
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00019987797437461867,
+      "loss": 0.0449,
+      "step": 1047
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00019969493593654665,
+      "loss": 0.0422,
+      "step": 1048
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00019951189749847466,
+      "loss": 0.0969,
+      "step": 1049
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00019932885906040267,
+      "loss": 0.0555,
+      "step": 1050
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00019914582062233069,
+      "loss": 0.0432,
+      "step": 1051
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00019896278218425867,
+      "loss": 0.0351,
+      "step": 1052
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00019877974374618668,
+      "loss": 0.0367,
+      "step": 1053
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00019859670530811467,
+      "loss": 0.0562,
+      "step": 1054
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00019841366687004268,
+      "loss": 0.0438,
+      "step": 1055
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001982306284319707,
+      "loss": 0.0824,
+      "step": 1056
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001980475899938987,
+      "loss": 0.1256,
+      "step": 1057
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00019786455155582672,
+      "loss": 0.0341,
+      "step": 1058
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00019768151311775473,
+      "loss": 0.0394,
+      "step": 1059
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00019749847467968274,
+      "loss": 0.2587,
+      "step": 1060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00019731543624161072,
+      "loss": 0.1328,
+      "step": 1061
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001971323978035387,
+      "loss": 0.0683,
+      "step": 1062
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00019694935936546672,
+      "loss": 0.1324,
+      "step": 1063
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00019676632092739473,
+      "loss": 0.0246,
+      "step": 1064
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00019658328248932274,
+      "loss": 0.0947,
+      "step": 1065
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00019640024405125076,
+      "loss": 0.0928,
+      "step": 1066
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00019621720561317874,
+      "loss": 0.0697,
+      "step": 1067
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00019603416717510675,
+      "loss": 0.0741,
+      "step": 1068
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00019585112873703476,
+      "loss": 0.0309,
+      "step": 1069
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00019566809029896278,
+      "loss": 0.0324,
+      "step": 1070
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001954850518608908,
+      "loss": 0.0524,
+      "step": 1071
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00019530201342281877,
+      "loss": 0.0426,
+      "step": 1072
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00019511897498474676,
+      "loss": 0.0517,
+      "step": 1073
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00019493593654667477,
+      "loss": 0.0497,
+      "step": 1074
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00019475289810860278,
+      "loss": 0.065,
+      "step": 1075
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001945698596705308,
+      "loss": 0.0339,
+      "step": 1076
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001943868212324588,
+      "loss": 0.0546,
+      "step": 1077
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00019420378279438682,
+      "loss": 0.0333,
+      "step": 1078
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00019402074435631483,
+      "loss": 0.075,
+      "step": 1079
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001938377059182428,
+      "loss": 0.2019,
+      "step": 1080
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00019365466748017083,
+      "loss": 0.0307,
+      "step": 1081
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001934716290420988,
+      "loss": 0.0365,
+      "step": 1082
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00019328859060402682,
+      "loss": 0.0578,
+      "step": 1083
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00019310555216595483,
+      "loss": 0.0408,
+      "step": 1084
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00019292251372788285,
+      "loss": 0.0696,
+      "step": 1085
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00019273947528981083,
+      "loss": 0.0509,
+      "step": 1086
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00019255643685173884,
+      "loss": 0.1542,
+      "step": 1087
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00019237339841366685,
+      "loss": 0.0405,
+      "step": 1088
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00019219035997559487,
+      "loss": 0.0142,
+      "step": 1089
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00019200732153752288,
+      "loss": 0.0709,
+      "step": 1090
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001918242830994509,
+      "loss": 0.0278,
+      "step": 1091
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00019164124466137885,
+      "loss": 0.0447,
+      "step": 1092
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00019145820622330686,
+      "loss": 0.1014,
+      "step": 1093
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00019127516778523487,
+      "loss": 0.0327,
+      "step": 1094
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00019109212934716288,
+      "loss": 0.028,
+      "step": 1095
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001909090909090909,
+      "loss": 0.0447,
+      "step": 1096
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001907260524710189,
+      "loss": 0.034,
+      "step": 1097
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00019054301403294692,
+      "loss": 0.1114,
+      "step": 1098
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001903599755948749,
+      "loss": 0.1447,
+      "step": 1099
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00019017693715680292,
+      "loss": 0.0542,
+      "step": 1100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00018999389871873093,
+      "loss": 0.0158,
+      "step": 1101
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001898108602806589,
+      "loss": 0.0195,
+      "step": 1102
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00018962782184258692,
+      "loss": 0.0792,
+      "step": 1103
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001894447834045149,
+      "loss": 0.0703,
+      "step": 1104
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00018926174496644292,
+      "loss": 0.0115,
+      "step": 1105
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00018907870652837093,
+      "loss": 0.0496,
+      "step": 1106
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00018889566809029894,
+      "loss": 0.0322,
+      "step": 1107
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00018871262965222696,
+      "loss": 0.0951,
+      "step": 1108
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00018852959121415497,
+      "loss": 0.0318,
+      "step": 1109
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00018834655277608298,
+      "loss": 0.0094,
+      "step": 1110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000188163514338011,
+      "loss": 0.0892,
+      "step": 1111
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00018798047589993898,
+      "loss": 0.2906,
+      "step": 1112
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00018779743746186696,
+      "loss": 0.1275,
+      "step": 1113
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00018761439902379497,
+      "loss": 0.0988,
+      "step": 1114
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00018743136058572298,
+      "loss": 0.0585,
+      "step": 1115
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000187248322147651,
+      "loss": 0.0748,
+      "step": 1116
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00018706528370957898,
+      "loss": 0.053,
+      "step": 1117
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000186882245271507,
+      "loss": 0.0266,
+      "step": 1118
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000186699206833435,
+      "loss": 0.1014,
+      "step": 1119
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00018651616839536302,
+      "loss": 0.0348,
+      "step": 1120
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00018633312995729103,
+      "loss": 0.0303,
+      "step": 1121
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00018615009151921904,
+      "loss": 0.0422,
+      "step": 1122
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000185967053081147,
+      "loss": 0.0252,
+      "step": 1123
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000185784014643075,
+      "loss": 0.0605,
+      "step": 1124
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00018560097620500302,
+      "loss": 0.0751,
+      "step": 1125
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00018541793776693103,
+      "loss": 0.0555,
+      "step": 1126
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00018523489932885905,
+      "loss": 0.0619,
+      "step": 1127
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00018505186089078706,
+      "loss": 0.0845,
+      "step": 1128
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00018486882245271507,
+      "loss": 0.0384,
+      "step": 1129
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00018468578401464308,
+      "loss": 0.0624,
+      "step": 1130
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018450274557657107,
+      "loss": 0.0967,
+      "step": 1131
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018431970713849908,
+      "loss": 0.0608,
+      "step": 1132
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018413666870042706,
+      "loss": 0.0211,
+      "step": 1133
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018395363026235507,
+      "loss": 0.0781,
+      "step": 1134
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018377059182428309,
+      "loss": 0.1161,
+      "step": 1135
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018358755338621107,
+      "loss": 0.046,
+      "step": 1136
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018340451494813908,
+      "loss": 0.0836,
+      "step": 1137
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001832214765100671,
+      "loss": 0.0719,
+      "step": 1138
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001830384380719951,
+      "loss": 0.0295,
+      "step": 1139
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018285539963392312,
+      "loss": 0.0465,
+      "step": 1140
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018267236119585113,
+      "loss": 0.149,
+      "step": 1141
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018248932275777914,
+      "loss": 0.0538,
+      "step": 1142
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001823062843197071,
+      "loss": 0.0343,
+      "step": 1143
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001821232458816351,
+      "loss": 0.0796,
+      "step": 1144
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018194020744356312,
+      "loss": 0.0519,
+      "step": 1145
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018175716900549114,
+      "loss": 0.0277,
+      "step": 1146
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018157413056741915,
+      "loss": 0.052,
+      "step": 1147
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018139109212934716,
+      "loss": 0.0205,
+      "step": 1148
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018120805369127514,
+      "loss": 0.0443,
+      "step": 1149
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018102501525320316,
+      "loss": 0.0336,
+      "step": 1150
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018084197681513117,
+      "loss": 0.0229,
+      "step": 1151
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018065893837705918,
+      "loss": 0.0513,
+      "step": 1152
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018047589993898716,
+      "loss": 0.0426,
+      "step": 1153
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018029286150091518,
+      "loss": 0.0622,
+      "step": 1154
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018010982306284316,
+      "loss": 0.0311,
+      "step": 1155
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017992678462477117,
+      "loss": 0.0387,
+      "step": 1156
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017974374618669918,
+      "loss": 0.0411,
+      "step": 1157
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001795607077486272,
+      "loss": 0.0097,
+      "step": 1158
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001793776693105552,
+      "loss": 0.1426,
+      "step": 1159
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00017919463087248322,
+      "loss": 0.0883,
+      "step": 1160
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00017901159243441123,
+      "loss": 0.0541,
+      "step": 1161
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00017882855399633922,
+      "loss": 0.0744,
+      "step": 1162
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001786455155582672,
+      "loss": 0.1138,
+      "step": 1163
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001784624771201952,
+      "loss": 0.0383,
+      "step": 1164
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00017827943868212323,
+      "loss": 0.0227,
+      "step": 1165
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00017809640024405124,
+      "loss": 0.0428,
+      "step": 1166
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017791336180597925,
+      "loss": 0.0211,
+      "step": 1167
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017773032336790723,
+      "loss": 0.0642,
+      "step": 1168
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017754728492983525,
+      "loss": 0.0299,
+      "step": 1169
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017736424649176326,
+      "loss": 0.0239,
+      "step": 1170
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017718120805369127,
+      "loss": 0.0466,
+      "step": 1171
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017699816961561928,
+      "loss": 0.0364,
+      "step": 1172
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017681513117754727,
+      "loss": 0.0409,
+      "step": 1173
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017663209273947525,
+      "loss": 0.0648,
+      "step": 1174
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017644905430140326,
+      "loss": 0.0442,
+      "step": 1175
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017626601586333127,
+      "loss": 0.0362,
+      "step": 1176
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017608297742525929,
+      "loss": 0.0455,
+      "step": 1177
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001758999389871873,
+      "loss": 0.0607,
+      "step": 1178
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001757169005491153,
+      "loss": 0.0137,
+      "step": 1179
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017553386211104332,
+      "loss": 0.0238,
+      "step": 1180
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001753508236729713,
+      "loss": 0.0423,
+      "step": 1181
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017516778523489932,
+      "loss": 0.0224,
+      "step": 1182
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001749847467968273,
+      "loss": 0.0302,
+      "step": 1183
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017480170835875532,
+      "loss": 0.0323,
+      "step": 1184
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017461866992068333,
+      "loss": 0.0582,
+      "step": 1185
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017443563148261134,
+      "loss": 0.0757,
+      "step": 1186
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017425259304453932,
+      "loss": 0.039,
+      "step": 1187
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017406955460646734,
+      "loss": 0.0394,
+      "step": 1188
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017388651616839535,
+      "loss": 0.0063,
+      "step": 1189
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017370347773032336,
+      "loss": 0.0478,
+      "step": 1190
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017352043929225137,
+      "loss": 0.0235,
+      "step": 1191
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017333740085417938,
+      "loss": 0.0228,
+      "step": 1192
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001731543624161074,
+      "loss": 0.0184,
+      "step": 1193
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017297132397803535,
+      "loss": 0.098,
+      "step": 1194
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017278828553996336,
+      "loss": 0.0261,
+      "step": 1195
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017260524710189138,
+      "loss": 0.0423,
+      "step": 1196
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001724222086638194,
+      "loss": 0.0165,
+      "step": 1197
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001722391702257474,
+      "loss": 0.0351,
+      "step": 1198
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001720561317876754,
+      "loss": 0.0459,
+      "step": 1199
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001718730933496034,
+      "loss": 0.0353,
+      "step": 1200
+    },
+    {
+      "epoch": 1.68,
+      "eval_loss": 0.10321791470050812,
+      "eval_runtime": 12622.7025,
+      "eval_samples_per_second": 0.209,
+      "eval_steps_per_second": 0.105,
+      "eval_wer": 0.08322094406981356,
+      "step": 1200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001716900549115314,
+      "loss": 0.1118,
+      "step": 1201
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00017150701647345942,
+      "loss": 0.0531,
+      "step": 1202
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00017132397803538743,
+      "loss": 0.048,
+      "step": 1203
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00017114093959731542,
+      "loss": 0.0386,
+      "step": 1204
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00017095790115924343,
+      "loss": 0.0488,
+      "step": 1205
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001707748627211714,
+      "loss": 0.0367,
+      "step": 1206
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00017059182428309943,
+      "loss": 0.0378,
+      "step": 1207
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00017040878584502744,
+      "loss": 0.0345,
+      "step": 1208
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00017022574740695545,
+      "loss": 0.0132,
+      "step": 1209
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00017004270896888346,
+      "loss": 0.0094,
+      "step": 1210
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016985967053081147,
+      "loss": 0.068,
+      "step": 1211
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016967663209273948,
+      "loss": 0.119,
+      "step": 1212
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016949359365466747,
+      "loss": 0.0761,
+      "step": 1213
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016931055521659545,
+      "loss": 0.0401,
+      "step": 1214
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016912751677852347,
+      "loss": 0.103,
+      "step": 1215
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016894447834045148,
+      "loss": 0.0777,
+      "step": 1216
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001687614399023795,
+      "loss": 0.044,
+      "step": 1217
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001685784014643075,
+      "loss": 0.0276,
+      "step": 1218
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016839536302623549,
+      "loss": 0.1067,
+      "step": 1219
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001682123245881635,
+      "loss": 0.0319,
+      "step": 1220
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001680292861500915,
+      "loss": 0.0337,
+      "step": 1221
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016784624771201952,
+      "loss": 0.0316,
+      "step": 1222
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016766320927394753,
+      "loss": 0.0665,
+      "step": 1223
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001674801708358755,
+      "loss": 0.1549,
+      "step": 1224
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001672971323978035,
+      "loss": 0.059,
+      "step": 1225
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016711409395973151,
+      "loss": 0.0471,
+      "step": 1226
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016693105552165953,
+      "loss": 0.1383,
+      "step": 1227
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016674801708358754,
+      "loss": 0.0211,
+      "step": 1228
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016656497864551555,
+      "loss": 0.0292,
+      "step": 1229
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016638194020744356,
+      "loss": 0.062,
+      "step": 1230
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016619890176937157,
+      "loss": 0.0553,
+      "step": 1231
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016601586333129956,
+      "loss": 0.0423,
+      "step": 1232
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016583282489322757,
+      "loss": 0.1004,
+      "step": 1233
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016564978645515556,
+      "loss": 0.045,
+      "step": 1234
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016546674801708357,
+      "loss": 0.087,
+      "step": 1235
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016528370957901158,
+      "loss": 0.0588,
+      "step": 1236
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001651006711409396,
+      "loss": 0.084,
+      "step": 1237
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016491763270286758,
+      "loss": 0.0793,
+      "step": 1238
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001647345942647956,
+      "loss": 0.0517,
+      "step": 1239
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001645515558267236,
+      "loss": 0.0241,
+      "step": 1240
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001643685173886516,
+      "loss": 0.0434,
+      "step": 1241
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016418547895057962,
+      "loss": 0.0194,
+      "step": 1242
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016400244051250764,
+      "loss": 0.0358,
+      "step": 1243
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001638194020744356,
+      "loss": 0.0526,
+      "step": 1244
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001636363636363636,
+      "loss": 0.0232,
+      "step": 1245
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016345332519829162,
+      "loss": 0.0454,
+      "step": 1246
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016327028676021963,
+      "loss": 0.0409,
+      "step": 1247
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016308724832214764,
+      "loss": 0.0205,
+      "step": 1248
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016290420988407565,
+      "loss": 0.0341,
+      "step": 1249
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016272117144600366,
+      "loss": 0.0604,
+      "step": 1250
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016253813300793165,
+      "loss": 0.0263,
+      "step": 1251
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016235509456985966,
+      "loss": 0.0166,
+      "step": 1252
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016217205613178767,
+      "loss": 0.1201,
+      "step": 1253
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016198901769371566,
+      "loss": 0.0392,
+      "step": 1254
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016180597925564367,
+      "loss": 0.0168,
+      "step": 1255
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016162294081757165,
+      "loss": 0.0432,
+      "step": 1256
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016143990237949967,
+      "loss": 0.1493,
+      "step": 1257
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016125686394142768,
+      "loss": 0.0558,
+      "step": 1258
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001610738255033557,
+      "loss": 0.0204,
+      "step": 1259
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001608907870652837,
+      "loss": 0.1088,
+      "step": 1260
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001607077486272117,
+      "loss": 0.0168,
+      "step": 1261
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016052471018913973,
+      "loss": 0.0927,
+      "step": 1262
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016034167175106774,
+      "loss": 0.0807,
+      "step": 1263
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001601586333129957,
+      "loss": 0.057,
+      "step": 1264
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001599755948749237,
+      "loss": 0.0348,
+      "step": 1265
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015979255643685172,
+      "loss": 0.0204,
+      "step": 1266
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015960951799877973,
+      "loss": 0.0444,
+      "step": 1267
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015942647956070774,
+      "loss": 0.0234,
+      "step": 1268
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015924344112263573,
+      "loss": 0.0524,
+      "step": 1269
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015906040268456374,
+      "loss": 0.0515,
+      "step": 1270
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015887736424649175,
+      "loss": 0.0255,
+      "step": 1271
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015869432580841976,
+      "loss": 0.0686,
+      "step": 1272
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015851128737034777,
+      "loss": 0.0688,
+      "step": 1273
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00015832824893227576,
+      "loss": 0.0628,
+      "step": 1274
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00015814521049420374,
+      "loss": 0.0483,
+      "step": 1275
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00015796217205613176,
+      "loss": 0.0448,
+      "step": 1276
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00015777913361805977,
+      "loss": 0.0802,
+      "step": 1277
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00015759609517998778,
+      "loss": 0.0536,
+      "step": 1278
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001574130567419158,
+      "loss": 0.0352,
+      "step": 1279
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001572300183038438,
+      "loss": 0.0661,
+      "step": 1280
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00015704697986577181,
+      "loss": 0.0615,
+      "step": 1281
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001568639414276998,
+      "loss": 0.0152,
+      "step": 1282
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001566809029896278,
+      "loss": 0.0654,
+      "step": 1283
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00015649786455155582,
+      "loss": 0.0377,
+      "step": 1284
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001563148261134838,
+      "loss": 0.0509,
+      "step": 1285
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00015613178767541182,
+      "loss": 0.0212,
+      "step": 1286
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00015594874923733983,
+      "loss": 0.116,
+      "step": 1287
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00015576571079926782,
+      "loss": 0.0129,
+      "step": 1288
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00015558267236119583,
+      "loss": 0.0614,
+      "step": 1289
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00015539963392312384,
+      "loss": 0.0267,
+      "step": 1290
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00015521659548505185,
+      "loss": 0.0974,
+      "step": 1291
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00015503355704697986,
+      "loss": 0.0262,
+      "step": 1292
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00015485051860890788,
+      "loss": 0.0207,
+      "step": 1293
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001546674801708359,
+      "loss": 0.0167,
+      "step": 1294
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00015448444173276385,
+      "loss": 0.0504,
+      "step": 1295
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00015430140329469186,
+      "loss": 0.0331,
+      "step": 1296
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00015411836485661987,
+      "loss": 0.022,
+      "step": 1297
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00015393532641854788,
+      "loss": 0.0454,
+      "step": 1298
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001537522879804759,
+      "loss": 0.0606,
+      "step": 1299
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001535692495424039,
+      "loss": 0.0158,
+      "step": 1300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001533862111043319,
+      "loss": 0.1187,
+      "step": 1301
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001532031726662599,
+      "loss": 0.0524,
+      "step": 1302
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001530201342281879,
+      "loss": 0.0188,
+      "step": 1303
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00015283709579011593,
+      "loss": 0.0311,
+      "step": 1304
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001526540573520439,
+      "loss": 0.0395,
+      "step": 1305
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00015247101891397192,
+      "loss": 0.0192,
+      "step": 1306
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001522879804758999,
+      "loss": 0.0467,
+      "step": 1307
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00015210494203782792,
+      "loss": 0.0695,
+      "step": 1308
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00015192190359975593,
+      "loss": 0.0267,
+      "step": 1309
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00015173886516168394,
+      "loss": 0.0501,
+      "step": 1310
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00015155582672361195,
+      "loss": 0.0364,
+      "step": 1311
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00015137278828553997,
+      "loss": 0.0297,
+      "step": 1312
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00015118974984746798,
+      "loss": 0.0171,
+      "step": 1313
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00015100671140939596,
+      "loss": 0.0586,
+      "step": 1314
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00015082367297132395,
+      "loss": 0.05,
+      "step": 1315
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00015064063453325196,
+      "loss": 0.031,
+      "step": 1316
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00015045759609517997,
+      "loss": 0.0312,
+      "step": 1317
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00015027455765710798,
+      "loss": 0.0698,
+      "step": 1318
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000150091519219036,
+      "loss": 0.0667,
+      "step": 1319
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014990848078096398,
+      "loss": 0.0441,
+      "step": 1320
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000149725442342892,
+      "loss": 0.0492,
+      "step": 1321
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014954240390482,
+      "loss": 0.0284,
+      "step": 1322
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000149359365466748,
+      "loss": 0.0274,
+      "step": 1323
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.000149176327028676,
+      "loss": 0.0763,
+      "step": 1324
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.000148993288590604,
+      "loss": 0.0396,
+      "step": 1325
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014881025015253202,
+      "loss": 0.0248,
+      "step": 1326
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014862721171446004,
+      "loss": 0.013,
+      "step": 1327
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014844417327638802,
+      "loss": 0.0253,
+      "step": 1328
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014826113483831603,
+      "loss": 0.0222,
+      "step": 1329
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014807809640024404,
+      "loss": 0.0379,
+      "step": 1330
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014789505796217206,
+      "loss": 0.0062,
+      "step": 1331
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014771201952410007,
+      "loss": 0.0175,
+      "step": 1332
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014752898108602805,
+      "loss": 0.0248,
+      "step": 1333
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014734594264795606,
+      "loss": 0.0289,
+      "step": 1334
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014716290420988408,
+      "loss": 0.0192,
+      "step": 1335
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014697986577181206,
+      "loss": 0.0182,
+      "step": 1336
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014679682733374007,
+      "loss": 0.0823,
+      "step": 1337
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014661378889566808,
+      "loss": 0.0476,
+      "step": 1338
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014643075045759607,
+      "loss": 0.0721,
+      "step": 1339
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014624771201952408,
+      "loss": 0.0625,
+      "step": 1340
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001460646735814521,
+      "loss": 0.0553,
+      "step": 1341
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001458816351433801,
+      "loss": 0.026,
+      "step": 1342
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001456985967053081,
+      "loss": 0.0616,
+      "step": 1343
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001455155582672361,
+      "loss": 0.0219,
+      "step": 1344
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001453325198291641,
+      "loss": 0.0298,
+      "step": 1345
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014514948139109212,
+      "loss": 0.0343,
+      "step": 1346
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014496644295302014,
+      "loss": 0.0225,
+      "step": 1347
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014478340451494812,
+      "loss": 0.0397,
+      "step": 1348
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014460036607687613,
+      "loss": 0.0617,
+      "step": 1349
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014441732763880415,
+      "loss": 0.0991,
+      "step": 1350
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014423428920073216,
+      "loss": 0.0359,
+      "step": 1351
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014405125076266014,
+      "loss": 0.0335,
+      "step": 1352
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014386821232458815,
+      "loss": 0.1212,
+      "step": 1353
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014368517388651617,
+      "loss": 0.0403,
+      "step": 1354
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014350213544844415,
+      "loss": 0.0158,
+      "step": 1355
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014331909701037216,
+      "loss": 0.0166,
+      "step": 1356
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014313605857230017,
+      "loss": 0.0359,
+      "step": 1357
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014295302013422816,
+      "loss": 0.0811,
+      "step": 1358
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014276998169615617,
+      "loss": 0.0188,
+      "step": 1359
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014258694325808418,
+      "loss": 0.036,
+      "step": 1360
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001424039048200122,
+      "loss": 0.064,
+      "step": 1361
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001422208663819402,
+      "loss": 0.0346,
+      "step": 1362
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001420378279438682,
+      "loss": 0.1841,
+      "step": 1363
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001418547895057962,
+      "loss": 0.0592,
+      "step": 1364
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014167175106772421,
+      "loss": 0.0743,
+      "step": 1365
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014148871262965223,
+      "loss": 0.0202,
+      "step": 1366
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014130567419158024,
+      "loss": 0.0346,
+      "step": 1367
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014112263575350822,
+      "loss": 0.0568,
+      "step": 1368
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014093959731543624,
+      "loss": 0.0512,
+      "step": 1369
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014075655887736425,
+      "loss": 0.0318,
+      "step": 1370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014057352043929223,
+      "loss": 0.0383,
+      "step": 1371
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014039048200122024,
+      "loss": 0.022,
+      "step": 1372
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014020744356314823,
+      "loss": 0.0693,
+      "step": 1373
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014002440512507624,
+      "loss": 0.0359,
+      "step": 1374
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00013984136668700425,
+      "loss": 0.0359,
+      "step": 1375
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00013965832824893226,
+      "loss": 0.024,
+      "step": 1376
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00013947528981086028,
+      "loss": 0.0929,
+      "step": 1377
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00013929225137278826,
+      "loss": 0.0398,
+      "step": 1378
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00013910921293471627,
+      "loss": 0.0226,
+      "step": 1379
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00013892617449664428,
+      "loss": 0.0385,
+      "step": 1380
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001387431360585723,
+      "loss": 0.0196,
+      "step": 1381
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001385600976205003,
+      "loss": 0.0347,
+      "step": 1382
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001383770591824283,
+      "loss": 0.0541,
+      "step": 1383
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001381940207443563,
+      "loss": 0.0105,
+      "step": 1384
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00013801098230628432,
+      "loss": 0.062,
+      "step": 1385
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00013782794386821233,
+      "loss": 0.0944,
+      "step": 1386
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001376449054301403,
+      "loss": 0.0294,
+      "step": 1387
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013746186699206832,
+      "loss": 0.0501,
+      "step": 1388
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001372788285539963,
+      "loss": 0.0335,
+      "step": 1389
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013709579011592432,
+      "loss": 0.0913,
+      "step": 1390
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013691275167785233,
+      "loss": 0.0731,
+      "step": 1391
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013672971323978035,
+      "loss": 0.0336,
+      "step": 1392
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013654667480170833,
+      "loss": 0.0182,
+      "step": 1393
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013636363636363634,
+      "loss": 0.0121,
+      "step": 1394
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013618059792556435,
+      "loss": 0.071,
+      "step": 1395
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013599755948749237,
+      "loss": 0.0498,
+      "step": 1396
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013581452104942038,
+      "loss": 0.0338,
+      "step": 1397
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013563148261134836,
+      "loss": 0.0178,
+      "step": 1398
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013544844417327637,
+      "loss": 0.0272,
+      "step": 1399
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013526540573520439,
+      "loss": 0.0421,
+      "step": 1400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001350823672971324,
+      "loss": 0.0158,
+      "step": 1401
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013489932885906038,
+      "loss": 0.0648,
+      "step": 1402
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001347162904209884,
+      "loss": 0.0258,
+      "step": 1403
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001345332519829164,
+      "loss": 0.0118,
+      "step": 1404
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001343502135448444,
+      "loss": 0.0464,
+      "step": 1405
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001341671751067724,
+      "loss": 0.0153,
+      "step": 1406
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013398413666870041,
+      "loss": 0.0431,
+      "step": 1407
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013380109823062843,
+      "loss": 0.0635,
+      "step": 1408
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001336180597925564,
+      "loss": 0.1599,
+      "step": 1409
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013343502135448442,
+      "loss": 0.0136,
+      "step": 1410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013325198291641244,
+      "loss": 0.0206,
+      "step": 1411
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013306894447834045,
+      "loss": 0.0357,
+      "step": 1412
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013288590604026846,
+      "loss": 0.0546,
+      "step": 1413
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013270286760219644,
+      "loss": 0.0129,
+      "step": 1414
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013251982916412446,
+      "loss": 0.0266,
+      "step": 1415
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013233679072605247,
+      "loss": 0.0291,
+      "step": 1416
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013215375228798048,
+      "loss": 0.047,
+      "step": 1417
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013197071384990846,
+      "loss": 0.0569,
+      "step": 1418
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013178767541183648,
+      "loss": 0.017,
+      "step": 1419
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001316046369737645,
+      "loss": 0.0665,
+      "step": 1420
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013142159853569247,
+      "loss": 0.0926,
+      "step": 1421
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013123856009762048,
+      "loss": 0.0214,
+      "step": 1422
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001310555216595485,
+      "loss": 0.0082,
+      "step": 1423
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013087248322147648,
+      "loss": 0.0277,
+      "step": 1424
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001306894447834045,
+      "loss": 0.0448,
+      "step": 1425
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001305064063453325,
+      "loss": 0.1196,
+      "step": 1426
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013032336790726052,
+      "loss": 0.0168,
+      "step": 1427
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013014032946918853,
+      "loss": 0.01,
+      "step": 1428
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001299572910311165,
+      "loss": 0.0436,
+      "step": 1429
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00012977425259304452,
+      "loss": 0.0143,
+      "step": 1430
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00012959121415497254,
+      "loss": 0.0381,
+      "step": 1431
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00012940817571690055,
+      "loss": 0.069,
+      "step": 1432
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00012922513727882856,
+      "loss": 0.0089,
+      "step": 1433
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00012904209884075655,
+      "loss": 0.0484,
+      "step": 1434
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00012885906040268456,
+      "loss": 0.0324,
+      "step": 1435
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00012867602196461257,
+      "loss": 0.0354,
+      "step": 1436
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00012849298352654055,
+      "loss": 0.0459,
+      "step": 1437
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00012830994508846857,
+      "loss": 0.0451,
+      "step": 1438
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00012812690665039658,
+      "loss": 0.0167,
+      "step": 1439
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00012794386821232456,
+      "loss": 0.0361,
+      "step": 1440
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00012776082977425257,
+      "loss": 0.0275,
+      "step": 1441
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00012757779133618059,
+      "loss": 0.0167,
+      "step": 1442
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001273947528981086,
+      "loss": 0.0278,
+      "step": 1443
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00012721171446003658,
+      "loss": 0.0067,
+      "step": 1444
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001270286760219646,
+      "loss": 0.0247,
+      "step": 1445
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001268456375838926,
+      "loss": 0.012,
+      "step": 1446
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00012666259914582062,
+      "loss": 0.0156,
+      "step": 1447
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00012647956070774863,
+      "loss": 0.0036,
+      "step": 1448
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00012629652226967661,
+      "loss": 0.0119,
+      "step": 1449
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00012611348383160463,
+      "loss": 0.0332,
+      "step": 1450
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00012593044539353264,
+      "loss": 0.077,
+      "step": 1451
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00012574740695546065,
+      "loss": 0.0364,
+      "step": 1452
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00012556436851738863,
+      "loss": 0.0285,
+      "step": 1453
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00012538133007931665,
+      "loss": 0.0308,
+      "step": 1454
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00012519829164124466,
+      "loss": 0.0341,
+      "step": 1455
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00012501525320317264,
+      "loss": 0.0096,
+      "step": 1456
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00012483221476510066,
+      "loss": 0.033,
+      "step": 1457
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00012464917632702867,
+      "loss": 0.0156,
+      "step": 1458
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00012446613788895665,
+      "loss": 0.0235,
+      "step": 1459
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00012428309945088466,
+      "loss": 0.0207,
+      "step": 1460
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00012410006101281268,
+      "loss": 0.0612,
+      "step": 1461
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001239170225747407,
+      "loss": 0.0261,
+      "step": 1462
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001237339841366687,
+      "loss": 0.0029,
+      "step": 1463
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00012355094569859668,
+      "loss": 0.0066,
+      "step": 1464
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001233679072605247,
+      "loss": 0.0217,
+      "step": 1465
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001231848688224527,
+      "loss": 0.0109,
+      "step": 1466
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00012300183038438072,
+      "loss": 0.0083,
+      "step": 1467
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00012281879194630873,
+      "loss": 0.0204,
+      "step": 1468
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00012263575350823672,
+      "loss": 0.0528,
+      "step": 1469
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00012245271507016473,
+      "loss": 0.0312,
+      "step": 1470
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00012226967663209274,
+      "loss": 0.0137,
+      "step": 1471
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00012208663819402072,
+      "loss": 0.023,
+      "step": 1472
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00012190359975594875,
+      "loss": 0.0044,
+      "step": 1473
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00012172056131787673,
+      "loss": 0.061,
+      "step": 1474
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00012153752287980475,
+      "loss": 0.0378,
+      "step": 1475
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00012135448444173276,
+      "loss": 0.0537,
+      "step": 1476
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00012117144600366076,
+      "loss": 0.0237,
+      "step": 1477
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00012098840756558877,
+      "loss": 0.0157,
+      "step": 1478
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00012080536912751675,
+      "loss": 0.0458,
+      "step": 1479
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00012062233068944477,
+      "loss": 0.0264,
+      "step": 1480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00012043929225137278,
+      "loss": 0.0481,
+      "step": 1481
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00012025625381330079,
+      "loss": 0.0212,
+      "step": 1482
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00012007321537522879,
+      "loss": 0.0637,
+      "step": 1483
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00011989017693715679,
+      "loss": 0.0444,
+      "step": 1484
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001197071384990848,
+      "loss": 0.0299,
+      "step": 1485
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001195241000610128,
+      "loss": 0.0665,
+      "step": 1486
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00011934106162294081,
+      "loss": 0.0549,
+      "step": 1487
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00011915802318486882,
+      "loss": 0.0294,
+      "step": 1488
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001189749847467968,
+      "loss": 0.0028,
+      "step": 1489
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00011879194630872482,
+      "loss": 0.0154,
+      "step": 1490
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00011860890787065283,
+      "loss": 0.0298,
+      "step": 1491
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00011842586943258084,
+      "loss": 0.0358,
+      "step": 1492
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00011824283099450884,
+      "loss": 0.0175,
+      "step": 1493
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011805979255643685,
+      "loss": 0.0058,
+      "step": 1494
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011787675411836483,
+      "loss": 0.036,
+      "step": 1495
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011769371568029285,
+      "loss": 0.0151,
+      "step": 1496
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011751067724222086,
+      "loss": 0.0231,
+      "step": 1497
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011732763880414887,
+      "loss": 0.0064,
+      "step": 1498
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011714460036607687,
+      "loss": 0.0455,
+      "step": 1499
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011696156192800487,
+      "loss": 0.051,
+      "step": 1500
+    },
+    {
+      "epoch": 2.1,
+      "eval_loss": 0.09693796932697296,
+      "eval_runtime": 13282.1925,
+      "eval_samples_per_second": 0.199,
+      "eval_steps_per_second": 0.099,
+      "eval_wer": 0.0774494248314161,
+      "step": 1500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011677852348993288,
+      "loss": 0.111,
+      "step": 1501
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011659548505186088,
+      "loss": 0.1146,
+      "step": 1502
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011641244661378889,
+      "loss": 0.0234,
+      "step": 1503
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001162294081757169,
+      "loss": 0.0145,
+      "step": 1504
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011604636973764489,
+      "loss": 0.0144,
+      "step": 1505
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001158633312995729,
+      "loss": 0.0123,
+      "step": 1506
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011568029286150091,
+      "loss": 0.0212,
+      "step": 1507
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011549725442342891,
+      "loss": 0.0431,
+      "step": 1508
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011531421598535692,
+      "loss": 0.0127,
+      "step": 1509
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011513117754728492,
+      "loss": 0.0439,
+      "step": 1510
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011494813910921292,
+      "loss": 0.0065,
+      "step": 1511
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011476510067114093,
+      "loss": 0.0234,
+      "step": 1512
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011458206223306894,
+      "loss": 0.0157,
+      "step": 1513
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011439902379499695,
+      "loss": 0.0085,
+      "step": 1514
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011421598535692494,
+      "loss": 0.0292,
+      "step": 1515
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011403294691885295,
+      "loss": 0.02,
+      "step": 1516
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011384990848078096,
+      "loss": 0.0376,
+      "step": 1517
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011366687004270896,
+      "loss": 0.0135,
+      "step": 1518
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011348383160463697,
+      "loss": 0.023,
+      "step": 1519
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011330079316656497,
+      "loss": 0.1056,
+      "step": 1520
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011311775472849297,
+      "loss": 0.0795,
+      "step": 1521
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011293471629042098,
+      "loss": 0.0112,
+      "step": 1522
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011275167785234899,
+      "loss": 0.0187,
+      "step": 1523
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011256863941427699,
+      "loss": 0.0386,
+      "step": 1524
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011238560097620499,
+      "loss": 0.0046,
+      "step": 1525
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.000112202562538133,
+      "loss": 0.2234,
+      "step": 1526
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.000112019524100061,
+      "loss": 0.0235,
+      "step": 1527
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011183648566198901,
+      "loss": 0.0234,
+      "step": 1528
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011165344722391702,
+      "loss": 0.0238,
+      "step": 1529
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.000111470408785845,
+      "loss": 0.0058,
+      "step": 1530
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011128737034777302,
+      "loss": 0.0476,
+      "step": 1531
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011110433190970103,
+      "loss": 0.0528,
+      "step": 1532
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011092129347162904,
+      "loss": 0.0416,
+      "step": 1533
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011073825503355704,
+      "loss": 0.0115,
+      "step": 1534
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011055521659548504,
+      "loss": 0.0106,
+      "step": 1535
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011037217815741305,
+      "loss": 0.0322,
+      "step": 1536
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00011018913971934105,
+      "loss": 0.0323,
+      "step": 1537
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00011000610128126906,
+      "loss": 0.0666,
+      "step": 1538
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00010982306284319707,
+      "loss": 0.0219,
+      "step": 1539
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00010964002440512506,
+      "loss": 0.0335,
+      "step": 1540
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00010945698596705307,
+      "loss": 0.0134,
+      "step": 1541
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00010927394752898108,
+      "loss": 0.0227,
+      "step": 1542
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00010909090909090908,
+      "loss": 0.0326,
+      "step": 1543
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00010890787065283709,
+      "loss": 0.1463,
+      "step": 1544
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00010872483221476509,
+      "loss": 0.0095,
+      "step": 1545
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00010854179377669309,
+      "loss": 0.0291,
+      "step": 1546
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001083587553386211,
+      "loss": 0.0189,
+      "step": 1547
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00010817571690054911,
+      "loss": 0.0156,
+      "step": 1548
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00010799267846247712,
+      "loss": 0.0327,
+      "step": 1549
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00010780964002440511,
+      "loss": 0.0227,
+      "step": 1550
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00010762660158633312,
+      "loss": 0.0563,
+      "step": 1551
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00010744356314826113,
+      "loss": 0.0284,
+      "step": 1552
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00010726052471018913,
+      "loss": 0.0331,
+      "step": 1553
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00010707748627211714,
+      "loss": 0.012,
+      "step": 1554
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00010689444783404513,
+      "loss": 0.0313,
+      "step": 1555
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00010671140939597314,
+      "loss": 0.0308,
+      "step": 1556
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00010652837095790115,
+      "loss": 0.0076,
+      "step": 1557
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00010634533251982916,
+      "loss": 0.0188,
+      "step": 1558
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00010616229408175716,
+      "loss": 0.1164,
+      "step": 1559
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00010597925564368516,
+      "loss": 0.0104,
+      "step": 1560
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00010579621720561317,
+      "loss": 0.0133,
+      "step": 1561
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00010561317876754117,
+      "loss": 0.0469,
+      "step": 1562
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00010543014032946918,
+      "loss": 0.0341,
+      "step": 1563
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00010524710189139719,
+      "loss": 0.0601,
+      "step": 1564
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00010506406345332518,
+      "loss": 0.0123,
+      "step": 1565
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00010488102501525319,
+      "loss": 0.01,
+      "step": 1566
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001046979865771812,
+      "loss": 0.0158,
+      "step": 1567
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001045149481391092,
+      "loss": 0.0031,
+      "step": 1568
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00010433190970103721,
+      "loss": 0.0406,
+      "step": 1569
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00010414887126296521,
+      "loss": 0.0458,
+      "step": 1570
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00010396583282489321,
+      "loss": 0.0425,
+      "step": 1571
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00010378279438682122,
+      "loss": 0.0779,
+      "step": 1572
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00010359975594874923,
+      "loss": 0.0106,
+      "step": 1573
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00010341671751067724,
+      "loss": 0.0237,
+      "step": 1574
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00010323367907260523,
+      "loss": 0.0288,
+      "step": 1575
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00010305064063453324,
+      "loss": 0.006,
+      "step": 1576
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00010286760219646125,
+      "loss": 0.0182,
+      "step": 1577
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00010268456375838925,
+      "loss": 0.0135,
+      "step": 1578
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00010250152532031726,
+      "loss": 0.0145,
+      "step": 1579
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00010231848688224527,
+      "loss": 0.043,
+      "step": 1580
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00010213544844417326,
+      "loss": 0.027,
+      "step": 1581
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00010195241000610127,
+      "loss": 0.0237,
+      "step": 1582
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00010176937156802928,
+      "loss": 0.0129,
+      "step": 1583
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00010158633312995728,
+      "loss": 0.009,
+      "step": 1584
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00010140329469188529,
+      "loss": 0.0099,
+      "step": 1585
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00010122025625381329,
+      "loss": 0.0394,
+      "step": 1586
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00010103721781574129,
+      "loss": 0.0318,
+      "step": 1587
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001008541793776693,
+      "loss": 0.0366,
+      "step": 1588
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00010067114093959731,
+      "loss": 0.0326,
+      "step": 1589
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00010048810250152532,
+      "loss": 0.0038,
+      "step": 1590
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00010030506406345331,
+      "loss": 0.0246,
+      "step": 1591
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00010012202562538132,
+      "loss": 0.0431,
+      "step": 1592
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.993898718730933e-05,
+      "loss": 0.0398,
+      "step": 1593
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.975594874923733e-05,
+      "loss": 0.0442,
+      "step": 1594
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.957291031116534e-05,
+      "loss": 0.0258,
+      "step": 1595
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.938987187309334e-05,
+      "loss": 0.0258,
+      "step": 1596
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.920683343502134e-05,
+      "loss": 0.0149,
+      "step": 1597
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.902379499694935e-05,
+      "loss": 0.0181,
+      "step": 1598
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.884075655887736e-05,
+      "loss": 0.0306,
+      "step": 1599
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.865771812080536e-05,
+      "loss": 0.0053,
+      "step": 1600
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.847467968273336e-05,
+      "loss": 0.041,
+      "step": 1601
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.829164124466137e-05,
+      "loss": 0.0117,
+      "step": 1602
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.810860280658937e-05,
+      "loss": 0.0318,
+      "step": 1603
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.792556436851738e-05,
+      "loss": 0.0139,
+      "step": 1604
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.77425259304454e-05,
+      "loss": 0.0207,
+      "step": 1605
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.755948749237338e-05,
+      "loss": 0.0094,
+      "step": 1606
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.737644905430139e-05,
+      "loss": 0.0199,
+      "step": 1607
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.71934106162294e-05,
+      "loss": 0.0186,
+      "step": 1608
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.701037217815741e-05,
+      "loss": 0.047,
+      "step": 1609
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.682733374008541e-05,
+      "loss": 0.033,
+      "step": 1610
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.664429530201341e-05,
+      "loss": 0.0098,
+      "step": 1611
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.646125686394142e-05,
+      "loss": 0.0268,
+      "step": 1612
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.627821842586942e-05,
+      "loss": 0.0096,
+      "step": 1613
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.609517998779743e-05,
+      "loss": 0.0074,
+      "step": 1614
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.591214154972544e-05,
+      "loss": 0.0178,
+      "step": 1615
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.572910311165343e-05,
+      "loss": 0.034,
+      "step": 1616
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.554606467358144e-05,
+      "loss": 0.012,
+      "step": 1617
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.536302623550945e-05,
+      "loss": 0.0122,
+      "step": 1618
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.517998779743745e-05,
+      "loss": 0.0092,
+      "step": 1619
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.499694935936546e-05,
+      "loss": 0.031,
+      "step": 1620
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.481391092129346e-05,
+      "loss": 0.0098,
+      "step": 1621
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.463087248322146e-05,
+      "loss": 0.004,
+      "step": 1622
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.444783404514947e-05,
+      "loss": 0.014,
+      "step": 1623
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.426479560707748e-05,
+      "loss": 0.0464,
+      "step": 1624
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.40817571690055e-05,
+      "loss": 0.0241,
+      "step": 1625
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.389871873093348e-05,
+      "loss": 0.2683,
+      "step": 1626
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.371568029286149e-05,
+      "loss": 0.0161,
+      "step": 1627
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.353264185478949e-05,
+      "loss": 0.0056,
+      "step": 1628
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.33496034167175e-05,
+      "loss": 0.013,
+      "step": 1629
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.316656497864551e-05,
+      "loss": 0.0322,
+      "step": 1630
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.29835265405735e-05,
+      "loss": 0.0173,
+      "step": 1631
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.280048810250151e-05,
+      "loss": 0.0259,
+      "step": 1632
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.261744966442952e-05,
+      "loss": 0.0332,
+      "step": 1633
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.243441122635753e-05,
+      "loss": 0.0243,
+      "step": 1634
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.225137278828553e-05,
+      "loss": 0.041,
+      "step": 1635
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.206833435021353e-05,
+      "loss": 0.0285,
+      "step": 1636
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.188529591214154e-05,
+      "loss": 0.0239,
+      "step": 1637
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.170225747406954e-05,
+      "loss": 0.0524,
+      "step": 1638
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.151921903599755e-05,
+      "loss": 0.0085,
+      "step": 1639
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.133618059792557e-05,
+      "loss": 0.0185,
+      "step": 1640
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.115314215985355e-05,
+      "loss": 0.0198,
+      "step": 1641
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.097010372178156e-05,
+      "loss": 0.0101,
+      "step": 1642
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.078706528370957e-05,
+      "loss": 0.0285,
+      "step": 1643
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.060402684563757e-05,
+      "loss": 0.0118,
+      "step": 1644
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.042098840756558e-05,
+      "loss": 0.0244,
+      "step": 1645
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.023794996949358e-05,
+      "loss": 0.0108,
+      "step": 1646
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.005491153142158e-05,
+      "loss": 0.0574,
+      "step": 1647
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.987187309334959e-05,
+      "loss": 0.0049,
+      "step": 1648
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.96888346552776e-05,
+      "loss": 0.0025,
+      "step": 1649
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.950579621720562e-05,
+      "loss": 0.029,
+      "step": 1650
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.93227577791336e-05,
+      "loss": 0.0331,
+      "step": 1651
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.913971934106161e-05,
+      "loss": 0.0239,
+      "step": 1652
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.895668090298962e-05,
+      "loss": 0.0073,
+      "step": 1653
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.877364246491762e-05,
+      "loss": 0.0059,
+      "step": 1654
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.859060402684563e-05,
+      "loss": 0.0428,
+      "step": 1655
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.840756558877363e-05,
+      "loss": 0.0139,
+      "step": 1656
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.822452715070163e-05,
+      "loss": 0.0136,
+      "step": 1657
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.804148871262964e-05,
+      "loss": 0.0294,
+      "step": 1658
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.785845027455765e-05,
+      "loss": 0.0075,
+      "step": 1659
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.767541183648565e-05,
+      "loss": 0.0308,
+      "step": 1660
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.749237339841365e-05,
+      "loss": 0.0093,
+      "step": 1661
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.730933496034166e-05,
+      "loss": 0.0176,
+      "step": 1662
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.712629652226966e-05,
+      "loss": 0.0187,
+      "step": 1663
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.694325808419767e-05,
+      "loss": 0.0277,
+      "step": 1664
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.676021964612569e-05,
+      "loss": 0.009,
+      "step": 1665
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.65771812080537e-05,
+      "loss": 0.0328,
+      "step": 1666
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.639414276998168e-05,
+      "loss": 0.0383,
+      "step": 1667
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.62111043319097e-05,
+      "loss": 0.0497,
+      "step": 1668
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.60280658938377e-05,
+      "loss": 0.0086,
+      "step": 1669
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.58450274557657e-05,
+      "loss": 0.026,
+      "step": 1670
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.566198901769372e-05,
+      "loss": 0.0097,
+      "step": 1671
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.547895057962171e-05,
+      "loss": 0.0422,
+      "step": 1672
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.529591214154971e-05,
+      "loss": 0.0166,
+      "step": 1673
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.511287370347772e-05,
+      "loss": 0.0444,
+      "step": 1674
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.492983526540574e-05,
+      "loss": 0.0232,
+      "step": 1675
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.474679682733373e-05,
+      "loss": 0.1412,
+      "step": 1676
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.456375838926173e-05,
+      "loss": 0.0138,
+      "step": 1677
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.438071995118974e-05,
+      "loss": 0.0172,
+      "step": 1678
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.419768151311774e-05,
+      "loss": 0.0181,
+      "step": 1679
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.401464307504575e-05,
+      "loss": 0.0142,
+      "step": 1680
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.383160463697377e-05,
+      "loss": 0.1001,
+      "step": 1681
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.364856619890175e-05,
+      "loss": 0.0461,
+      "step": 1682
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.346552776082976e-05,
+      "loss": 0.0201,
+      "step": 1683
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.328248932275778e-05,
+      "loss": 0.0166,
+      "step": 1684
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.309945088468579e-05,
+      "loss": 0.0082,
+      "step": 1685
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.291641244661379e-05,
+      "loss": 0.0122,
+      "step": 1686
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.273337400854178e-05,
+      "loss": 0.011,
+      "step": 1687
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.25503355704698e-05,
+      "loss": 0.0221,
+      "step": 1688
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.23672971323978e-05,
+      "loss": 0.013,
+      "step": 1689
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.21842586943258e-05,
+      "loss": 0.0163,
+      "step": 1690
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.200122025625382e-05,
+      "loss": 0.0027,
+      "step": 1691
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.18181818181818e-05,
+      "loss": 0.0115,
+      "step": 1692
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.163514338010981e-05,
+      "loss": 0.0161,
+      "step": 1693
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.145210494203783e-05,
+      "loss": 0.0577,
+      "step": 1694
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.126906650396582e-05,
+      "loss": 0.0057,
+      "step": 1695
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.108602806589384e-05,
+      "loss": 0.0252,
+      "step": 1696
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.090298962782183e-05,
+      "loss": 0.0112,
+      "step": 1697
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.071995118974983e-05,
+      "loss": 0.0243,
+      "step": 1698
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.053691275167784e-05,
+      "loss": 0.014,
+      "step": 1699
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.035387431360586e-05,
+      "loss": 0.0242,
+      "step": 1700
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.017083587553387e-05,
+      "loss": 0.0038,
+      "step": 1701
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.998779743746185e-05,
+      "loss": 0.0226,
+      "step": 1702
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.980475899938987e-05,
+      "loss": 0.0474,
+      "step": 1703
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.962172056131786e-05,
+      "loss": 0.0135,
+      "step": 1704
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.943868212324588e-05,
+      "loss": 0.0127,
+      "step": 1705
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.925564368517389e-05,
+      "loss": 0.017,
+      "step": 1706
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.907260524710187e-05,
+      "loss": 0.003,
+      "step": 1707
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.888956680902988e-05,
+      "loss": 0.0321,
+      "step": 1708
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.87065283709579e-05,
+      "loss": 0.0133,
+      "step": 1709
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.852348993288591e-05,
+      "loss": 0.0259,
+      "step": 1710
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.83404514948139e-05,
+      "loss": 0.0569,
+      "step": 1711
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.81574130567419e-05,
+      "loss": 0.007,
+      "step": 1712
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.797437461866992e-05,
+      "loss": 0.0395,
+      "step": 1713
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.779133618059791e-05,
+      "loss": 0.0417,
+      "step": 1714
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.760829774252593e-05,
+      "loss": 0.0213,
+      "step": 1715
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.742525930445394e-05,
+      "loss": 0.0172,
+      "step": 1716
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.724222086638192e-05,
+      "loss": 0.021,
+      "step": 1717
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.705918242830993e-05,
+      "loss": 0.0197,
+      "step": 1718
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.687614399023795e-05,
+      "loss": 0.0193,
+      "step": 1719
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.669310555216594e-05,
+      "loss": 0.0162,
+      "step": 1720
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.651006711409396e-05,
+      "loss": 0.0939,
+      "step": 1721
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.632702867602195e-05,
+      "loss": 0.0088,
+      "step": 1722
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.614399023794995e-05,
+      "loss": 0.0429,
+      "step": 1723
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.596095179987797e-05,
+      "loss": 0.0796,
+      "step": 1724
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.577791336180598e-05,
+      "loss": 0.038,
+      "step": 1725
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.559487492373399e-05,
+      "loss": 0.0428,
+      "step": 1726
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.541183648566197e-05,
+      "loss": 0.0219,
+      "step": 1727
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.522879804758999e-05,
+      "loss": 0.0199,
+      "step": 1728
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.5045759609518e-05,
+      "loss": 0.0569,
+      "step": 1729
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.4862721171446e-05,
+      "loss": 0.0119,
+      "step": 1730
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.4679682733374e-05,
+      "loss": 0.056,
+      "step": 1731
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.4496644295302e-05,
+      "loss": 0.0485,
+      "step": 1732
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.431360585723002e-05,
+      "loss": 0.0154,
+      "step": 1733
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.413056741915802e-05,
+      "loss": 0.0079,
+      "step": 1734
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.394752898108603e-05,
+      "loss": 0.0146,
+      "step": 1735
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.376449054301403e-05,
+      "loss": 0.0216,
+      "step": 1736
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.358145210494204e-05,
+      "loss": 0.0136,
+      "step": 1737
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.339841366687004e-05,
+      "loss": 0.0068,
+      "step": 1738
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.321537522879803e-05,
+      "loss": 0.0043,
+      "step": 1739
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.303233679072605e-05,
+      "loss": 0.0278,
+      "step": 1740
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.284929835265404e-05,
+      "loss": 0.0047,
+      "step": 1741
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.266625991458206e-05,
+      "loss": 0.0595,
+      "step": 1742
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.248322147651007e-05,
+      "loss": 0.0093,
+      "step": 1743
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.230018303843807e-05,
+      "loss": 0.0286,
+      "step": 1744
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.211714460036608e-05,
+      "loss": 0.0421,
+      "step": 1745
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.193410616229408e-05,
+      "loss": 0.0356,
+      "step": 1746
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.175106772422208e-05,
+      "loss": 0.0056,
+      "step": 1747
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.156802928615009e-05,
+      "loss": 0.0141,
+      "step": 1748
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.138499084807809e-05,
+      "loss": 0.0693,
+      "step": 1749
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.12019524100061e-05,
+      "loss": 0.0207,
+      "step": 1750
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.10189139719341e-05,
+      "loss": 0.0047,
+      "step": 1751
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.083587553386211e-05,
+      "loss": 0.0063,
+      "step": 1752
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.065283709579012e-05,
+      "loss": 0.0039,
+      "step": 1753
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.046979865771812e-05,
+      "loss": 0.0051,
+      "step": 1754
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.028676021964612e-05,
+      "loss": 0.0206,
+      "step": 1755
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.010372178157411e-05,
+      "loss": 0.0032,
+      "step": 1756
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.992068334350213e-05,
+      "loss": 0.0071,
+      "step": 1757
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.973764490543014e-05,
+      "loss": 0.0035,
+      "step": 1758
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.955460646735814e-05,
+      "loss": 0.0193,
+      "step": 1759
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.937156802928615e-05,
+      "loss": 0.0021,
+      "step": 1760
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.918852959121415e-05,
+      "loss": 0.003,
+      "step": 1761
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.900549115314216e-05,
+      "loss": 0.004,
+      "step": 1762
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.882245271507016e-05,
+      "loss": 0.0264,
+      "step": 1763
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.863941427699815e-05,
+      "loss": 0.0354,
+      "step": 1764
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.845637583892617e-05,
+      "loss": 0.0064,
+      "step": 1765
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.827333740085416e-05,
+      "loss": 0.031,
+      "step": 1766
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.809029896278218e-05,
+      "loss": 0.0102,
+      "step": 1767
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.790726052471019e-05,
+      "loss": 0.0047,
+      "step": 1768
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.772422208663819e-05,
+      "loss": 0.008,
+      "step": 1769
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.75411836485662e-05,
+      "loss": 0.0142,
+      "step": 1770
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.73581452104942e-05,
+      "loss": 0.028,
+      "step": 1771
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.71751067724222e-05,
+      "loss": 0.0751,
+      "step": 1772
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.699206833435021e-05,
+      "loss": 0.0427,
+      "step": 1773
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.68090298962782e-05,
+      "loss": 0.0176,
+      "step": 1774
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.662599145820622e-05,
+      "loss": 0.1036,
+      "step": 1775
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.644295302013423e-05,
+      "loss": 0.0184,
+      "step": 1776
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.625991458206223e-05,
+      "loss": 0.0091,
+      "step": 1777
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.607687614399024e-05,
+      "loss": 0.0253,
+      "step": 1778
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.589383770591824e-05,
+      "loss": 0.0381,
+      "step": 1779
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.571079926784624e-05,
+      "loss": 0.0026,
+      "step": 1780
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.552776082977425e-05,
+      "loss": 0.0092,
+      "step": 1781
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.534472239170225e-05,
+      "loss": 0.0521,
+      "step": 1782
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.516168395363026e-05,
+      "loss": 0.0233,
+      "step": 1783
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.497864551555826e-05,
+      "loss": 0.0336,
+      "step": 1784
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.479560707748627e-05,
+      "loss": 0.0107,
+      "step": 1785
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.461256863941428e-05,
+      "loss": 0.0109,
+      "step": 1786
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.442953020134228e-05,
+      "loss": 0.0217,
+      "step": 1787
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.424649176327028e-05,
+      "loss": 0.0051,
+      "step": 1788
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.406345332519829e-05,
+      "loss": 0.0083,
+      "step": 1789
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.388041488712629e-05,
+      "loss": 0.0293,
+      "step": 1790
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.36973764490543e-05,
+      "loss": 0.0056,
+      "step": 1791
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.35143380109823e-05,
+      "loss": 0.0112,
+      "step": 1792
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.333129957291031e-05,
+      "loss": 0.0202,
+      "step": 1793
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.314826113483831e-05,
+      "loss": 0.0341,
+      "step": 1794
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.296522269676632e-05,
+      "loss": 0.0254,
+      "step": 1795
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.278218425869432e-05,
+      "loss": 0.0116,
+      "step": 1796
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.259914582062233e-05,
+      "loss": 0.0148,
+      "step": 1797
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.241610738255033e-05,
+      "loss": 0.0073,
+      "step": 1798
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.223306894447833e-05,
+      "loss": 0.0427,
+      "step": 1799
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.205003050640634e-05,
+      "loss": 0.0254,
+      "step": 1800
+    },
+    {
+      "epoch": 2.52,
+      "eval_loss": 0.09300445765256882,
+      "eval_runtime": 13233.6251,
+      "eval_samples_per_second": 0.2,
+      "eval_steps_per_second": 0.1,
+      "eval_wer": 0.07151923839746133,
+      "step": 1800
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.186699206833435e-05,
+      "loss": 0.0373,
+      "step": 1801
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.168395363026235e-05,
+      "loss": 0.0331,
+      "step": 1802
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.150091519219036e-05,
+      "loss": 0.1211,
+      "step": 1803
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.131787675411836e-05,
+      "loss": 0.008,
+      "step": 1804
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.113483831604637e-05,
+      "loss": 0.0119,
+      "step": 1805
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.0951799877974375e-05,
+      "loss": 0.0261,
+      "step": 1806
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.0768761439902373e-05,
+      "loss": 0.0272,
+      "step": 1807
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.058572300183038e-05,
+      "loss": 0.0115,
+      "step": 1808
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.040268456375838e-05,
+      "loss": 0.0342,
+      "step": 1809
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.021964612568639e-05,
+      "loss": 0.0385,
+      "step": 1810
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.0036607687614394e-05,
+      "loss": 0.0084,
+      "step": 1811
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.98535692495424e-05,
+      "loss": 0.1,
+      "step": 1812
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.9670530811470404e-05,
+      "loss": 0.0273,
+      "step": 1813
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.94874923733984e-05,
+      "loss": 0.0303,
+      "step": 1814
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.9304453935326414e-05,
+      "loss": 0.0143,
+      "step": 1815
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.912141549725442e-05,
+      "loss": 0.0197,
+      "step": 1816
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.893837705918242e-05,
+      "loss": 0.0059,
+      "step": 1817
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.875533862111043e-05,
+      "loss": 0.0578,
+      "step": 1818
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.8572300183038434e-05,
+      "loss": 0.002,
+      "step": 1819
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.838926174496644e-05,
+      "loss": 0.0421,
+      "step": 1820
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.8206223306894445e-05,
+      "loss": 0.007,
+      "step": 1821
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.802318486882244e-05,
+      "loss": 0.0262,
+      "step": 1822
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.7840146430750455e-05,
+      "loss": 0.0056,
+      "step": 1823
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.765710799267846e-05,
+      "loss": 0.0372,
+      "step": 1824
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.747406955460646e-05,
+      "loss": 0.0652,
+      "step": 1825
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.729103111653447e-05,
+      "loss": 0.0202,
+      "step": 1826
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.710799267846247e-05,
+      "loss": 0.0345,
+      "step": 1827
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.692495424039048e-05,
+      "loss": 0.0104,
+      "step": 1828
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.6741915802318485e-05,
+      "loss": 0.0195,
+      "step": 1829
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.6558877364246484e-05,
+      "loss": 0.0255,
+      "step": 1830
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.6375838926174495e-05,
+      "loss": 0.0093,
+      "step": 1831
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.6192800488102494e-05,
+      "loss": 0.0324,
+      "step": 1832
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.60097620500305e-05,
+      "loss": 0.0142,
+      "step": 1833
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.582672361195851e-05,
+      "loss": 0.0166,
+      "step": 1834
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.564368517388651e-05,
+      "loss": 0.0075,
+      "step": 1835
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.546064673581452e-05,
+      "loss": 0.0124,
+      "step": 1836
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.527760829774252e-05,
+      "loss": 0.0116,
+      "step": 1837
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.5094569859670524e-05,
+      "loss": 0.0047,
+      "step": 1838
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.4911531421598536e-05,
+      "loss": 0.0297,
+      "step": 1839
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.4728492983526534e-05,
+      "loss": 0.0023,
+      "step": 1840
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.454545454545454e-05,
+      "loss": 0.0159,
+      "step": 1841
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.4362416107382545e-05,
+      "loss": 0.0318,
+      "step": 1842
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.417937766931055e-05,
+      "loss": 0.0122,
+      "step": 1843
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.399633923123856e-05,
+      "loss": 0.0088,
+      "step": 1844
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.381330079316656e-05,
+      "loss": 0.0412,
+      "step": 1845
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3630262355094565e-05,
+      "loss": 0.0604,
+      "step": 1846
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.344722391702256e-05,
+      "loss": 0.0421,
+      "step": 1847
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3264185478950575e-05,
+      "loss": 0.0139,
+      "step": 1848
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.308114704087858e-05,
+      "loss": 0.0051,
+      "step": 1849
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.2898108602806585e-05,
+      "loss": 0.032,
+      "step": 1850
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.271507016473459e-05,
+      "loss": 0.0162,
+      "step": 1851
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.253203172666259e-05,
+      "loss": 0.0302,
+      "step": 1852
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.23489932885906e-05,
+      "loss": 0.0141,
+      "step": 1853
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2165954850518606e-05,
+      "loss": 0.0107,
+      "step": 1854
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.1982916412446604e-05,
+      "loss": 0.0066,
+      "step": 1855
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.1799877974374616e-05,
+      "loss": 0.0106,
+      "step": 1856
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.1616839536302614e-05,
+      "loss": 0.0303,
+      "step": 1857
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1433801098230626e-05,
+      "loss": 0.0042,
+      "step": 1858
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.125076266015863e-05,
+      "loss": 0.0343,
+      "step": 1859
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.106772422208663e-05,
+      "loss": 0.0134,
+      "step": 1860
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.088468578401464e-05,
+      "loss": 0.0085,
+      "step": 1861
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.0701647345942646e-05,
+      "loss": 0.0123,
+      "step": 1862
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.0518608907870644e-05,
+      "loss": 0.0148,
+      "step": 1863
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.0335570469798656e-05,
+      "loss": 0.0305,
+      "step": 1864
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.0152532031726655e-05,
+      "loss": 0.0367,
+      "step": 1865
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.9969493593654667e-05,
+      "loss": 0.0036,
+      "step": 1866
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.978645515558267e-05,
+      "loss": 0.0118,
+      "step": 1867
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.960341671751067e-05,
+      "loss": 0.0544,
+      "step": 1868
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.942037827943868e-05,
+      "loss": 0.0082,
+      "step": 1869
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.923733984136668e-05,
+      "loss": 0.0089,
+      "step": 1870
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.9054301403294685e-05,
+      "loss": 0.0252,
+      "step": 1871
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.88712629652227e-05,
+      "loss": 0.0184,
+      "step": 1872
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.8688224527150695e-05,
+      "loss": 0.0286,
+      "step": 1873
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.850518608907871e-05,
+      "loss": 0.0112,
+      "step": 1874
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.8322147651006705e-05,
+      "loss": 0.0167,
+      "step": 1875
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.813910921293471e-05,
+      "loss": 0.1396,
+      "step": 1876
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.795607077486272e-05,
+      "loss": 0.0254,
+      "step": 1877
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.777303233679072e-05,
+      "loss": 0.0037,
+      "step": 1878
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.7589993898718726e-05,
+      "loss": 0.0104,
+      "step": 1879
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.740695546064673e-05,
+      "loss": 0.0115,
+      "step": 1880
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.7223917022574736e-05,
+      "loss": 0.026,
+      "step": 1881
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.704087858450275e-05,
+      "loss": 0.0084,
+      "step": 1882
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.6857840146430746e-05,
+      "loss": 0.0244,
+      "step": 1883
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.667480170835875e-05,
+      "loss": 0.0333,
+      "step": 1884
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.649176327028675e-05,
+      "loss": 0.0309,
+      "step": 1885
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.630872483221476e-05,
+      "loss": 0.0098,
+      "step": 1886
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.6125686394142766e-05,
+      "loss": 0.0082,
+      "step": 1887
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.594264795607077e-05,
+      "loss": 0.0433,
+      "step": 1888
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.575960951799878e-05,
+      "loss": 0.0705,
+      "step": 1889
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5576571079926775e-05,
+      "loss": 0.0213,
+      "step": 1890
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.539353264185479e-05,
+      "loss": 0.0043,
+      "step": 1891
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.521049420378279e-05,
+      "loss": 0.0143,
+      "step": 1892
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.502745576571079e-05,
+      "loss": 0.0612,
+      "step": 1893
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.48444173276388e-05,
+      "loss": 0.0201,
+      "step": 1894
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.46613788895668e-05,
+      "loss": 0.0347,
+      "step": 1895
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.447834045149481e-05,
+      "loss": 0.0351,
+      "step": 1896
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.429530201342282e-05,
+      "loss": 0.0226,
+      "step": 1897
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.4112263575350816e-05,
+      "loss": 0.0119,
+      "step": 1898
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.392922513727883e-05,
+      "loss": 0.024,
+      "step": 1899
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.3746186699206826e-05,
+      "loss": 0.0379,
+      "step": 1900
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.356314826113483e-05,
+      "loss": 0.04,
+      "step": 1901
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.338010982306284e-05,
+      "loss": 0.0107,
+      "step": 1902
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.319707138499084e-05,
+      "loss": 0.0063,
+      "step": 1903
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.301403294691885e-05,
+      "loss": 0.005,
+      "step": 1904
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.283099450884686e-05,
+      "loss": 0.0041,
+      "step": 1905
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.2647956070774856e-05,
+      "loss": 0.0115,
+      "step": 1906
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.246491763270287e-05,
+      "loss": 0.0405,
+      "step": 1907
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.2281879194630866e-05,
+      "loss": 0.0383,
+      "step": 1908
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.209884075655887e-05,
+      "loss": 0.0179,
+      "step": 1909
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.1915802318486883e-05,
+      "loss": 0.0197,
+      "step": 1910
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.173276388041488e-05,
+      "loss": 0.0167,
+      "step": 1911
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.1549725442342894e-05,
+      "loss": 0.0106,
+      "step": 1912
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.136668700427089e-05,
+      "loss": 0.0098,
+      "step": 1913
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.11836485661989e-05,
+      "loss": 0.027,
+      "step": 1914
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.100061012812691e-05,
+      "loss": 0.013,
+      "step": 1915
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.081757169005491e-05,
+      "loss": 0.0222,
+      "step": 1916
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.063453325198291e-05,
+      "loss": 0.0517,
+      "step": 1917
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.045149481391092e-05,
+      "loss": 0.0072,
+      "step": 1918
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.026845637583892e-05,
+      "loss": 0.0159,
+      "step": 1919
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.0085417937766934e-05,
+      "loss": 0.0218,
+      "step": 1920
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.990237949969493e-05,
+      "loss": 0.1058,
+      "step": 1921
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.971934106162294e-05,
+      "loss": 0.0542,
+      "step": 1922
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.9536302623550936e-05,
+      "loss": 0.0775,
+      "step": 1923
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.935326418547895e-05,
+      "loss": 0.0654,
+      "step": 1924
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.917022574740695e-05,
+      "loss": 0.0752,
+      "step": 1925
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.898718730933496e-05,
+      "loss": 0.0156,
+      "step": 1926
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.880414887126296e-05,
+      "loss": 0.0348,
+      "step": 1927
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.862111043319096e-05,
+      "loss": 0.0158,
+      "step": 1928
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.843807199511897e-05,
+      "loss": 0.0163,
+      "step": 1929
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.825503355704698e-05,
+      "loss": 0.0275,
+      "step": 1930
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.8071995118974977e-05,
+      "loss": 0.005,
+      "step": 1931
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.788895668090299e-05,
+      "loss": 0.023,
+      "step": 1932
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.770591824283099e-05,
+      "loss": 0.0176,
+      "step": 1933
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.7522879804759e-05,
+      "loss": 0.03,
+      "step": 1934
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.7339841366687e-05,
+      "loss": 0.0108,
+      "step": 1935
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.715680292861501e-05,
+      "loss": 0.013,
+      "step": 1936
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6973764490543014e-05,
+      "loss": 0.0101,
+      "step": 1937
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.679072605247102e-05,
+      "loss": 0.0075,
+      "step": 1938
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.660768761439902e-05,
+      "loss": 0.051,
+      "step": 1939
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.642464917632702e-05,
+      "loss": 0.0307,
+      "step": 1940
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6241610738255034e-05,
+      "loss": 0.0248,
+      "step": 1941
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.605857230018304e-05,
+      "loss": 0.0228,
+      "step": 1942
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.587553386211104e-05,
+      "loss": 0.0095,
+      "step": 1943
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.569249542403904e-05,
+      "loss": 0.0381,
+      "step": 1944
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.550945698596705e-05,
+      "loss": 0.0511,
+      "step": 1945
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.532641854789506e-05,
+      "loss": 0.0304,
+      "step": 1946
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.514338010982306e-05,
+      "loss": 0.0415,
+      "step": 1947
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.496034167175106e-05,
+      "loss": 0.0075,
+      "step": 1948
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.477730323367907e-05,
+      "loss": 0.0578,
+      "step": 1949
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.459426479560707e-05,
+      "loss": 0.013,
+      "step": 1950
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.441122635753508e-05,
+      "loss": 0.0489,
+      "step": 1951
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.422818791946308e-05,
+      "loss": 0.017,
+      "step": 1952
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.404514948139109e-05,
+      "loss": 0.0063,
+      "step": 1953
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.3862111043319094e-05,
+      "loss": 0.059,
+      "step": 1954
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.36790726052471e-05,
+      "loss": 0.0365,
+      "step": 1955
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.3496034167175104e-05,
+      "loss": 0.0144,
+      "step": 1956
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.331299572910311e-05,
+      "loss": 0.056,
+      "step": 1957
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.3129957291031114e-05,
+      "loss": 0.0574,
+      "step": 1958
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.294691885295912e-05,
+      "loss": 0.0173,
+      "step": 1959
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.2763880414887124e-05,
+      "loss": 0.0022,
+      "step": 1960
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.258084197681513e-05,
+      "loss": 0.0097,
+      "step": 1961
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.2397803538743134e-05,
+      "loss": 0.0027,
+      "step": 1962
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.221476510067114e-05,
+      "loss": 0.0275,
+      "step": 1963
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.2031726662599144e-05,
+      "loss": 0.045,
+      "step": 1964
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.184868822452715e-05,
+      "loss": 0.0058,
+      "step": 1965
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1665649786455155e-05,
+      "loss": 0.0472,
+      "step": 1966
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.148261134838316e-05,
+      "loss": 0.0201,
+      "step": 1967
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1299572910311165e-05,
+      "loss": 0.0611,
+      "step": 1968
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.111653447223916e-05,
+      "loss": 0.0109,
+      "step": 1969
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.0933496034167175e-05,
+      "loss": 0.063,
+      "step": 1970
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.075045759609518e-05,
+      "loss": 0.0094,
+      "step": 1971
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0567419158023185e-05,
+      "loss": 0.0187,
+      "step": 1972
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0384380719951187e-05,
+      "loss": 0.089,
+      "step": 1973
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.020134228187919e-05,
+      "loss": 0.0336,
+      "step": 1974
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0018303843807197e-05,
+      "loss": 0.0206,
+      "step": 1975
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9835265405735202e-05,
+      "loss": 0.025,
+      "step": 1976
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9652226967663207e-05,
+      "loss": 0.0342,
+      "step": 1977
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.946918852959121e-05,
+      "loss": 0.0031,
+      "step": 1978
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9286150091519217e-05,
+      "loss": 0.0088,
+      "step": 1979
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9103111653447222e-05,
+      "loss": 0.0229,
+      "step": 1980
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8920073215375227e-05,
+      "loss": 0.0103,
+      "step": 1981
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.873703477730323e-05,
+      "loss": 0.0798,
+      "step": 1982
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8553996339231234e-05,
+      "loss": 0.0354,
+      "step": 1983
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8370957901159243e-05,
+      "loss": 0.0073,
+      "step": 1984
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8187919463087248e-05,
+      "loss": 0.0136,
+      "step": 1985
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.800488102501525e-05,
+      "loss": 0.0404,
+      "step": 1986
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7821842586943254e-05,
+      "loss": 0.0127,
+      "step": 1987
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.763880414887126e-05,
+      "loss": 0.0206,
+      "step": 1988
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7455765710799268e-05,
+      "loss": 0.0478,
+      "step": 1989
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.727272727272727e-05,
+      "loss": 0.0426,
+      "step": 1990
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7089688834655275e-05,
+      "loss": 0.0092,
+      "step": 1991
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.690665039658328e-05,
+      "loss": 0.0125,
+      "step": 1992
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.672361195851128e-05,
+      "loss": 0.0065,
+      "step": 1993
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.654057352043929e-05,
+      "loss": 0.0365,
+      "step": 1994
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.6357535082367295e-05,
+      "loss": 0.0105,
+      "step": 1995
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.61744966442953e-05,
+      "loss": 0.0469,
+      "step": 1996
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.5991458206223302e-05,
+      "loss": 0.0345,
+      "step": 1997
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.5808419768151307e-05,
+      "loss": 0.0112,
+      "step": 1998
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.5625381330079315e-05,
+      "loss": 0.0488,
+      "step": 1999
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.544234289200732e-05,
+      "loss": 0.0082,
+      "step": 2000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.5259304453935322e-05,
+      "loss": 0.0197,
+      "step": 2001
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.5076266015863327e-05,
+      "loss": 0.0085,
+      "step": 2002
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.4893227577791336e-05,
+      "loss": 0.0119,
+      "step": 2003
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.471018913971934e-05,
+      "loss": 0.0166,
+      "step": 2004
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.4527150701647343e-05,
+      "loss": 0.0263,
+      "step": 2005
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.4344112263575348e-05,
+      "loss": 0.0153,
+      "step": 2006
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.4161073825503353e-05,
+      "loss": 0.0081,
+      "step": 2007
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.397803538743136e-05,
+      "loss": 0.0462,
+      "step": 2008
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.3794996949359363e-05,
+      "loss": 0.0113,
+      "step": 2009
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.3611958511287368e-05,
+      "loss": 0.0027,
+      "step": 2010
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.3428920073215373e-05,
+      "loss": 0.0249,
+      "step": 2011
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.3245881635143375e-05,
+      "loss": 0.009,
+      "step": 2012
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.3062843197071383e-05,
+      "loss": 0.0107,
+      "step": 2013
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.287980475899939e-05,
+      "loss": 0.0152,
+      "step": 2014
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.2696766320927393e-05,
+      "loss": 0.0363,
+      "step": 2015
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.2513727882855395e-05,
+      "loss": 0.0526,
+      "step": 2016
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.23306894447834e-05,
+      "loss": 0.0134,
+      "step": 2017
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.214765100671141e-05,
+      "loss": 0.027,
+      "step": 2018
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1964612568639414e-05,
+      "loss": 0.02,
+      "step": 2019
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1781574130567415e-05,
+      "loss": 0.0986,
+      "step": 2020
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.159853569249542e-05,
+      "loss": 0.0017,
+      "step": 2021
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.141549725442343e-05,
+      "loss": 0.0116,
+      "step": 2022
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1232458816351434e-05,
+      "loss": 0.0343,
+      "step": 2023
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1049420378279436e-05,
+      "loss": 0.0545,
+      "step": 2024
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.086638194020744e-05,
+      "loss": 0.1772,
+      "step": 2025
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0683343502135446e-05,
+      "loss": 0.0356,
+      "step": 2026
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0500305064063454e-05,
+      "loss": 0.0103,
+      "step": 2027
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0317266625991456e-05,
+      "loss": 0.0179,
+      "step": 2028
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.013422818791946e-05,
+      "loss": 0.009,
+      "step": 2029
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9951189749847466e-05,
+      "loss": 0.04,
+      "step": 2030
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9768151311775468e-05,
+      "loss": 0.0029,
+      "step": 2031
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9585112873703476e-05,
+      "loss": 0.0101,
+      "step": 2032
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.940207443563148e-05,
+      "loss": 0.0555,
+      "step": 2033
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9219035997559487e-05,
+      "loss": 0.0157,
+      "step": 2034
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9035997559487488e-05,
+      "loss": 0.0166,
+      "step": 2035
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8852959121415493e-05,
+      "loss": 0.0271,
+      "step": 2036
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.86699206833435e-05,
+      "loss": 0.0299,
+      "step": 2037
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8486882245271507e-05,
+      "loss": 0.0044,
+      "step": 2038
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.830384380719951e-05,
+      "loss": 0.0082,
+      "step": 2039
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8120805369127517e-05,
+      "loss": 0.0083,
+      "step": 2040
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.793776693105552e-05,
+      "loss": 0.0291,
+      "step": 2041
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.7754728492983524e-05,
+      "loss": 0.0349,
+      "step": 2042
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.757169005491153e-05,
+      "loss": 0.0019,
+      "step": 2043
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.7388651616839534e-05,
+      "loss": 0.0083,
+      "step": 2044
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.720561317876754e-05,
+      "loss": 0.0078,
+      "step": 2045
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.7022574740695544e-05,
+      "loss": 0.0121,
+      "step": 2046
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.683953630262355e-05,
+      "loss": 0.0217,
+      "step": 2047
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6656497864551554e-05,
+      "loss": 0.0539,
+      "step": 2048
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.647345942647956e-05,
+      "loss": 0.0172,
+      "step": 2049
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6290420988407565e-05,
+      "loss": 0.0138,
+      "step": 2050
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.610738255033557e-05,
+      "loss": 0.0132,
+      "step": 2051
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5924344112263575e-05,
+      "loss": 0.0719,
+      "step": 2052
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.574130567419158e-05,
+      "loss": 0.0231,
+      "step": 2053
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.555826723611958e-05,
+      "loss": 0.0073,
+      "step": 2054
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.537522879804759e-05,
+      "loss": 0.0233,
+      "step": 2055
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5192190359975593e-05,
+      "loss": 0.0018,
+      "step": 2056
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5009151921903598e-05,
+      "loss": 0.0622,
+      "step": 2057
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4826113483831604e-05,
+      "loss": 0.0154,
+      "step": 2058
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4643075045759609e-05,
+      "loss": 0.0264,
+      "step": 2059
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4460036607687614e-05,
+      "loss": 0.0024,
+      "step": 2060
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4276998169615617e-05,
+      "loss": 0.0043,
+      "step": 2061
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4093959731543624e-05,
+      "loss": 0.0192,
+      "step": 2062
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3910921293471627e-05,
+      "loss": 0.0103,
+      "step": 2063
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3727882855399634e-05,
+      "loss": 0.0282,
+      "step": 2064
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3544844417327637e-05,
+      "loss": 0.0287,
+      "step": 2065
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.336180597925564e-05,
+      "loss": 0.0075,
+      "step": 2066
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3178767541183648e-05,
+      "loss": 0.0131,
+      "step": 2067
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2995729103111651e-05,
+      "loss": 0.0091,
+      "step": 2068
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2812690665039658e-05,
+      "loss": 0.0816,
+      "step": 2069
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2629652226967661e-05,
+      "loss": 0.0222,
+      "step": 2070
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2446613788895668e-05,
+      "loss": 0.0114,
+      "step": 2071
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.2263575350823671e-05,
+      "loss": 0.0121,
+      "step": 2072
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.2080536912751676e-05,
+      "loss": 0.0493,
+      "step": 2073
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1897498474679681e-05,
+      "loss": 0.0585,
+      "step": 2074
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1714460036607687e-05,
+      "loss": 0.0019,
+      "step": 2075
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1531421598535692e-05,
+      "loss": 0.0767,
+      "step": 2076
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1348383160463697e-05,
+      "loss": 0.0064,
+      "step": 2077
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.11653447223917e-05,
+      "loss": 0.0321,
+      "step": 2078
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0982306284319707e-05,
+      "loss": 0.0282,
+      "step": 2079
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.079926784624771e-05,
+      "loss": 0.0052,
+      "step": 2080
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0616229408175717e-05,
+      "loss": 0.0211,
+      "step": 2081
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.043319097010372e-05,
+      "loss": 0.0073,
+      "step": 2082
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0250152532031727e-05,
+      "loss": 0.0048,
+      "step": 2083
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.006711409395973e-05,
+      "loss": 0.0165,
+      "step": 2084
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.884075655887734e-06,
+      "loss": 0.0118,
+      "step": 2085
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.70103721781574e-06,
+      "loss": 0.0036,
+      "step": 2086
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.517998779743744e-06,
+      "loss": 0.0128,
+      "step": 2087
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.33496034167175e-06,
+      "loss": 0.0074,
+      "step": 2088
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.151921903599754e-06,
+      "loss": 0.0213,
+      "step": 2089
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.96888346552776e-06,
+      "loss": 0.0089,
+      "step": 2090
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.785845027455764e-06,
+      "loss": 0.0309,
+      "step": 2091
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.60280658938377e-06,
+      "loss": 0.0075,
+      "step": 2092
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.419768151311775e-06,
+      "loss": 0.0161,
+      "step": 2093
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.23672971323978e-06,
+      "loss": 0.0243,
+      "step": 2094
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.053691275167785e-06,
+      "loss": 0.0063,
+      "step": 2095
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.87065283709579e-06,
+      "loss": 0.0209,
+      "step": 2096
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.687614399023795e-06,
+      "loss": 0.0194,
+      "step": 2097
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.504575960951799e-06,
+      "loss": 0.0031,
+      "step": 2098
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.321537522879804e-06,
+      "loss": 0.0111,
+      "step": 2099
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.1384990848078085e-06,
+      "loss": 0.0579,
+      "step": 2100
+    },
+    {
+      "epoch": 2.94,
+      "eval_loss": 0.08942369371652603,
+      "eval_runtime": 13281.7764,
+      "eval_samples_per_second": 0.199,
+      "eval_steps_per_second": 0.099,
+      "eval_wer": 0.06604522015073383,
+      "step": 2100
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.955460646735814e-06,
+      "loss": 0.0244,
+      "step": 2101
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.772422208663819e-06,
+      "loss": 0.0056,
+      "step": 2102
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.589383770591824e-06,
+      "loss": 0.0304,
+      "step": 2103
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.406345332519829e-06,
+      "loss": 0.0459,
+      "step": 2104
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.223306894447834e-06,
+      "loss": 0.0198,
+      "step": 2105
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.040268456375838e-06,
+      "loss": 0.0026,
+      "step": 2106
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.857230018303843e-06,
+      "loss": 0.0322,
+      "step": 2107
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.674191580231848e-06,
+      "loss": 0.0118,
+      "step": 2108
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.4911531421598534e-06,
+      "loss": 0.0397,
+      "step": 2109
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.3081147040878585e-06,
+      "loss": 0.0413,
+      "step": 2110
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.125076266015864e-06,
+      "loss": 0.0181,
+      "step": 2111
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.942037827943867e-06,
+      "loss": 0.0088,
+      "step": 2112
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.758999389871872e-06,
+      "loss": 0.0087,
+      "step": 2113
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.575960951799877e-06,
+      "loss": 0.0029,
+      "step": 2114
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.392922513727882e-06,
+      "loss": 0.0029,
+      "step": 2115
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.209884075655887e-06,
+      "loss": 0.0288,
+      "step": 2116
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.026845637583892e-06,
+      "loss": 0.0169,
+      "step": 2117
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.8438071995118975e-06,
+      "loss": 0.0274,
+      "step": 2118
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.660768761439902e-06,
+      "loss": 0.0243,
+      "step": 2119
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.477730323367907e-06,
+      "loss": 0.0148,
+      "step": 2120
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.294691885295912e-06,
+      "loss": 0.0234,
+      "step": 2121
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.111653447223917e-06,
+      "loss": 0.0393,
+      "step": 2122
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9286150091519216e-06,
+      "loss": 0.0025,
+      "step": 2123
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7455765710799267e-06,
+      "loss": 0.0166,
+      "step": 2124
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.562538133007932e-06,
+      "loss": 0.2041,
+      "step": 2125
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.379499694935936e-06,
+      "loss": 0.2378,
+      "step": 2126
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.196461256863941e-06,
+      "loss": 0.0179,
+      "step": 2127
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.013422818791946e-06,
+      "loss": 0.0511,
+      "step": 2128
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.830384380719951e-06,
+      "loss": 0.0115,
+      "step": 2129
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.647345942647956e-06,
+      "loss": 0.0099,
+      "step": 2130
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4643075045759608e-06,
+      "loss": 0.0131,
+      "step": 2131
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.281269066503966e-06,
+      "loss": 0.0148,
+      "step": 2132
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0982306284319706e-06,
+      "loss": 0.0168,
+      "step": 2133
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.151921903599755e-07,
+      "loss": 0.0389,
+      "step": 2134
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.321537522879804e-07,
+      "loss": 0.0326,
+      "step": 2135
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.491153142159853e-07,
+      "loss": 0.01,
+      "step": 2136
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.660768761439902e-07,
+      "loss": 0.0079,
+      "step": 2137
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.830384380719951e-07,
+      "loss": 0.0349,
+      "step": 2138
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.1415,
+      "step": 2139
+    },
+    {
+      "epoch": 3.0,
+      "step": 2139,
+      "total_flos": 3.3715855099813274e+18,
+      "train_loss": 0.05018302534102554,
+      "train_runtime": 453623.4225,
+      "train_samples_per_second": 0.019,
+      "train_steps_per_second": 0.005
+    }
+  ],
+  "max_steps": 2139,
+  "num_train_epochs": 3,
+  "total_flos": 3.3715855099813274e+18,
+  "trial_name": null,
+  "trial_params": null
+}